diff --git "a/traces/regular.txt" "b/traces/regular.txt" new file mode 100644--- /dev/null +++ "b/traces/regular.txt" @@ -0,0 +1,5156 @@ +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] Output code: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # AOT ID: ['0_inference'] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from ctypes import c_void_p, c_long +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import torch +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import random +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import os +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import tempfile +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from math import inf, nan +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.hooks import run_intermediate_hooks +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.utils import maybe_profile +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.codegen.memory_planning import _align as align +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch import device, empty_strided +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.async_compile import AsyncCompile +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.select_algorithm import extern_kernels +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.codegen.multi_kernel import MultiKernelCall +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] aten = torch.ops.aten +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_ops = torch.ops.inductor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] _quantized = torch.ops._quantized +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride = torch._C._dynamo.guards.assert_size_stride +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] empty_strided_cpu = torch._C._dynamo.guards._empty_strided_cpu +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] empty_strided_cuda = torch._C._dynamo.guards._empty_strided_cuda +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] reinterpret_tensor = torch._C._dynamo.guards._reinterpret_tensor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] alloc_from_pool = torch.ops.inductor._alloc_from_pool +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] async_compile = AsyncCompile() +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/de/cde7jszcnlkvp4rfgrvjw6xlnnue4zbdrhajl5tacta5otakmb63.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [emb_3], Original ATen: [aten.cat] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # emb_3 => cat +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_cat_0 = async_compile.triton('triton_poi_fused_cat_0', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[512], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*i64', 1: '*fp32', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_cat_0', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_cat_0(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 512 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = xindex < xnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 256 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_ptr0 + (0)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tl.broadcast_to(tmp5, [XBLOCK]) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = x0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.full([1], 0, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 >= tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.full([1], 128, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp0 < tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp6.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp0.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = -9.210340371976184 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tmp8 * tmp9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = 0.0078125 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = tmp10 * tmp11 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = tl_math.exp(tmp12) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tmp7 * tmp13 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp14 * tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp17 = tl_math.sin(tmp16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18 = tl.full(tmp17.shape, 0.0, tmp17.dtype) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19 = tl.where(tmp4, tmp17, tmp18) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp20 = tmp0 >= tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp21 = tl.full([1], 256, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp22 = tmp0 < tmp21 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp23 = (-128) + x0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp24 = tmp23.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp25 = tmp24 * tmp9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp26 = tmp25 * tmp11 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp27 = tl_math.exp(tmp26) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp28 = tmp7 * tmp27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp29 = tmp28 * tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp30 = tl_math.cos(tmp29) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp31 = tl.full(tmp30.shape, 0.0, tmp30.dtype) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp32 = tl.where(tmp20, tmp30, tmp31) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp33 = tl.where(tmp4, tmp19, tmp32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (x2), tmp33, xmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_heuristics import grid, split_scan_grid, grid_combo_kernels, start_graph, end_graph +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._C import _cuda_getCurrentRawStream as get_raw_stream +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/bl/cblwbw2hjtjxrekj6apfmjf5l5x2gjh5imd6mds67uvxms3anoxi.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [emb_4, to_2], Original ATen: [aten._to_copy, aten.cat] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # emb_4 => cat_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # to_2 => convert_element_type_3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused__to_copy_cat_1 = async_compile.triton('triton_poi_fused__to_copy_cat_1', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[512], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*fp32', 1: '*bf16', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused__to_copy_cat_1', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused__to_copy_cat_1(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 512 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = xindex < xnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 256 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x1 = (xindex // 256) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = x0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.full([1], 0, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 >= tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.full([1], 128, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp0 < tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_ptr0 + (128 + (256*x1) + x0), tmp4 & xmask, eviction_policy='evict_last', other=0.0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tmp0 >= tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tl.full([1], 256, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp0 < tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tl.load(in_ptr0 + ((256*x1) + ((-128) + x0)), tmp6 & xmask, eviction_policy='evict_last', other=0.0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.where(tmp4, tmp5, tmp9) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tmp10.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (x2), tmp11, xmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/u3/cu3sqq3how4rmiuqcsp7uieydcf5d4sr2g3d35dtluktnfl77vig.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [emb_4, sample_1, to_2], Original ATen: [aten._to_copy, aten.cat, aten.silu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # emb_4 => cat_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # sample_1 => convert_element_type_7, convert_element_type_8, mul_5, sigmoid +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # to_2 => convert_element_type_3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused__to_copy_cat_silu_2 = async_compile.triton('triton_tem_fused__to_copy_cat_silu_2', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=1, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=2, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused__to_copy_cat_silu_2', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused__to_copy_cat_silu_2(arg_A, arg_B, in_ptr2, out_ptr1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 256 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 256 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 256 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (1152*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr2 + (tl.broadcast_to(idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = acc + tmp0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp1.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.sigmoid(tmp2) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp2 * tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tmp4.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr1 + (tl.broadcast_to(idx_n + (1152*idx_m), acc.shape)), tmp5, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import torch._inductor.kernel.mm_common +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta0 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 16, 'BLOCK_N': 32, 'BLOCK_K': 16} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/zn/cznvdrtenmlcvw5mpcvgrif4kxyext2p5wgfzwceerhmhdoargb2.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [sample_1, silu_1], Original ATen: [aten.silu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # sample_1 => convert_element_type_7, convert_element_type_8, mul_5, sigmoid +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # silu_1 => convert_element_type_12, convert_element_type_13, mul_6, sigmoid_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_silu_3 = async_compile.triton('triton_tem_fused_silu_3', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=5, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=2, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_silu_3', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_silu_3(arg_A, arg_B, in_ptr2, out_ptr0, out_ptr1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (1152*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (tl.broadcast_to(xindex, acc.shape)), acc, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr2 + (tl.broadcast_to(idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = acc + tmp0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp1.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.sigmoid(tmp2) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp2 * tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tmp4.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr1 + (tl.broadcast_to(idx_n + (1152*idx_m), acc.shape)), tmp5, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta1 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 16, 'BLOCK_N': 32, 'BLOCK_K': 128} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/nd/cndti5g4mx6r2cgwieh3gxf5cp5ib7i3av7oa533gh2lkq5j3ug6.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [silu_1], Original ATen: [aten.silu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # silu_1 => convert_element_type_12, convert_element_type_13, mul_6, sigmoid_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_silu_4 = async_compile.triton('triton_tem_fused_silu_4', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=5, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_silu_4', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_silu_4(arg_A, arg_B, out_ptr0): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 6912 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (6912*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (tl.broadcast_to(xindex, acc.shape)), acc, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta2 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 16, 'BLOCK_N': 64, 'BLOCK_K': 128} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/4y/c4yp7ug2ilia5rusvxvoynjkw7m7sgacoiqn4wwrs446kfdxwbss.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [latent], Original ATen: [aten.convolution] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # latent => convolution +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_convolution_5 = async_compile.triton('triton_poi_fused_convolution_5', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[8, 16384], tile_hint=TileHint.SQUARE, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: 'i32', 3: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 3), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_convolution_5', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_convolution_5(in_ptr0, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ynumel = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 16384 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] yoffset = tl.program_id(1) * YBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] yindex = yoffset + tl.arange(0, YBLOCK)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ymask = yindex < ynumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK, YBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] y3 = yindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] y0 = yindex % 4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] y1 = (yindex // 4) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (x2 + (16384*y3)), ymask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (y0 + (4*x2) + (65536*y1)), tmp0, ymask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/7s/c7swnybxfpeleuqpgcd6mqauq6lxxa7gmaznwmhfrdegqcdobhfc.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [latent], Original ATen: [aten.convolution] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # latent => convolution +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_convolution_6 = async_compile.triton('triton_tem_fused_convolution_6', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=2, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_convolution_6', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_convolution_6(arg_X, arg_W, out_ptr0): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] KERNEL_H : tl.constexpr = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] KERNEL_W : tl.constexpr = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] STRIDE_H : tl.constexpr = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] STRIDE_W : tl.constexpr = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] PADDING_H : tl.constexpr = 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] PADDING_W : tl.constexpr = 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUPS : tl.constexpr = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] UNROLL : tl.constexpr = False +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 256 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] X = arg_X +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] W = arg_W +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Tensor dimensions +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BATCH = 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] IN_C = 4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] IN_H = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] IN_W = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] OUT_C = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] OUT_H = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] OUT_W = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Strides: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_xn = 65536 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_xc = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_xh = 512 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_xw = 4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_wc_out = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_wc_in = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_wh = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ww = 4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] nhw = tl.program_id(0) * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_y_w = nhw % OUT_W +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] nh = nhw // OUT_W +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_y_h = nh % OUT_H +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = nh // OUT_H +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_y_c = tl.program_id(1) * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group = 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_IN_C = IN_C +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_OUT_C = OUT_C +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x_base = X + (group * stride_xc * GROUP_IN_C + idx_n * stride_xn)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] w_base = ( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] W + (group * stride_wc_out * GROUP_OUT_C + idx_y_c * stride_wc_out)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Could be simplified, but slightly slower: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # for i in range(KERNEL_H): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # for j in range(KERNEL_W): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # for k in range(0, GROUP_IN_C, BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K_COUNT = (GROUP_IN_C + BLOCK_K - 1) // BLOCK_K +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for ijk in range(KERNEL_H * KERNEL_W * BLOCK_K_COUNT): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] k = (ijk % BLOCK_K_COUNT) * BLOCK_K +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ij = ijk // BLOCK_K_COUNT +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] i = ij // KERNEL_W +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] j = ij % KERNEL_W +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_x_h = i - PADDING_H + idx_y_h * STRIDE_H +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_x_w = j - PADDING_W + idx_y_w * STRIDE_W +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_x_c = tl.arange(0, BLOCK_K) + k +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x_ptrs = x_base + ( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] (idx_x_h * stride_xh)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] + (idx_x_w * stride_xw)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] + (idx_x_c * stride_xc)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask_x = ( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] (idx_n < BATCH)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_x_h >= 0)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_x_h < IN_H)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_x_w >= 0)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_x_w < IN_W)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_x_c < GROUP_IN_C)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] matrix_x = tl.load(x_ptrs, mask=mask_x, other=0.0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] w_ptrs = w_base + ( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] (idx_x_c * stride_wc_in)[:, None] + (i * stride_wh) + (j * stride_ww) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask_w = (idx_x_c[:, None] < GROUP_IN_C) & (idx_y_c[None, :] < GROUP_OUT_C) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] matrix_w = tl.load(w_ptrs, mask=mask_w, other=0.0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(matrix_x, matrix_w, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = ( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] (idx_n < BATCH)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_y_h < OUT_H)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_y_w < OUT_W)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] & (idx_y_c < GROUP_OUT_C)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = idx_n[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_c = idx_y_c[None, :] + group * GROUP_OUT_C +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_h = idx_y_h[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_w = idx_y_w[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_w + (64*idx_h) + (4096*idx_c) + (4718592*idx_n) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (tl.broadcast_to(idx_c + (1152*idx_w) + (73728*idx_h) + (4718592*idx_n), acc.shape)), acc, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import torch._inductor.kernel.conv +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta3 = {'KERNEL_H': 2, 'KERNEL_W': 2, 'STRIDE_H': 2, 'STRIDE_W': 2, 'PADDING_H': 0, 'PADDING_W': 0, 'GROUPS': 1, 'UNROLL': False, 'ALLOW_TF32': True, 'BLOCK_M': 64, 'BLOCK_N': 256, 'BLOCK_K': 16} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/ky/ckyh232ncqjqh2omw3i7heq7abhq6chpqeh5y7khbn5avz3xttb6.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add, add_2, mul_4, norm_hidden_states, norm_hidden_states_1], Original ATen: [aten.add, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # add => add +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # add_2 => add_6 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # mul_4 => mul_14 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # norm_hidden_states => add_5, convert_element_type_25, convert_element_type_26, mul_13, rsqrt, sub_1, var_mean +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # norm_hidden_states_1 => add_7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_mul_native_layer_norm_7 = async_compile.triton('triton_red_fused_add_mul_native_layer_norm_7', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.reduction( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[8192, 2048], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] reduction_hint=ReductionHint.INNER, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16', 5: '*bf16', 6: '*bf16', 7: 'i32', 8: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_red_fused_add_mul_native_layer_norm_7', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 12, 'num_reduction': 2, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_red_fused_add_mul_native_layer_norm_7(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr, RBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 8192 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rnumel = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbase = tl.arange(0, RBLOCK)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 4096 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_mean = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_m2 = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_weight = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r2 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (r2 + (1152*x3)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr1 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.load(in_ptr2 + (r2 + (1152*x0)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 + tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp2 + tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tmp4.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tl.broadcast_to(tmp5, [XBLOCK, RBLOCK]) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_mean_next, tmp7_m2_next, tmp7_weight_next = triton_helpers.welford_reduce( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6, tmp7_mean, tmp7_m2, tmp7_weight, roffset == 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_mean = tl.where(rmask, tmp7_mean_next, tmp7_mean) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_m2 = tl.where(rmask, tmp7_m2_next, tmp7_m2) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_weight = tl.where(rmask, tmp7_weight_next, tmp7_weight) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_tmp, tmp8_tmp, tmp9_tmp = triton_helpers.welford( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7_mean, tmp7_m2, tmp7_weight, 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp7_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp8_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tmp9_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x1 = (xindex // 4096) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r2 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.load(in_ptr0 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tl.load(in_ptr1 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = tl.load(in_ptr2 + (r2 + (1152*x0)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp24 = tl.load(in_ptr3 + (1152 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp25 = tl.load(in_ptr4 + (1152 + r2 + (6912*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp26 = tl.load(in_ptr5 + (1152 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp32 = tl.load(in_ptr3 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp33 = tl.load(in_ptr4 + (r2 + (6912*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp34 = tl.load(in_ptr5 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = tmp10 + tmp11 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tmp12 + tmp13 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = tmp14.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp15 - tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp17 = 1152.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18 = tmp8 / tmp17 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19 = 1e-06 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp20 = tmp18 + tmp19 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp21 = libdevice.rsqrt(tmp20) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp22 = tmp16 * tmp21 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp23 = tmp22.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp27 = tmp25 + tmp26 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp28 = tmp24 + tmp27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp29 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp30 = tmp28 + tmp29 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp31 = tmp23 * tmp30 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp35 = tmp33 + tmp34 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp36 = tmp32 + tmp35 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp37 = tmp31 + tmp36 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr2 + (r2 + (1152*x3)), tmp37, rmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/ki/ckibfeohmeixrmfcuk5oki6iblzyraixxsyip7b4ovqns6ntraxe.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_5], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_5 => clone +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8 = async_compile.triton('triton_poi_fused_clone_8', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[16777216], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_8', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_clone_8(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 9437184 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 72 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x1 = (xindex // 72) % 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = (xindex // 1152) % 4096 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = (xindex // 4718592) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x4 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (x0 + (72*x2) + (294912*x1) + (4718592*x3)), None).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (x4), tmp0, None) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/q2/cq25omdqgimcuhsp4qzny4pqeqlouvnaiv3oxz3aiagkaxnxxyax.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add, attn_output, hidden_states_10, hidden_states_9], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # add => add +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # attn_output => mul_15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_10 => add_8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_9 => div_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_add_div_mul_9 = async_compile.triton('triton_poi_fused_add_div_mul_9', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[16777216], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16', 5: '*bf16', 6: '*bf16', 7: '*bf16', 8: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_add_div_mul_9', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 8, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_add_div_mul_9(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 9437184 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = (xindex // 4718592) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x4 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x5 = xindex % 4718592 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (2304 + x0), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr1 + (2304 + x0 + (6912*x2)), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tl.load(in_ptr2 + (2304 + x0), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_out_ptr0 + (x4), None).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tl.load(in_ptr3 + (x0), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tl.load(in_ptr4 + (x4), None).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = tl.load(in_ptr5 + (x0), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tl.load(in_ptr6 + (x5), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tmp1 + tmp2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp0 + tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp5 + tmp6 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tmp7 * tmp8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tmp4 * tmp9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = tmp11 + tmp12 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = tmp13 + tmp14 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp10 + tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(in_out_ptr0 + (x4), tmp16, None) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/ra/cranq4bcckawbphmj2lsoxnmhjb2qyuifjpvtfquik7jmnlla2qs.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_2], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_2 => add_2, add_3, convert_element_type_20, convert_element_type_21, mul_10, mul_11, mul_12, mul_7, mul_8, mul_9, tanh +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_gelu_10 = async_compile.triton('triton_tem_fused_gelu_10', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=2, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_gelu_10', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_gelu_10(arg_A, arg_B, in_ptr2, out_ptr1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 600 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 4096 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 4096 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 4096 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (1152*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr2 + (tl.broadcast_to(idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = acc + tmp0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp1.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = 0.5 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp2 * tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tmp2 * tmp2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tmp5 * tmp2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = 0.044715 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp6 * tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tmp2 + tmp8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = 0.7978845608028654 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tmp9 * tmp10 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = libdevice.tanh(tmp11) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tmp12 + tmp13 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = tmp4 * tmp14 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp15.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr1 + (tl.broadcast_to(idx_n + (1152*idx_m), acc.shape)), tmp16, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta4 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 32, 'BLOCK_N': 32, 'BLOCK_K': 128} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/xe/cxe5hf5k5os5qqlu7bulaleexfcqduwen3xu327dj7lwykikw2xk.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_3], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_3 => addmm_4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11 = async_compile.triton('triton_tem_fused_addmm_11', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_addmm_11', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_addmm_11(in_ptr0, arg_A, arg_B, out_ptr0): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 600 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (1152*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (tl.broadcast_to(idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = acc + tmp0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (tl.broadcast_to(xindex, acc.shape)), tmp1, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta5 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 64, 'BLOCK_N': 128, 'BLOCK_K': 128} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/bb/cbbrshzzkdma5ca42jidnaqspw44u4f33qzsqilgwvnp5xqkbizf.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_11, hidden_states_30, hidden_states_49], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_11 => constant_pad_nd +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_30 => constant_pad_nd_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_49 => constant_pad_nd_2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12 = async_compile.triton('triton_poi_fused_constant_pad_nd_12', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[16384], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*i64', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_constant_pad_nd_12', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_constant_pad_nd_12(in_ptr0, out_ptr0, out_ptr1, out_ptr2, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 9728 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = xindex < xnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 304 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = (xindex // 4864) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = x0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.full([1], 300, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 < tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.load(in_ptr0 + (x0 + (300*x2)), tmp2 & xmask, eviction_policy='evict_last', other=0.0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp3.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tmp5 - tmp4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = -10000.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp6 * tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tl.full(tmp8.shape, 0.0, tmp8.dtype) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.where(tmp2, tmp8, tmp9) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (x3), tmp10, xmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr1 + (x3), tmp10, xmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr2 + (x3), tmp10, xmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/4l/c4lxmko46u4zufblekvo2p4ywit2erom2ybagpya5xn43s7swrvr.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_6, hidden_states_16, hidden_states_17, mul_6, norm_hidden_states_2, norm_hidden_states_3], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # add_6 => add_11 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_16 => div_2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_17 => add_9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # mul_6 => mul_17 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # norm_hidden_states_2 => add_10, convert_element_type_51, convert_element_type_52, mul_16, rsqrt_1, sub_2, var_mean_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # norm_hidden_states_3 => add_12 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13 = async_compile.triton('triton_red_fused_add_div_mul_native_layer_norm_13', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.reduction( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[8192, 2048], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] reduction_hint=ReductionHint.INNER, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16', 5: '*bf16', 6: '*bf16', 7: 'i32', 8: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_red_fused_add_div_mul_native_layer_norm_13', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 12, 'num_reduction': 2, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_red_fused_add_div_mul_native_layer_norm_13(in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr, RBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 8192 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rnumel = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbase = tl.arange(0, RBLOCK)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_mean = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_m2 = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_weight = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r1 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (r1 + (1152*x0)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr1 + (r1), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_ptr2 + (r1 + (1152*x0)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 + tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp2 * tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tmp4 + tmp5 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp6.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tl.broadcast_to(tmp7, [XBLOCK, RBLOCK]) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_mean_next, tmp9_m2_next, tmp9_weight_next = triton_helpers.welford_reduce( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8, tmp9_mean, tmp9_m2, tmp9_weight, roffset == 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_mean = tl.where(rmask, tmp9_mean_next, tmp9_mean) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_m2 = tl.where(rmask, tmp9_m2_next, tmp9_m2) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_weight = tl.where(rmask, tmp9_weight_next, tmp9_weight) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_tmp, tmp10_tmp, tmp11_tmp = triton_helpers.welford( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9_mean, tmp9_m2, tmp9_weight, 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tmp9_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tmp10_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tmp11_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = (xindex // 4096) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r1 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = tl.load(in_ptr0 + (r1 + (1152*x0)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = tl.load(in_ptr1 + (r1), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp17 = tl.load(in_ptr2 + (r1 + (1152*x0)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp28 = tl.load(in_ptr3 + (4608 + r1), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp29 = tl.load(in_ptr4 + (4608 + r1 + (6912*x3)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp30 = tl.load(in_ptr5 + (4608 + r1), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp35 = tl.load(in_ptr3 + (3456 + r1), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp36 = tl.load(in_ptr4 + (3456 + r1 + (6912*x3)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp37 = tl.load(in_ptr5 + (3456 + r1), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tmp12 + tmp13 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp14 * tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18 = tmp16 + tmp17 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19 = tmp18.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp20 = tmp19 - tmp9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp21 = 1152.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp22 = tmp10 / tmp21 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp23 = 1e-06 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp24 = tmp22 + tmp23 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp25 = libdevice.rsqrt(tmp24) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp26 = tmp20 * tmp25 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp27 = tmp26.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp31 = tmp29 + tmp30 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp32 = tmp28 + tmp31 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp33 = tmp32 + tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp34 = tmp27 * tmp33 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp38 = tmp36 + tmp37 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp39 = tmp35 + tmp38 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp40 = tmp34 + tmp39 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr2 + (r1 + (1152*x0)), tmp40, rmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/dl/cdlczkikibpe42lxtfv523s4nk4xay5tshesy2y6yyd7k6z36th2.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_19], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_19 => add_13, add_14, convert_element_type_56, convert_element_type_57, mul_18, mul_19, mul_20, mul_21, mul_22, mul_23, tanh_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14 = async_compile.triton('triton_poi_fused_gelu_14', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[67108864], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_gelu_14', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_gelu_14(in_out_ptr0, in_ptr0, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 37748736 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 4608 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_out_ptr0 + (x2), None).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr0 + (x0), None, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 + tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tmp2.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = 0.5 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tmp3 * tmp4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tmp3 * tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp6 * tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = 0.044715 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tmp7 * tmp8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tmp3 + tmp9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = 0.7978845608028654 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = tmp10 * tmp11 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = libdevice.tanh(tmp12) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = tmp13 + tmp14 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp5 * tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp17 = tmp16.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(in_out_ptr0 + (x2), tmp17, None) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/3o/c3oxrutw4zqeyml2ey2haeuatjlgfnykjbaeirinczlbrvp6ke2p.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_10, ff_output, hidden_states_16, hidden_states_17, hidden_states_22, mul_8, norm_hidden_states_4, norm_hidden_states_5], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # add_10 => add_18 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # ff_output => mul_24 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_16 => div_2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_17 => add_9 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_22 => add_15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # mul_8 => mul_26 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # norm_hidden_states_4 => add_17, convert_element_type_61, convert_element_type_62, mul_25, rsqrt_2, sub_3, var_mean_2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # norm_hidden_states_5 => add_19 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15 = async_compile.triton('triton_red_fused_add_div_mul_native_layer_norm_15', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.reduction( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[8192, 2048], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] reduction_hint=ReductionHint.INNER, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16', 5: '*bf16', 6: '*bf16', 7: '*bf16', 8: '*bf16', 9: '*bf16', 10: 'i32', 11: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_red_fused_add_div_mul_native_layer_norm_15', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 15, 'num_reduction': 2, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_red_fused_add_div_mul_native_layer_norm_15(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr, RBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 8192 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rnumel = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbase = tl.arange(0, RBLOCK)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x1 = (xindex // 4096) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_m2 = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_weight = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r2 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (5760 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr1 + (5760 + r2 + (6912*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tl.load(in_ptr2 + (5760 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_out_ptr0 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tl.load(in_ptr3 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tl.load(in_ptr4 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.load(in_ptr5 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tl.load(in_ptr6 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tmp1 + tmp2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp0 + tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp5 + tmp6 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp4 * tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tmp9 + tmp10 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = tmp11 * tmp12 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = tmp13 + tmp14 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp8 + tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp17 = tmp16.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18 = tl.broadcast_to(tmp17, [XBLOCK, RBLOCK]) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean_next, tmp19_m2_next, tmp19_weight_next = triton_helpers.welford_reduce( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18, tmp19_mean, tmp19_m2, tmp19_weight, roffset == 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean = tl.where(rmask, tmp19_mean_next, tmp19_mean) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_m2 = tl.where(rmask, tmp19_m2_next, tmp19_m2) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_weight = tl.where(rmask, tmp19_weight_next, tmp19_weight) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(in_out_ptr0 + (r2 + (1152*x3)), tmp16, rmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_tmp, tmp20_tmp, tmp21_tmp = triton_helpers.welford( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean, tmp19_m2, tmp19_weight, 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19 = tmp19_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp20 = tmp20_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp21 = tmp21_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r2 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp22 = tl.load(in_out_ptr0 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp32 = tl.load(in_ptr7 + (1152 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp33 = tl.load(in_ptr1 + (1152 + r2 + (6912*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp34 = tl.load(in_ptr2 + (1152 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp40 = tl.load(in_ptr7 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp41 = tl.load(in_ptr1 + (r2 + (6912*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp42 = tl.load(in_ptr2 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp23 = tmp22.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp24 = tmp23 - tmp19 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp25 = 1152.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp26 = tmp20 / tmp25 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp27 = 1e-06 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp28 = tmp26 + tmp27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp29 = libdevice.rsqrt(tmp28) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp30 = tmp24 * tmp29 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp31 = tmp30.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp35 = tmp33 + tmp34 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp36 = tmp32 + tmp35 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp37 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp38 = tmp36 + tmp37 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp39 = tmp31 * tmp38 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp43 = tmp41 + tmp42 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp44 = tmp40 + tmp43 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp45 = tmp39 + tmp44 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr2 + (r2 + (1152*x3)), tmp45, rmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/3z/c3zdh6zd7jxrws54jyk7qiqrqrdhkm73azcomgvobyedtgfj4qa3.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_1, hidden_states_28, hidden_states_29], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # attn_output_1 => mul_27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_28 => div_3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_29 => add_20 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16 = async_compile.triton('triton_tem_fused_add_div_mul_16', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=3, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16', 5: '*bf16', 6: '*bf16', 7: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_add_div_mul_16', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_add_div_mul_16(arg_A, arg_B, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, out_ptr1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 8192 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (1152*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x4 = (xindex // 4718592) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr2 + (tl.broadcast_to(2304 + idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr3 + (2304 + idx_n + (6912*x4)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tl.load(in_ptr4 + (tl.broadcast_to(2304 + idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_ptr5 + (tl.broadcast_to(idx_n, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.load(in_ptr6 + (tl.broadcast_to(xindex, acc.shape)), mask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tmp1 + tmp2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp0 + tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = acc + tmp5 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp6 * tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tmp4 * tmp8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tmp9 + tmp10 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr1 + (tl.broadcast_to(xindex, acc.shape)), tmp11, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta6 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 64, 'BLOCK_N': 128, 'BLOCK_K': 64} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/7v/c7v3uk7fqux5bxi3psju5nxg7vlgmwkdljjjw5z665aj4myhis2j.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_524], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_524 => constant_pad_nd_27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_17 = async_compile.triton('triton_poi_fused_constant_pad_nd_17', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[16384], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*i64', 1: '*bf16', 2: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_constant_pad_nd_17', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 1, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_constant_pad_nd_17(in_ptr0, out_ptr0, xnumel, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 9728 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = xindex < xnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x0 = xindex % 304 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = (xindex // 4864) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = x0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.full([1], 300, tl.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 < tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tl.load(in_ptr0 + (x0 + (300*x2)), tmp2 & xmask, eviction_policy='evict_last', other=0.0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp3.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tmp5 - tmp4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = -10000.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp6 * tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tl.full(tmp8.shape, 0.0, tmp8.dtype) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.where(tmp2, tmp8, tmp9) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (x3), tmp10, xmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/o5/co5umixi3hmlkkyp5z7bcq4bhs6wlu2r6amuvavtj35wwwjruv4c.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_226, ff_output_27, hidden_states_529, hidden_states_530, hidden_states_535, hidden_states_536, hidden_states_537, mul_116], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # add_226 => add_342 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # ff_output_27 => mul_348 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_529 => div_56 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_530 => add_333 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_535 => add_339 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_536 => add_341, convert_element_type_1033, convert_element_type_1034, mul_349, rsqrt_56, sub_57, var_mean_56 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # hidden_states_537 => add_343 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # mul_116 => mul_350 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_18 = async_compile.triton('triton_red_fused_add_div_mul_native_layer_norm_18', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.reduction( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[8192, 2048], +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] reduction_hint=ReductionHint.INNER, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: '*bf16', 4: '*bf16', 5: '*bf16', 6: '*bf16', 7: '*bf16', 8: '*bf16', 9: '*bf16', 10: '*bf16', 11: '*bf16', 12: 'i32', 13: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_red_fused_add_div_mul_native_layer_norm_18', 'mutated_arg_names': ['in_out_ptr0'], 'no_x_dim': False, 'num_load': 13, 'num_reduction': 2, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_red_fused_add_div_mul_native_layer_norm_18(in_out_ptr0, in_ptr0, in_ptr1, in_ptr2, in_ptr3, in_ptr4, in_ptr5, in_ptr6, in_ptr7, in_ptr8, in_ptr9, out_ptr2, xnumel, rnumel, XBLOCK : tl.constexpr, RBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 8192 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rnumel = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK, RBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbase = tl.arange(0, RBLOCK)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x1 = (xindex // 4096) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_m2 = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_weight = tl.zeros([XBLOCK, RBLOCK], tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r2 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (5760 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr1 + (5760 + r2 + (6912*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tl.load(in_ptr2 + (5760 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp5 = tl.load(in_out_ptr0 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp6 = tl.load(in_ptr3 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp9 = tl.load(in_ptr4 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp10 = tl.load(in_ptr5 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp14 = tl.load(in_ptr6 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp3 = tmp1 + tmp2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp4 = tmp0 + tmp3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp7 = tmp5 + tmp6 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp8 = tmp4 * tmp7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp11 = tmp9 + tmp10 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp12 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp13 = tmp11 * tmp12 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp15 = tmp13 + tmp14 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp16 = tmp8 + tmp15 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp17 = tmp16.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18 = tl.broadcast_to(tmp17, [XBLOCK, RBLOCK]) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean_next, tmp19_m2_next, tmp19_weight_next = triton_helpers.welford_reduce( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp18, tmp19_mean, tmp19_m2, tmp19_weight, roffset == 0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean = tl.where(rmask, tmp19_mean_next, tmp19_mean) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_m2 = tl.where(rmask, tmp19_m2_next, tmp19_m2) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_weight = tl.where(rmask, tmp19_weight_next, tmp19_weight) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(in_out_ptr0 + (r2 + (1152*x3)), tmp16, rmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_tmp, tmp20_tmp, tmp21_tmp = triton_helpers.welford( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19_mean, tmp19_m2, tmp19_weight, 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp19 = tmp19_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp20 = tmp20_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp21 = tmp21_tmp[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for roffset in range(0, rnumel, RBLOCK): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rindex = roffset + rbase +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rmask = rindex < rnumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] r2 = rindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp22 = tl.load(in_out_ptr0 + (r2 + (1152*x3)), rmask, eviction_policy='evict_first', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp32 = tl.load(in_ptr7 + (1152 + r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp33 = tl.load(in_ptr8 + (r2 + (1152*x1)), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp34 = tl.load(in_ptr9 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp40 = tl.load(in_ptr7 + (r2), rmask, eviction_policy='evict_last', other=0.0).to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp23 = tmp22.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp24 = tmp23 - tmp19 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp25 = 1152.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp26 = tmp20 / tmp25 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp27 = 1e-06 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp28 = tmp26 + tmp27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp29 = libdevice.rsqrt(tmp28) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp30 = tmp24 * tmp29 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp31 = tmp30.to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp35 = tmp33 + tmp34 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp36 = tmp32 + tmp35 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp37 = 1.0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp38 = tmp36 + tmp37 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp39 = tmp31 * tmp38 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp41 = tmp40 + tmp35 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp42 = tmp39 + tmp41 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr2 + (r2 + (1152*x3)), tmp42, rmask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/pn/cpndl4h2tgpqwa3v53dicnosfwuo6vtp5acfyzuwduunzbkjx6ef.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_19 = async_compile.triton('triton_tem_fused_19', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.template( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_stages=5, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] num_warps=4, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'kernel_name': 'triton_tem_fused_19', 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_tem_fused_19(arg_A, arg_B, out_ptr0): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] GROUP_M : tl.constexpr = 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] EVEN_K : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ALLOW_TF32 : tl.constexpr = True +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ACC_TYPE : tl.constexpr = tl.float32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B_PROLOGUE_CAST_TYPE : tl.constexpr = None +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_M : tl.constexpr = 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_N : tl.constexpr = 32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] BLOCK_K : tl.constexpr = 128 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = arg_A +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = arg_B +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] M = 8192 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] N = 32 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] K = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if M * N == 0: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # early exit due to zero-size input(s) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_am = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_ak = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bk = 1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stride_bn = 1152 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # based on triton.ops.matmul +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid = tl.program_id(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_m = (M + BLOCK_M - 1) // BLOCK_M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] grid_n = (N + BLOCK_N - 1) // BLOCK_N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # re-order program ID for better L2 performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] width = GROUP_M * grid_n +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_id = pid // width +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] group_size = min(grid_m - group_id * GROUP_M, GROUP_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_m = group_id * GROUP_M + (pid % group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] pid_n = (pid % width) // (group_size) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_am == 1 and stride_ak == M) or (stride_am == K and stride_ak == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = tl.max_contiguous(tl.multiple_of(rm % M, BLOCK_M), BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ram = rm % M +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if (stride_bk == 1 and stride_bn == K) or (stride_bk == N and stride_bn == 1): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = tl.max_contiguous(tl.multiple_of(rn % N, BLOCK_N), BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rbn = rn % N +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rk = tl.arange(0, BLOCK_K) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A = A + (ram[:, None] * stride_am + rk[None, :] * stride_ak) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B = B + (rk[:, None] * stride_bk + rbn[None, :] * stride_bn) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc = tl.zeros((BLOCK_M, BLOCK_N), dtype=ACC_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] for k in range(K, 0, -BLOCK_K): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if EVEN_K: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] else: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] a = tl.load(A, mask=rk[None, :] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = tl.load(B, mask=rk[:, None] < k, other=0.) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if B_PROLOGUE_CAST_TYPE is not None: +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] b = b.to(B_PROLOGUE_CAST_TYPE) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] acc += tl.dot(a, b, allow_tf32=ALLOW_TF32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] A += BLOCK_K * stride_ak +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] B += BLOCK_K * stride_bk +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # rematerialize rm and rn to save registers +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rm = pid_m * BLOCK_M + tl.arange(0, BLOCK_M) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] rn = pid_n * BLOCK_N + tl.arange(0, BLOCK_N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_m = rm[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] idx_n = rn[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] mask = (idx_m < M) & (idx_n < N) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # inductor generates a suffix +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = idx_n + (32*idx_m) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (tl.broadcast_to(xindex, acc.shape)), acc, mask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] meta7 = {'GROUP_M': 8, 'EVEN_K': True, 'ALLOW_TF32': True, 'ACC_TYPE': 'tl.float32', 'B_PROLOGUE_CAST_TYPE': None, 'BLOCK_M': 64, 'BLOCK_N': 32, 'BLOCK_K': 128} +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # kernel path: /tmp/torchinductor_sayak/oa/coadiitwxe7hbol4x276rojcmvv4pyheia3ifaorra5idaysnrcf.py +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [output], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # output => clone_140 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_20 = async_compile.triton('triton_poi_fused_clone_20', ''' +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] import triton.language as tl +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from triton.compiler.compiler import AttrsDescriptor +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime import triton_helpers, triton_heuristics +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.triton_helpers import libdevice, math as tl_math +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.runtime.hints import AutotuneHint, ReductionHint, TileHint, instance_descriptor, DeviceProperties +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton_heuristics.pointwise( +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] size_hints=[16, 16384], tile_hint=TileHint.DEFAULT, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] filename=__file__, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_meta={'signature': {0: '*bf16', 1: '*bf16', 2: '*bf16', 3: 'i32', 4: 'i32'}, 'device': DeviceProperties(type='cuda', index=0, cc=90, major=9, regs_per_multiprocessor=65536, max_threads_per_multi_processor=2048, multi_processor_count=132), 'constants': {}, 'configs': [AttrsDescriptor(divisible_by_16=(0, 1, 2, 3, 4), equal_to_1=())]}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] inductor_meta={'autotune_hints': set(), 'kernel_name': 'triton_poi_fused_clone_20', 'mutated_arg_names': [], 'no_x_dim': False, 'num_load': 2, 'num_reduction': 0, 'backend_hash': '002A1A9B1115CD8E0489B47343AA1BAA75B3F6181CDF90468122931EFBBE395F', 'are_deterministic_algorithms_enabled': False, 'assert_indirect_indexing': True, 'autotune_local_cache': True, 'autotune_pointwise': True, 'autotune_remote_cache': None, 'force_disable_caches': False, 'dynamic_scale_rblock': True, 'max_autotune': True, 'max_autotune_pointwise': False, 'min_split_scan_rblock': 256, 'spill_threshold': 16, 'store_cubin': False}, +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] min_elem_per_thread=0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] @triton.jit +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def triton_poi_fused_clone_20(in_ptr0, in_ptr1, out_ptr0, ynumel, xnumel, YBLOCK : tl.constexpr, XBLOCK : tl.constexpr): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ynumel = 16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xnumel = 16384 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] yoffset = tl.program_id(1) * YBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] yindex = yoffset + tl.arange(0, YBLOCK)[None, :] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ymask = yindex < ynumel +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xoffset = tl.program_id(0) * XBLOCK +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xindex = xoffset + tl.arange(0, XBLOCK)[:, None] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] xmask = tl.full([XBLOCK, YBLOCK], True, tl.int1) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x2 = xindex % 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x3 = (xindex // 2) % 64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x4 = (xindex // 128) % 2 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x5 = (xindex // 256) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] y0 = yindex % 8 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] y1 = (yindex // 8) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] x7 = xindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] y6 = yindex +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp0 = tl.load(in_ptr0 + (y0 + (8*x2) + (16*x4) + (32*x3) + (2048*x5) + (131072*y1)), ymask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp1 = tl.load(in_ptr1 + (y0 + (8*x2) + (16*x4)), ymask, eviction_policy='evict_last').to(tl.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tmp2 = tmp0 + tmp1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] tl.store(out_ptr0 + (x7 + (16384*y6)), tmp2, ymask) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] ''', device_str='cuda') +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] async_compile.wait(globals()) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del async_compile +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def call(args): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg0_1, arg1_1, arg2_1, arg3_1, arg4_1, arg5_1, arg6_1, arg7_1, arg8_1, arg9_1, arg10_1, arg11_1, arg12_1, arg13_1, arg14_1, arg15_1, arg16_1, arg17_1, arg18_1, arg19_1, arg20_1, arg21_1, arg22_1, arg23_1, arg24_1, arg25_1, arg26_1, arg27_1, arg28_1, arg29_1, arg30_1, arg31_1, arg32_1, arg33_1, arg34_1, arg35_1, arg36_1, arg37_1, arg38_1, arg39_1, arg40_1, arg41_1, arg42_1, arg43_1, arg44_1, arg45_1, arg46_1, arg47_1, arg48_1, arg49_1, arg50_1, arg51_1, arg52_1, arg53_1, arg54_1, arg55_1, arg56_1, arg57_1, arg58_1, arg59_1, arg60_1, arg61_1, arg62_1, arg63_1, arg64_1, arg65_1, arg66_1, arg67_1, arg68_1, arg69_1, arg70_1, arg71_1, arg72_1, arg73_1, arg74_1, arg75_1, arg76_1, arg77_1, arg78_1, arg79_1, arg80_1, arg81_1, arg82_1, arg83_1, arg84_1, arg85_1, arg86_1, arg87_1, arg88_1, arg89_1, arg90_1, arg91_1, arg92_1, arg93_1, arg94_1, arg95_1, arg96_1, arg97_1, arg98_1, arg99_1, arg100_1, arg101_1, arg102_1, arg103_1, arg104_1, arg105_1, arg106_1, arg107_1, arg108_1, arg109_1, arg110_1, arg111_1, arg112_1, arg113_1, arg114_1, arg115_1, arg116_1, arg117_1, arg118_1, arg119_1, arg120_1, arg121_1, arg122_1, arg123_1, arg124_1, arg125_1, arg126_1, arg127_1, arg128_1, arg129_1, arg130_1, arg131_1, arg132_1, arg133_1, arg134_1, arg135_1, arg136_1, arg137_1, arg138_1, arg139_1, arg140_1, arg141_1, arg142_1, arg143_1, arg144_1, arg145_1, arg146_1, arg147_1, arg148_1, arg149_1, arg150_1, arg151_1, arg152_1, arg153_1, arg154_1, arg155_1, arg156_1, arg157_1, arg158_1, arg159_1, arg160_1, arg161_1, arg162_1, arg163_1, arg164_1, arg165_1, arg166_1, arg167_1, arg168_1, arg169_1, arg170_1, arg171_1, arg172_1, arg173_1, arg174_1, arg175_1, arg176_1, arg177_1, arg178_1, arg179_1, arg180_1, arg181_1, arg182_1, arg183_1, arg184_1, arg185_1, arg186_1, arg187_1, arg188_1, arg189_1, arg190_1, arg191_1, arg192_1, arg193_1, arg194_1, arg195_1, arg196_1, arg197_1, arg198_1, arg199_1, arg200_1, arg201_1, arg202_1, arg203_1, arg204_1, arg205_1, arg206_1, arg207_1, arg208_1, arg209_1, arg210_1, arg211_1, arg212_1, arg213_1, arg214_1, arg215_1, arg216_1, arg217_1, arg218_1, arg219_1, arg220_1, arg221_1, arg222_1, arg223_1, arg224_1, arg225_1, arg226_1, arg227_1, arg228_1, arg229_1, arg230_1, arg231_1, arg232_1, arg233_1, arg234_1, arg235_1, arg236_1, arg237_1, arg238_1, arg239_1, arg240_1, arg241_1, arg242_1, arg243_1, arg244_1, arg245_1, arg246_1, arg247_1, arg248_1, arg249_1, arg250_1, arg251_1, arg252_1, arg253_1, arg254_1, arg255_1, arg256_1, arg257_1, arg258_1, arg259_1, arg260_1, arg261_1, arg262_1, arg263_1, arg264_1, arg265_1, arg266_1, arg267_1, arg268_1, arg269_1, arg270_1, arg271_1, arg272_1, arg273_1, arg274_1, arg275_1, arg276_1, arg277_1, arg278_1, arg279_1, arg280_1, arg281_1, arg282_1, arg283_1, arg284_1, arg285_1, arg286_1, arg287_1, arg288_1, arg289_1, arg290_1, arg291_1, arg292_1, arg293_1, arg294_1, arg295_1, arg296_1, arg297_1, arg298_1, arg299_1, arg300_1, arg301_1, arg302_1, arg303_1, arg304_1, arg305_1, arg306_1, arg307_1, arg308_1, arg309_1, arg310_1, arg311_1, arg312_1, arg313_1, arg314_1, arg315_1, arg316_1, arg317_1, arg318_1, arg319_1, arg320_1, arg321_1, arg322_1, arg323_1, arg324_1, arg325_1, arg326_1, arg327_1, arg328_1, arg329_1, arg330_1, arg331_1, arg332_1, arg333_1, arg334_1, arg335_1, arg336_1, arg337_1, arg338_1, arg339_1, arg340_1, arg341_1, arg342_1, arg343_1, arg344_1, arg345_1, arg346_1, arg347_1, arg348_1, arg349_1, arg350_1, arg351_1, arg352_1, arg353_1, arg354_1, arg355_1, arg356_1, arg357_1, arg358_1, arg359_1, arg360_1, arg361_1, arg362_1, arg363_1, arg364_1, arg365_1, arg366_1, arg367_1, arg368_1, arg369_1, arg370_1, arg371_1, arg372_1, arg373_1, arg374_1, arg375_1, arg376_1, arg377_1, arg378_1, arg379_1, arg380_1, arg381_1, arg382_1, arg383_1, arg384_1, arg385_1, arg386_1, arg387_1, arg388_1, arg389_1, arg390_1, arg391_1, arg392_1, arg393_1, arg394_1, arg395_1, arg396_1, arg397_1, arg398_1, arg399_1, arg400_1, arg401_1, arg402_1, arg403_1, arg404_1, arg405_1, arg406_1, arg407_1, arg408_1, arg409_1, arg410_1, arg411_1, arg412_1, arg413_1, arg414_1, arg415_1, arg416_1, arg417_1, arg418_1, arg419_1, arg420_1, arg421_1, arg422_1, arg423_1, arg424_1, arg425_1, arg426_1, arg427_1, arg428_1, arg429_1, arg430_1, arg431_1, arg432_1, arg433_1, arg434_1, arg435_1, arg436_1, arg437_1, arg438_1, arg439_1, arg440_1, arg441_1, arg442_1, arg443_1, arg444_1, arg445_1, arg446_1, arg447_1, arg448_1, arg449_1, arg450_1, arg451_1, arg452_1, arg453_1, arg454_1, arg455_1, arg456_1, arg457_1, arg458_1, arg459_1, arg460_1, arg461_1, arg462_1, arg463_1, arg464_1, arg465_1, arg466_1, arg467_1, arg468_1, arg469_1, arg470_1, arg471_1, arg472_1, arg473_1, arg474_1, arg475_1, arg476_1, arg477_1, arg478_1, arg479_1, arg480_1, arg481_1, arg482_1, arg483_1, arg484_1, arg485_1, arg486_1, arg487_1, arg488_1, arg489_1, arg490_1, arg491_1, arg492_1, arg493_1, arg494_1, arg495_1, arg496_1, arg497_1, arg498_1, arg499_1, arg500_1, arg501_1, arg502_1, arg503_1, arg504_1, arg505_1, arg506_1, arg507_1, arg508_1, arg509_1, arg510_1, arg511_1, arg512_1, arg513_1, arg514_1, arg515_1, arg516_1, arg517_1, arg518_1, arg519_1, arg520_1, arg521_1, arg522_1, arg523_1, arg524_1, arg525_1, arg526_1, arg527_1, arg528_1, arg529_1, arg530_1, arg531_1, arg532_1, arg533_1, arg534_1, arg535_1, arg536_1, arg537_1, arg538_1, arg539_1, arg540_1, arg541_1, arg542_1, arg543_1, arg544_1, arg545_1, arg546_1, arg547_1, arg548_1, arg549_1, arg550_1, arg551_1, arg552_1, arg553_1, arg554_1, arg555_1, arg556_1, arg557_1, arg558_1, arg559_1, arg560_1, arg561_1, arg562_1, arg563_1, arg564_1, arg565_1, arg566_1, arg567_1, arg568_1, arg569_1, arg570_1, arg571_1, arg572_1, arg573_1, arg574_1, arg575_1, arg576_1, arg577_1, arg578_1, arg579_1, arg580_1, arg581_1, arg582_1, arg583_1, arg584_1, arg585_1, arg586_1, arg587_1, arg588_1, arg589_1, arg590_1, arg591_1, arg592_1, arg593_1, arg594_1, arg595_1, arg596_1, arg597_1, arg598_1, arg599_1, arg600_1, arg601_1, arg602_1, arg603_1, arg604_1, arg605_1, arg606_1, arg607_1 = args +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] args.clear() +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg0_1, (2, 300), (300, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg1_1, (2, 4, 128, 128), (65536, 16384, 128, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg2_1, (1152, 4, 2, 2), (16, 1, 8, 4)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg3_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg4_1, (1, 4096, 1152), (4718592, 1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg5_1, (2, ), (0, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg6_1, (1152, 256), (256, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg7_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg8_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg9_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg10_1, (6912, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg11_1, (6912, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg12_1, (1152, 4096), (4096, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg13_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg14_1, (2, 300, 4096), (1228800, 4096, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg15_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg16_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg17_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg18_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg19_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg20_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg21_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg22_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg23_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg24_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg25_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg26_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg27_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg28_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg29_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg30_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg31_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg32_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg33_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg34_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg35_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg36_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg37_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg38_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg39_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg40_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg41_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg42_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg43_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg44_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg45_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg46_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg47_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg48_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg49_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg50_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg51_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg52_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg53_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg54_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg55_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg56_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg57_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg58_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg59_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg60_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg61_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg62_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg63_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg64_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg65_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg66_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg67_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg68_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg69_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg70_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg71_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg72_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg73_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg74_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg75_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg76_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg77_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg78_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg79_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg80_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg81_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg82_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg83_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg84_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg85_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg86_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg87_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg88_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg89_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg90_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg91_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg92_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg93_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg94_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg95_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg96_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg97_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg98_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg99_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg100_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg101_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg102_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg103_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg104_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg105_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg106_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg107_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg108_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg109_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg110_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg111_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg112_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg113_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg114_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg115_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg116_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg117_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg118_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg119_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg120_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg121_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg122_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg123_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg124_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg125_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg126_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg127_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg128_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg129_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg130_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg131_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg132_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg133_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg134_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg135_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg136_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg137_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg138_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg139_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg140_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg141_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg142_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg143_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg144_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg145_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg146_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg147_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg148_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg149_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg150_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg151_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg152_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg153_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg154_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg155_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg156_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg157_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg158_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg159_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg160_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg161_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg162_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg163_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg164_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg165_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg166_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg167_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg168_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg169_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg170_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg171_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg172_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg173_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg174_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg175_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg176_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg177_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg178_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg179_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg180_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg181_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg182_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg183_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg184_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg185_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg186_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg187_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg188_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg189_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg190_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg191_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg192_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg193_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg194_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg195_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg196_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg197_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg198_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg199_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg200_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg201_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg202_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg203_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg204_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg205_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg206_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg207_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg208_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg209_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg210_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg211_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg212_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg213_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg214_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg215_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg216_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg217_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg218_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg219_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg220_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg221_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg222_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg223_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg224_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg225_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg226_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg227_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg228_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg229_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg230_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg231_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg232_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg233_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg234_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg235_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg236_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg237_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg238_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg239_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg240_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg241_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg242_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg243_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg244_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg245_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg246_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg247_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg248_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg249_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg250_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg251_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg252_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg253_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg254_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg255_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg256_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg257_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg258_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg259_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg260_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg261_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg262_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg263_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg264_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg265_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg266_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg267_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg268_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg269_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg270_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg271_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg272_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg273_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg274_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg275_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg276_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg277_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg278_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg279_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg280_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg281_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg282_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg283_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg284_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg285_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg286_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg287_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg288_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg289_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg290_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg291_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg292_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg293_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg294_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg295_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg296_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg297_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg298_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg299_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg300_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg301_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg302_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg303_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg304_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg305_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg306_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg307_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg308_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg309_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg310_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg311_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg312_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg313_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg314_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg315_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg316_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg317_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg318_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg319_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg320_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg321_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg322_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg323_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg324_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg325_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg326_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg327_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg328_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg329_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg330_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg331_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg332_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg333_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg334_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg335_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg336_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg337_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg338_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg339_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg340_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg341_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg342_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg343_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg344_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg345_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg346_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg347_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg348_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg349_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg350_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg351_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg352_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg353_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg354_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg355_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg356_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg357_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg358_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg359_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg360_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg361_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg362_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg363_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg364_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg365_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg366_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg367_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg368_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg369_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg370_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg371_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg372_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg373_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg374_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg375_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg376_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg377_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg378_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg379_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg380_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg381_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg382_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg383_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg384_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg385_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg386_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg387_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg388_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg389_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg390_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg391_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg392_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg393_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg394_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg395_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg396_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg397_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg398_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg399_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg400_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg401_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg402_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg403_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg404_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg405_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg406_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg407_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg408_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg409_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg410_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg411_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg412_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg413_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg414_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg415_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg416_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg417_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg418_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg419_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg420_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg421_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg422_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg423_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg424_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg425_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg426_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg427_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg428_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg429_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg430_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg431_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg432_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg433_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg434_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg435_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg436_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg437_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg438_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg439_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg440_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg441_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg442_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg443_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg444_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg445_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg446_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg447_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg448_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg449_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg450_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg451_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg452_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg453_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg454_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg455_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg456_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg457_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg458_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg459_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg460_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg461_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg462_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg463_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg464_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg465_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg466_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg467_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg468_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg469_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg470_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg471_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg472_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg473_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg474_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg475_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg476_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg477_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg478_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg479_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg480_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg481_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg482_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg483_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg484_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg485_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg486_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg487_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg488_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg489_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg490_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg491_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg492_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg493_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg494_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg495_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg496_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg497_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg498_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg499_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg500_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg501_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg502_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg503_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg504_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg505_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg506_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg507_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg508_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg509_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg510_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg511_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg512_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg513_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg514_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg515_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg516_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg517_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg518_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg519_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg520_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg521_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg522_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg523_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg524_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg525_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg526_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg527_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg528_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg529_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg530_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg531_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg532_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg533_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg534_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg535_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg536_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg537_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg538_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg539_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg540_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg541_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg542_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg543_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg544_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg545_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg546_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg547_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg548_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg549_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg550_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg551_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg552_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg553_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg554_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg555_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg556_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg557_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg558_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg559_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg560_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg561_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg562_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg563_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg564_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg565_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg566_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg567_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg568_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg569_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg570_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg571_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg572_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg573_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg574_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg575_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg576_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg577_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg578_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg579_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg580_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg581_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg582_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg583_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg584_1, (6, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg585_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg586_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg587_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg588_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg589_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg590_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg591_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg592_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg593_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg594_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg595_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg596_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg597_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg598_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg599_1, (1152, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg600_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg601_1, (4608, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg602_1, (4608, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg603_1, (1152, 4608), (4608, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg604_1, (1152, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg605_1, (2, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg606_1, (32, 1152), (1152, 1)) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] assert_size_stride(arg607_1, (32, ), (1, )) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] with torch.cuda._DeviceGuard(0): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] torch.cuda.set_device(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf0 = empty_strided_cuda((2, 256), (256, 1), torch.float32) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [emb_3], Original ATen: [aten.cat] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] stream0 = get_raw_stream(0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_cat_0.run(arg5_1, buf0, 512, grid=grid(512), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg5_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf1 = empty_strided_cuda((2, 256), (256, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [emb_4, to_2], Original ATen: [aten._to_copy, aten.cat] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused__to_copy_cat_1.run(buf0, buf1, 512, grid=grid(512), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf0 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf3 = empty_strided_cuda((2, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [emb_4, sample_1, to_2], Original ATen: [aten._to_copy, aten.cat, aten.silu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused__to_copy_cat_silu_2.run(buf1, arg6_1, arg7_1, buf3, grid=torch._inductor.kernel.mm_common.mm_grid(2, 1152, meta0), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg6_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg7_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf4 = empty_strided_cuda((2, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf5 = empty_strided_cuda((2, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [sample_1, silu_1], Original ATen: [aten.silu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_silu_3.run(buf3, arg8_1, arg9_1, buf4, buf5, grid=torch._inductor.kernel.mm_common.mm_grid(2, 1152, meta1), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg8_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf3 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf6 = empty_strided_cuda((2, 6912), (6912, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [silu_1], Original ATen: [aten.silu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_silu_4.run(buf5, arg10_1, buf6, grid=torch._inductor.kernel.mm_common.mm_grid(2, 6912, meta2), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg10_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf5 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf7 = empty_strided_cuda((2, 4, 128, 128), (65536, 1, 512, 4), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [latent], Original ATen: [aten.convolution] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_convolution_5.run(arg1_1, buf7, 8, 16384, grid=grid(8, 16384), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg1_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf8 = empty_strided_cuda((2, 1152, 64, 64), (4718592, 1, 73728, 1152), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [latent], Original ATen: [aten.convolution] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_convolution_6.run(buf7, arg2_1, buf8, grid=torch._inductor.kernel.conv.conv2d_grid(2, 1152, 64, 64, meta3), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg2_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf7 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf12 = empty_strided_cuda((2, 4096, 1152), (4718592, 1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add, add_2, mul_4, norm_hidden_states, norm_hidden_states_1], Original ATen: [aten.add, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_mul_native_layer_norm_7.run(buf8, arg3_1, arg4_1, arg17_1, buf6, arg11_1, buf12, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf13 = empty_strided_cuda((8192, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg19_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf12, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg18_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf13) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg18_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg19_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf14 = empty_strided_cuda((8192, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg21_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf12, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg20_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf14) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg20_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg21_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf15 = empty_strided_cuda((8192, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg23_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf12, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg22_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf15) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg22_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg23_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf12 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_4], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf16 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf13, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf14, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf15, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf13 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf17 = buf16[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf16 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf21 = reinterpret_tensor(buf15, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf15 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_5], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf17, buf21, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf22 = reinterpret_tensor(buf17, (8192, 1152), (1152, 1), 0); del buf17 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf21, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg24_1, (1152, 1152), (1, 1152), 0), out=buf22) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg24_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf23 = reinterpret_tensor(buf22, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf22 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add, attn_output, hidden_states_10, hidden_states_9], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_add_div_mul_9.run(buf23, arg17_1, buf6, arg11_1, arg25_1, buf8, arg3_1, arg4_1, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg25_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg3_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg4_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf24 = reinterpret_tensor(buf8, (8192, 1152), (1152, 1), 0); del buf8 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg27_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf23, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg26_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf24) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg26_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg27_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf26 = empty_strided_cuda((2, 300, 1152), (345600, 1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_2], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_gelu_10.run(arg14_1, arg12_1, arg13_1, buf26, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta4), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg12_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg13_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg14_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf27 = empty_strided_cuda((600, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_3], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg16_1, buf26, arg15_1, buf27, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg15_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg16_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf28 = reinterpret_tensor(buf26, (600, 1152), (1152, 1), 0); del buf26 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_2], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg29_1, buf27, arg28_1, buf28, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg28_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg29_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf29 = empty_strided_cuda((600, 1152), (1152, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_2], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg31_1, buf27, arg30_1, buf29, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg30_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg31_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf30 = empty_strided_cuda((2, 16, 1, 304), (4864, 304, 304, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf63 = empty_strided_cuda((2, 16, 1, 304), (4864, 304, 304, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf96 = empty_strided_cuda((2, 16, 1, 304), (4864, 304, 304, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_11, hidden_states_30, hidden_states_49], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf30, buf63, buf96, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_11], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf31 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf24, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf28, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf29, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf30, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf32 = buf31[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf31 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf36 = buf24; del buf24 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf32, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg32_1, (1152, 1152), (1, 1152), 0), out=buf36) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg32_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf40 = reinterpret_tensor(buf32, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf32 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_6, hidden_states_16, hidden_states_17, mul_6, norm_hidden_states_2, norm_hidden_states_3], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf36, arg33_1, buf23, arg17_1, buf6, arg11_1, buf40, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf41 = empty_strided_cuda((8192, 4608), (4608, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf40, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg34_1, (1152, 4608), (1, 1152), 0), out=buf41) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg34_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf42 = reinterpret_tensor(buf41, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf41 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_19], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf42, arg35_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg35_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf43 = reinterpret_tensor(buf40, (8192, 1152), (1152, 1), 0); del buf40 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf42, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg36_1, (4608, 1152), (1, 4608), 0), out=buf43) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg36_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf44 = reinterpret_tensor(buf43, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf43 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf48 = reinterpret_tensor(buf21, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf21 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_10, ff_output, hidden_states_16, hidden_states_17, hidden_states_22, mul_8, norm_hidden_states_4, norm_hidden_states_5], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf44, arg17_1, buf6, arg11_1, arg37_1, buf36, arg33_1, buf23, arg38_1, buf48, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg17_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg33_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg37_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf49 = buf36; del buf36 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg40_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf48, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg39_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf49) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg39_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg40_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf50 = reinterpret_tensor(buf23, (8192, 1152), (1152, 1), 0); del buf23 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg42_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf48, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg41_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf50) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg41_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg42_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf51 = buf14; del buf14 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg44_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf48, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg43_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf51) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg43_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg44_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf48 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_23], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf52 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf49, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf50, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf51, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf49 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf53 = buf52[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf52 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf57 = reinterpret_tensor(buf51, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf51 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_24], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf53, buf57, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf59 = reinterpret_tensor(buf53, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf53 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_1, hidden_states_28, hidden_states_29], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf57, arg45_1, arg38_1, buf6, arg11_1, arg46_1, buf44, buf59, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg45_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg46_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf60 = reinterpret_tensor(buf57, (8192, 1152), (1152, 1), 0); del buf57 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg48_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf59, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg47_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf60) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg47_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg48_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf61 = buf29; del buf29 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_6], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg50_1, buf27, arg49_1, buf61, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg49_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg50_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf62 = buf28; del buf28 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_6], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg52_1, buf27, arg51_1, buf62, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg51_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg52_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_30], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf64 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf60, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf61, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf62, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf63, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf65 = buf64[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf64 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf69 = buf60; del buf60 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf65, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg53_1, (1152, 1152), (1, 1152), 0), out=buf69) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg53_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf73 = reinterpret_tensor(buf65, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf65 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_14, hidden_states_35, hidden_states_36, mul_10, norm_hidden_states_6, norm_hidden_states_7], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf69, arg54_1, buf59, arg38_1, buf6, arg11_1, buf73, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf74 = reinterpret_tensor(buf42, (8192, 4608), (4608, 1), 0); del buf42 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf73, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg55_1, (1152, 4608), (1, 1152), 0), out=buf74) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg55_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf75 = reinterpret_tensor(buf74, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf74 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_38], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf75, arg56_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg56_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf76 = reinterpret_tensor(buf73, (8192, 1152), (1152, 1), 0); del buf73 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf75, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg57_1, (4608, 1152), (1, 4608), 0), out=buf76) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg57_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf77 = reinterpret_tensor(buf76, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf76 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf81 = buf44; del buf44 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_18, ff_output_1, hidden_states_35, hidden_states_36, hidden_states_41, mul_12, norm_hidden_states_8, norm_hidden_states_9], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf77, arg38_1, buf6, arg11_1, arg58_1, buf69, arg54_1, buf59, arg59_1, buf81, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg38_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg54_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg58_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf82 = buf69; del buf69 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg61_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf81, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg60_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf82) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg60_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg61_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf83 = reinterpret_tensor(buf59, (8192, 1152), (1152, 1), 0); del buf59 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg63_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf81, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg62_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf83) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg62_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg63_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf84 = buf50; del buf50 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg65_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf81, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg64_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf84) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg64_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg65_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf81 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_42], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf85 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf82, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf83, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf84, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf82 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf86 = buf85[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf85 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf90 = reinterpret_tensor(buf84, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf84 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_43], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf86, buf90, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf92 = reinterpret_tensor(buf86, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf86 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_2, hidden_states_47, hidden_states_48], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf90, arg66_1, arg59_1, buf6, arg11_1, arg67_1, buf77, buf92, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg66_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg67_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf93 = reinterpret_tensor(buf90, (8192, 1152), (1152, 1), 0); del buf90 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg69_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf92, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg68_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf93) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg68_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg69_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf94 = buf62; del buf62 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_10], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg71_1, buf27, arg70_1, buf94, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg70_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg71_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf95 = buf61; del buf61 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_10], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg73_1, buf27, arg72_1, buf95, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg72_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg73_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_49], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf97 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf93, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf94, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf95, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf96, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf98 = buf97[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf97 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf102 = buf93; del buf93 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf98, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg74_1, (1152, 1152), (1, 1152), 0), out=buf102) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg74_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf106 = reinterpret_tensor(buf98, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf98 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_22, hidden_states_54, hidden_states_55, mul_14, norm_hidden_states_10, norm_hidden_states_11], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf102, arg75_1, buf92, arg59_1, buf6, arg11_1, buf106, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf107 = reinterpret_tensor(buf75, (8192, 4608), (4608, 1), 0); del buf75 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf106, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg76_1, (1152, 4608), (1, 1152), 0), out=buf107) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg76_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf108 = reinterpret_tensor(buf107, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf107 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_57], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf108, arg77_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg77_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf109 = reinterpret_tensor(buf106, (8192, 1152), (1152, 1), 0); del buf106 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf108, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg78_1, (4608, 1152), (1, 4608), 0), out=buf109) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg78_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf110 = reinterpret_tensor(buf109, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf109 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf114 = buf77; del buf77 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_26, ff_output_2, hidden_states_54, hidden_states_55, hidden_states_60, mul_16, norm_hidden_states_12, norm_hidden_states_13], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf110, arg59_1, buf6, arg11_1, arg79_1, buf102, arg75_1, buf92, arg80_1, buf114, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg59_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg75_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg79_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf115 = reinterpret_tensor(buf92, (8192, 1152), (1152, 1), 0); del buf92 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg82_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf114, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg81_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf115) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg81_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg82_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf116 = buf102; del buf102 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg84_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf114, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg83_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf116) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg83_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg84_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf117 = buf83; del buf83 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg86_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf114, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg85_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf117) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg85_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg86_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf114 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_61], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf118 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf115, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf116, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf117, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf115 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf119 = buf118[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf118 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf123 = reinterpret_tensor(buf117, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf117 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_62], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf119, buf123, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf125 = reinterpret_tensor(buf119, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf119 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_3, hidden_states_66, hidden_states_67], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf123, arg87_1, arg80_1, buf6, arg11_1, arg88_1, buf110, buf125, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg87_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg88_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf126 = reinterpret_tensor(buf123, (8192, 1152), (1152, 1), 0); del buf123 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg90_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf125, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg89_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf126) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg89_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg90_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf127 = buf95; del buf95 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_14], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg92_1, buf27, arg91_1, buf127, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg91_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg92_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf128 = buf94; del buf94 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_14], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg94_1, buf27, arg93_1, buf128, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg93_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg94_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf129 = buf96; del buf96 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf162 = buf63; del buf63 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf195 = buf30; del buf30 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_106, hidden_states_68, hidden_states_87], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf129, buf162, buf195, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_68], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf130 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf126, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf127, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf128, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf129, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf131 = buf130[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf130 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf135 = buf126; del buf126 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf131, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg95_1, (1152, 1152), (1, 1152), 0), out=buf135) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg95_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf139 = reinterpret_tensor(buf131, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf131 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_30, hidden_states_73, hidden_states_74, mul_18, norm_hidden_states_14, norm_hidden_states_15], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf135, arg96_1, buf125, arg80_1, buf6, arg11_1, buf139, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf140 = reinterpret_tensor(buf108, (8192, 4608), (4608, 1), 0); del buf108 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf139, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg97_1, (1152, 4608), (1, 1152), 0), out=buf140) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg97_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf141 = reinterpret_tensor(buf140, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf140 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_76], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf141, arg98_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg98_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf142 = reinterpret_tensor(buf139, (8192, 1152), (1152, 1), 0); del buf139 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf141, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg99_1, (4608, 1152), (1, 4608), 0), out=buf142) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg99_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf143 = reinterpret_tensor(buf142, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf142 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf147 = buf110; del buf110 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_34, ff_output_3, hidden_states_73, hidden_states_74, hidden_states_79, mul_20, norm_hidden_states_16, norm_hidden_states_17], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf143, arg80_1, buf6, arg11_1, arg100_1, buf135, arg96_1, buf125, arg101_1, buf147, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg100_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg80_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg96_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf148 = buf135; del buf135 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg103_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf147, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg102_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf148) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg102_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg103_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf149 = reinterpret_tensor(buf125, (8192, 1152), (1152, 1), 0); del buf125 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg105_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf147, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg104_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf149) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg104_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg105_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf150 = buf116; del buf116 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg107_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf147, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg106_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf150) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg106_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg107_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf147 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_80], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf151 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf148, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf149, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf150, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf148 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf152 = buf151[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf151 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf156 = reinterpret_tensor(buf150, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf150 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_81], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf152, buf156, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf158 = reinterpret_tensor(buf152, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf152 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_4, hidden_states_85, hidden_states_86], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf156, arg108_1, arg101_1, buf6, arg11_1, arg109_1, buf143, buf158, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg108_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg109_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf159 = reinterpret_tensor(buf156, (8192, 1152), (1152, 1), 0); del buf156 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg111_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf158, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg110_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf159) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg110_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg111_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf160 = buf128; del buf128 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_18], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg113_1, buf27, arg112_1, buf160, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg112_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg113_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf161 = buf127; del buf127 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_18], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg115_1, buf27, arg114_1, buf161, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg114_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg115_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_87], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf163 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf159, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf160, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf161, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf162, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf164 = buf163[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf163 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf168 = buf159; del buf159 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf164, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg116_1, (1152, 1152), (1, 1152), 0), out=buf168) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg116_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf172 = reinterpret_tensor(buf164, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf164 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_38, hidden_states_92, hidden_states_93, mul_22, norm_hidden_states_18, norm_hidden_states_19], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf168, arg117_1, buf158, arg101_1, buf6, arg11_1, buf172, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf173 = reinterpret_tensor(buf141, (8192, 4608), (4608, 1), 0); del buf141 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf172, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg118_1, (1152, 4608), (1, 1152), 0), out=buf173) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg118_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf174 = reinterpret_tensor(buf173, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf173 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_95], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf174, arg119_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg119_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf175 = reinterpret_tensor(buf172, (8192, 1152), (1152, 1), 0); del buf172 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf174, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg120_1, (4608, 1152), (1, 4608), 0), out=buf175) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg120_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf176 = reinterpret_tensor(buf175, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf175 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf180 = buf143; del buf143 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_42, ff_output_4, hidden_states_92, hidden_states_93, hidden_states_98, mul_24, norm_hidden_states_20, norm_hidden_states_21], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf176, arg101_1, buf6, arg11_1, arg121_1, buf168, arg117_1, buf158, arg122_1, buf180, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg101_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg117_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg121_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf181 = buf168; del buf168 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg124_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf180, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg123_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf181) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg123_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg124_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf182 = reinterpret_tensor(buf158, (8192, 1152), (1152, 1), 0); del buf158 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg126_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf180, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg125_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf182) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg125_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg126_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf183 = buf149; del buf149 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg128_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf180, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg127_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf183) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg127_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg128_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf180 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_99], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf184 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf181, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf182, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf183, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf181 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf185 = buf184[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf184 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf189 = reinterpret_tensor(buf183, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf183 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_100], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf185, buf189, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf191 = reinterpret_tensor(buf185, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf185 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_5, hidden_states_104, hidden_states_105], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf189, arg129_1, arg122_1, buf6, arg11_1, arg130_1, buf176, buf191, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg129_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg130_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf192 = reinterpret_tensor(buf189, (8192, 1152), (1152, 1), 0); del buf189 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg132_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf191, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg131_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf192) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg131_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg132_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf193 = buf161; del buf161 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_22], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg134_1, buf27, arg133_1, buf193, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg133_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg134_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf194 = buf160; del buf160 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_22], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg136_1, buf27, arg135_1, buf194, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg135_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg136_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_106], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf196 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf192, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf193, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf194, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf195, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf197 = buf196[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf196 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf201 = buf192; del buf192 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf197, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg137_1, (1152, 1152), (1, 1152), 0), out=buf201) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg137_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf205 = reinterpret_tensor(buf197, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf197 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_46, hidden_states_111, hidden_states_112, mul_26, norm_hidden_states_22, norm_hidden_states_23], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf201, arg138_1, buf191, arg122_1, buf6, arg11_1, buf205, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf206 = reinterpret_tensor(buf174, (8192, 4608), (4608, 1), 0); del buf174 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf205, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg139_1, (1152, 4608), (1, 1152), 0), out=buf206) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg139_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf207 = reinterpret_tensor(buf206, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf206 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_114], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf207, arg140_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg140_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf208 = reinterpret_tensor(buf205, (8192, 1152), (1152, 1), 0); del buf205 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf207, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg141_1, (4608, 1152), (1, 4608), 0), out=buf208) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg141_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf209 = reinterpret_tensor(buf208, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf208 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf213 = buf176; del buf176 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_50, ff_output_5, hidden_states_111, hidden_states_112, hidden_states_117, mul_28, norm_hidden_states_24, norm_hidden_states_25], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf209, arg122_1, buf6, arg11_1, arg142_1, buf201, arg138_1, buf191, arg143_1, buf213, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg122_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg138_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg142_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf214 = buf201; del buf201 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg145_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf213, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg144_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf214) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg144_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg145_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf215 = reinterpret_tensor(buf191, (8192, 1152), (1152, 1), 0); del buf191 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg147_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf213, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg146_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf215) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg146_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg147_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf216 = buf182; del buf182 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg149_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf213, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg148_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf216) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg148_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg149_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf213 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_118], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf217 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf214, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf215, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf216, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf214 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf218 = buf217[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf217 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf222 = reinterpret_tensor(buf216, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf216 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_119], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf218, buf222, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf224 = reinterpret_tensor(buf218, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf218 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_6, hidden_states_123, hidden_states_124], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf222, arg150_1, arg143_1, buf6, arg11_1, arg151_1, buf209, buf224, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg150_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg151_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf225 = reinterpret_tensor(buf222, (8192, 1152), (1152, 1), 0); del buf222 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg153_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf224, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg152_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf225) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg152_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg153_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf226 = buf194; del buf194 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_26], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg155_1, buf27, arg154_1, buf226, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg154_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg155_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf227 = buf193; del buf193 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_26], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg157_1, buf27, arg156_1, buf227, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg156_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg157_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf228 = buf195; del buf195 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf261 = buf162; del buf162 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf294 = buf129; del buf129 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_125, hidden_states_144, hidden_states_163], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf228, buf261, buf294, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_125], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf229 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf225, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf226, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf227, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf228, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf230 = buf229[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf229 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf234 = buf225; del buf225 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf230, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg158_1, (1152, 1152), (1, 1152), 0), out=buf234) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg158_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf238 = reinterpret_tensor(buf230, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf230 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_54, hidden_states_130, hidden_states_131, mul_30, norm_hidden_states_26, norm_hidden_states_27], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf234, arg159_1, buf224, arg143_1, buf6, arg11_1, buf238, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf239 = reinterpret_tensor(buf207, (8192, 4608), (4608, 1), 0); del buf207 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf238, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg160_1, (1152, 4608), (1, 1152), 0), out=buf239) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg160_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf240 = reinterpret_tensor(buf239, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf239 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_133], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf240, arg161_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg161_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf241 = reinterpret_tensor(buf238, (8192, 1152), (1152, 1), 0); del buf238 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf240, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg162_1, (4608, 1152), (1, 4608), 0), out=buf241) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg162_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf242 = reinterpret_tensor(buf241, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf241 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf246 = buf209; del buf209 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_58, ff_output_6, hidden_states_130, hidden_states_131, hidden_states_136, mul_32, norm_hidden_states_28, norm_hidden_states_29], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf242, arg143_1, buf6, arg11_1, arg163_1, buf234, arg159_1, buf224, arg164_1, buf246, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg143_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg159_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg163_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf247 = buf234; del buf234 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg166_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf246, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg165_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf247) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg165_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg166_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf248 = reinterpret_tensor(buf224, (8192, 1152), (1152, 1), 0); del buf224 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg168_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf246, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg167_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf248) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg167_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg168_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf249 = buf215; del buf215 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg170_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf246, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg169_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf249) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg169_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg170_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf246 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_137], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf250 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf247, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf248, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf249, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf247 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf251 = buf250[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf250 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf255 = reinterpret_tensor(buf249, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf249 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_138], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf251, buf255, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf257 = reinterpret_tensor(buf251, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf251 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_7, hidden_states_142, hidden_states_143], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf255, arg171_1, arg164_1, buf6, arg11_1, arg172_1, buf242, buf257, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg171_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg172_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf258 = reinterpret_tensor(buf255, (8192, 1152), (1152, 1), 0); del buf255 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg174_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf257, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg173_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf258) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg173_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg174_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf259 = buf227; del buf227 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_30], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg176_1, buf27, arg175_1, buf259, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg175_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg176_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf260 = buf226; del buf226 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_30], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg178_1, buf27, arg177_1, buf260, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg177_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg178_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_144], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf262 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf258, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf259, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf260, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf261, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf263 = buf262[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf262 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf267 = buf258; del buf258 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf263, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg179_1, (1152, 1152), (1, 1152), 0), out=buf267) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg179_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf271 = reinterpret_tensor(buf263, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf263 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_62, hidden_states_149, hidden_states_150, mul_34, norm_hidden_states_30, norm_hidden_states_31], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf267, arg180_1, buf257, arg164_1, buf6, arg11_1, buf271, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf272 = reinterpret_tensor(buf240, (8192, 4608), (4608, 1), 0); del buf240 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf271, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg181_1, (1152, 4608), (1, 1152), 0), out=buf272) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg181_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf273 = reinterpret_tensor(buf272, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf272 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_152], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf273, arg182_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg182_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf274 = reinterpret_tensor(buf271, (8192, 1152), (1152, 1), 0); del buf271 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf273, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg183_1, (4608, 1152), (1, 4608), 0), out=buf274) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg183_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf275 = reinterpret_tensor(buf274, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf274 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf279 = buf242; del buf242 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_66, ff_output_7, hidden_states_149, hidden_states_150, hidden_states_155, mul_36, norm_hidden_states_32, norm_hidden_states_33], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf275, arg164_1, buf6, arg11_1, arg184_1, buf267, arg180_1, buf257, arg185_1, buf279, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg164_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg180_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg184_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf280 = buf267; del buf267 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg187_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf279, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg186_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf280) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg186_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg187_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf281 = reinterpret_tensor(buf257, (8192, 1152), (1152, 1), 0); del buf257 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg189_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf279, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg188_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf281) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg188_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg189_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf282 = buf248; del buf248 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg191_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf279, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg190_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf282) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg190_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg191_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf279 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_156], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf283 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf280, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf281, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf282, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf280 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf284 = buf283[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf283 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf288 = reinterpret_tensor(buf282, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf282 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_157], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf284, buf288, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf290 = reinterpret_tensor(buf284, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf284 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_8, hidden_states_161, hidden_states_162], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf288, arg192_1, arg185_1, buf6, arg11_1, arg193_1, buf275, buf290, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg192_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg193_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf291 = reinterpret_tensor(buf288, (8192, 1152), (1152, 1), 0); del buf288 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg195_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf290, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg194_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf291) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg194_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg195_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf292 = buf260; del buf260 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_34], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg197_1, buf27, arg196_1, buf292, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg196_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg197_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf293 = buf259; del buf259 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_34], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg199_1, buf27, arg198_1, buf293, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg198_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg199_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_163], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf295 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf291, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf292, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf293, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf294, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf296 = buf295[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf295 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf300 = buf291; del buf291 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf296, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg200_1, (1152, 1152), (1, 1152), 0), out=buf300) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg200_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf304 = reinterpret_tensor(buf296, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf296 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_70, hidden_states_168, hidden_states_169, mul_38, norm_hidden_states_34, norm_hidden_states_35], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf300, arg201_1, buf290, arg185_1, buf6, arg11_1, buf304, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf305 = reinterpret_tensor(buf273, (8192, 4608), (4608, 1), 0); del buf273 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf304, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg202_1, (1152, 4608), (1, 1152), 0), out=buf305) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg202_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf306 = reinterpret_tensor(buf305, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf305 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_171], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf306, arg203_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg203_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf307 = reinterpret_tensor(buf304, (8192, 1152), (1152, 1), 0); del buf304 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf306, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg204_1, (4608, 1152), (1, 4608), 0), out=buf307) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg204_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf308 = reinterpret_tensor(buf307, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf307 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf312 = buf275; del buf275 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_74, ff_output_8, hidden_states_168, hidden_states_169, hidden_states_174, mul_40, norm_hidden_states_36, norm_hidden_states_37], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf308, arg185_1, buf6, arg11_1, arg205_1, buf300, arg201_1, buf290, arg206_1, buf312, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg185_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg201_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg205_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf313 = buf300; del buf300 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg208_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf312, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg207_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf313) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg207_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg208_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf314 = reinterpret_tensor(buf290, (8192, 1152), (1152, 1), 0); del buf290 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg210_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf312, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg209_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf314) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg209_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg210_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf315 = buf281; del buf281 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg212_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf312, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg211_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf315) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg211_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg212_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf312 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_175], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf316 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf313, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf314, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf315, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf313 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf317 = buf316[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf316 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf321 = reinterpret_tensor(buf315, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf315 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_176], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf317, buf321, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf323 = reinterpret_tensor(buf317, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf317 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_9, hidden_states_180, hidden_states_181], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf321, arg213_1, arg206_1, buf6, arg11_1, arg214_1, buf308, buf323, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg213_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg214_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf324 = reinterpret_tensor(buf321, (8192, 1152), (1152, 1), 0); del buf321 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg216_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf323, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg215_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf324) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg215_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg216_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf325 = buf293; del buf293 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_38], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg218_1, buf27, arg217_1, buf325, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg217_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg218_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf326 = buf292; del buf292 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_38], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg220_1, buf27, arg219_1, buf326, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg219_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg220_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf327 = buf294; del buf294 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf360 = buf261; del buf261 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf393 = buf228; del buf228 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_182, hidden_states_201, hidden_states_220], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf327, buf360, buf393, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_182], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf328 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf324, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf325, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf326, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf327, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf329 = buf328[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf328 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf333 = buf324; del buf324 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf329, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg221_1, (1152, 1152), (1, 1152), 0), out=buf333) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg221_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf337 = reinterpret_tensor(buf329, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf329 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_78, hidden_states_187, hidden_states_188, mul_42, norm_hidden_states_38, norm_hidden_states_39], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf333, arg222_1, buf323, arg206_1, buf6, arg11_1, buf337, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf338 = reinterpret_tensor(buf306, (8192, 4608), (4608, 1), 0); del buf306 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf337, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg223_1, (1152, 4608), (1, 1152), 0), out=buf338) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg223_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf339 = reinterpret_tensor(buf338, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf338 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_190], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf339, arg224_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg224_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf340 = reinterpret_tensor(buf337, (8192, 1152), (1152, 1), 0); del buf337 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf339, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg225_1, (4608, 1152), (1, 4608), 0), out=buf340) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg225_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf341 = reinterpret_tensor(buf340, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf340 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf345 = buf308; del buf308 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_82, ff_output_9, hidden_states_187, hidden_states_188, hidden_states_193, mul_44, norm_hidden_states_40, norm_hidden_states_41], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf341, arg206_1, buf6, arg11_1, arg226_1, buf333, arg222_1, buf323, arg227_1, buf345, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg206_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg222_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg226_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf346 = buf333; del buf333 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg229_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf345, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg228_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf346) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg228_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg229_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf347 = reinterpret_tensor(buf323, (8192, 1152), (1152, 1), 0); del buf323 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg231_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf345, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg230_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf347) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg230_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg231_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf348 = buf314; del buf314 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg233_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf345, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg232_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf348) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg232_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg233_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf345 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_194], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf349 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf346, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf347, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf348, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf346 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf350 = buf349[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf349 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf354 = reinterpret_tensor(buf348, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf348 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_195], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf350, buf354, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf356 = reinterpret_tensor(buf350, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf350 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_10, hidden_states_199, hidden_states_200], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf354, arg234_1, arg227_1, buf6, arg11_1, arg235_1, buf341, buf356, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg234_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg235_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf357 = reinterpret_tensor(buf354, (8192, 1152), (1152, 1), 0); del buf354 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg237_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf356, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg236_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf357) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg236_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg237_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf358 = buf326; del buf326 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_42], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg239_1, buf27, arg238_1, buf358, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg238_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg239_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf359 = buf325; del buf325 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_42], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg241_1, buf27, arg240_1, buf359, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg240_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg241_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_201], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf361 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf357, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf358, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf359, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf360, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf362 = buf361[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf361 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf366 = buf357; del buf357 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf362, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg242_1, (1152, 1152), (1, 1152), 0), out=buf366) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg242_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf370 = reinterpret_tensor(buf362, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf362 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_86, hidden_states_206, hidden_states_207, mul_46, norm_hidden_states_42, norm_hidden_states_43], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf366, arg243_1, buf356, arg227_1, buf6, arg11_1, buf370, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf371 = reinterpret_tensor(buf339, (8192, 4608), (4608, 1), 0); del buf339 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf370, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg244_1, (1152, 4608), (1, 1152), 0), out=buf371) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg244_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf372 = reinterpret_tensor(buf371, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf371 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_209], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf372, arg245_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg245_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf373 = reinterpret_tensor(buf370, (8192, 1152), (1152, 1), 0); del buf370 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf372, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg246_1, (4608, 1152), (1, 4608), 0), out=buf373) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg246_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf374 = reinterpret_tensor(buf373, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf373 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf378 = buf341; del buf341 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_90, ff_output_10, hidden_states_206, hidden_states_207, hidden_states_212, mul_48, norm_hidden_states_44, norm_hidden_states_45], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf374, arg227_1, buf6, arg11_1, arg247_1, buf366, arg243_1, buf356, arg248_1, buf378, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg227_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg243_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg247_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf379 = buf366; del buf366 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg250_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf378, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg249_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf379) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg249_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg250_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf380 = reinterpret_tensor(buf356, (8192, 1152), (1152, 1), 0); del buf356 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg252_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf378, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg251_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf380) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg251_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg252_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf381 = buf347; del buf347 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg254_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf378, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg253_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf381) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg253_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg254_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf378 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_213], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf382 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf379, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf380, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf381, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf379 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf383 = buf382[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf382 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf387 = reinterpret_tensor(buf381, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf381 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_214], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf383, buf387, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf389 = reinterpret_tensor(buf383, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf383 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_11, hidden_states_218, hidden_states_219], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf387, arg255_1, arg248_1, buf6, arg11_1, arg256_1, buf374, buf389, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg255_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg256_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf390 = reinterpret_tensor(buf387, (8192, 1152), (1152, 1), 0); del buf387 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg258_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf389, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg257_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf390) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg257_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg258_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf391 = buf359; del buf359 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_46], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg260_1, buf27, arg259_1, buf391, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg259_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg260_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf392 = buf358; del buf358 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_46], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg262_1, buf27, arg261_1, buf392, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg261_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg262_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_220], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf394 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf390, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf391, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf392, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf393, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf395 = buf394[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf394 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf399 = buf390; del buf390 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf395, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg263_1, (1152, 1152), (1, 1152), 0), out=buf399) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg263_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf403 = reinterpret_tensor(buf395, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf395 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_94, hidden_states_225, hidden_states_226, mul_50, norm_hidden_states_46, norm_hidden_states_47], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf399, arg264_1, buf389, arg248_1, buf6, arg11_1, buf403, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf404 = reinterpret_tensor(buf372, (8192, 4608), (4608, 1), 0); del buf372 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf403, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg265_1, (1152, 4608), (1, 1152), 0), out=buf404) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg265_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf405 = reinterpret_tensor(buf404, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf404 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_228], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf405, arg266_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg266_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf406 = reinterpret_tensor(buf403, (8192, 1152), (1152, 1), 0); del buf403 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf405, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg267_1, (4608, 1152), (1, 4608), 0), out=buf406) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg267_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf407 = reinterpret_tensor(buf406, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf406 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf411 = buf374; del buf374 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_98, ff_output_11, hidden_states_225, hidden_states_226, hidden_states_231, mul_52, norm_hidden_states_48, norm_hidden_states_49], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf407, arg248_1, buf6, arg11_1, arg268_1, buf399, arg264_1, buf389, arg269_1, buf411, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg248_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg264_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg268_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf412 = buf399; del buf399 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg271_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf411, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg270_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf412) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg270_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg271_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf413 = reinterpret_tensor(buf389, (8192, 1152), (1152, 1), 0); del buf389 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg273_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf411, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg272_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf413) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg272_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg273_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf414 = buf380; del buf380 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg275_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf411, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg274_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf414) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg274_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg275_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf411 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_232], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf415 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf412, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf413, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf414, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf412 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf416 = buf415[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf415 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf420 = reinterpret_tensor(buf414, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf414 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_233], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf416, buf420, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf422 = reinterpret_tensor(buf416, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf416 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_12, hidden_states_237, hidden_states_238], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf420, arg276_1, arg269_1, buf6, arg11_1, arg277_1, buf407, buf422, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg276_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg277_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf423 = reinterpret_tensor(buf420, (8192, 1152), (1152, 1), 0); del buf420 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg279_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf422, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg278_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf423) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg278_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg279_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf424 = buf392; del buf392 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_50], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg281_1, buf27, arg280_1, buf424, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg280_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg281_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf425 = buf391; del buf391 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_50], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg283_1, buf27, arg282_1, buf425, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg282_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg283_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf426 = buf393; del buf393 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf459 = buf360; del buf360 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf492 = buf327; del buf327 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_239, hidden_states_258, hidden_states_277], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf426, buf459, buf492, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_239], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf427 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf423, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf424, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf425, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf426, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf428 = buf427[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf427 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf432 = buf423; del buf423 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf428, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg284_1, (1152, 1152), (1, 1152), 0), out=buf432) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg284_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf436 = reinterpret_tensor(buf428, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf428 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_102, hidden_states_244, hidden_states_245, mul_54, norm_hidden_states_50, norm_hidden_states_51], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf432, arg285_1, buf422, arg269_1, buf6, arg11_1, buf436, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf437 = reinterpret_tensor(buf405, (8192, 4608), (4608, 1), 0); del buf405 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf436, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg286_1, (1152, 4608), (1, 1152), 0), out=buf437) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg286_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf438 = reinterpret_tensor(buf437, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf437 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_247], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf438, arg287_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg287_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf439 = reinterpret_tensor(buf436, (8192, 1152), (1152, 1), 0); del buf436 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf438, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg288_1, (4608, 1152), (1, 4608), 0), out=buf439) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg288_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf440 = reinterpret_tensor(buf439, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf439 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf444 = buf407; del buf407 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_106, ff_output_12, hidden_states_244, hidden_states_245, hidden_states_250, mul_56, norm_hidden_states_52, norm_hidden_states_53], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf440, arg269_1, buf6, arg11_1, arg289_1, buf432, arg285_1, buf422, arg290_1, buf444, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg269_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg285_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg289_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf445 = buf432; del buf432 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg292_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf444, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg291_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf445) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg291_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg292_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf446 = reinterpret_tensor(buf422, (8192, 1152), (1152, 1), 0); del buf422 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg294_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf444, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg293_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf446) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg293_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg294_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf447 = buf413; del buf413 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg296_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf444, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg295_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf447) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg295_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg296_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf444 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_251], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf448 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf445, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf446, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf447, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf445 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf449 = buf448[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf448 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf453 = reinterpret_tensor(buf447, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf447 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_252], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf449, buf453, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf455 = reinterpret_tensor(buf449, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf449 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_13, hidden_states_256, hidden_states_257], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf453, arg297_1, arg290_1, buf6, arg11_1, arg298_1, buf440, buf455, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg297_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg298_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf456 = reinterpret_tensor(buf453, (8192, 1152), (1152, 1), 0); del buf453 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg300_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf455, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg299_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf456) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg299_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg300_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf457 = buf425; del buf425 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_54], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg302_1, buf27, arg301_1, buf457, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg301_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg302_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf458 = buf424; del buf424 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_54], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg304_1, buf27, arg303_1, buf458, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg303_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg304_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_258], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf460 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf456, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf457, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf458, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf459, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf461 = buf460[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf460 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf465 = buf456; del buf456 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf461, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg305_1, (1152, 1152), (1, 1152), 0), out=buf465) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg305_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf469 = reinterpret_tensor(buf461, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf461 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_110, hidden_states_263, hidden_states_264, mul_58, norm_hidden_states_54, norm_hidden_states_55], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf465, arg306_1, buf455, arg290_1, buf6, arg11_1, buf469, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf470 = reinterpret_tensor(buf438, (8192, 4608), (4608, 1), 0); del buf438 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf469, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg307_1, (1152, 4608), (1, 1152), 0), out=buf470) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg307_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf471 = reinterpret_tensor(buf470, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf470 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_266], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf471, arg308_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg308_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf472 = reinterpret_tensor(buf469, (8192, 1152), (1152, 1), 0); del buf469 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf471, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg309_1, (4608, 1152), (1, 4608), 0), out=buf472) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg309_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf473 = reinterpret_tensor(buf472, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf472 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf477 = buf440; del buf440 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_114, ff_output_13, hidden_states_263, hidden_states_264, hidden_states_269, mul_60, norm_hidden_states_56, norm_hidden_states_57], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf473, arg290_1, buf6, arg11_1, arg310_1, buf465, arg306_1, buf455, arg311_1, buf477, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg290_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg306_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg310_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf478 = buf465; del buf465 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg313_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf477, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg312_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf478) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg312_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg313_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf479 = reinterpret_tensor(buf455, (8192, 1152), (1152, 1), 0); del buf455 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg315_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf477, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg314_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf479) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg314_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg315_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf480 = buf446; del buf446 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg317_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf477, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg316_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf480) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg316_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg317_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf477 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_270], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf481 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf478, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf479, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf480, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf478 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf482 = buf481[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf481 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf486 = reinterpret_tensor(buf480, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf480 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_271], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf482, buf486, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf488 = reinterpret_tensor(buf482, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf482 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_14, hidden_states_275, hidden_states_276], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf486, arg318_1, arg311_1, buf6, arg11_1, arg319_1, buf473, buf488, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg318_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg319_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf489 = reinterpret_tensor(buf486, (8192, 1152), (1152, 1), 0); del buf486 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg321_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf488, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg320_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf489) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg320_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg321_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf490 = buf458; del buf458 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_58], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg323_1, buf27, arg322_1, buf490, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg322_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg323_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf491 = buf457; del buf457 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_58], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg325_1, buf27, arg324_1, buf491, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg324_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg325_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_277], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf493 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf489, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf490, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf491, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf492, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf494 = buf493[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf493 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf498 = buf489; del buf489 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf494, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg326_1, (1152, 1152), (1, 1152), 0), out=buf498) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg326_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf502 = reinterpret_tensor(buf494, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf494 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_118, hidden_states_282, hidden_states_283, mul_62, norm_hidden_states_58, norm_hidden_states_59], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf498, arg327_1, buf488, arg311_1, buf6, arg11_1, buf502, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf503 = reinterpret_tensor(buf471, (8192, 4608), (4608, 1), 0); del buf471 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf502, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg328_1, (1152, 4608), (1, 1152), 0), out=buf503) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg328_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf504 = reinterpret_tensor(buf503, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf503 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_285], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf504, arg329_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg329_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf505 = reinterpret_tensor(buf502, (8192, 1152), (1152, 1), 0); del buf502 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf504, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg330_1, (4608, 1152), (1, 4608), 0), out=buf505) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg330_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf506 = reinterpret_tensor(buf505, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf505 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf510 = buf473; del buf473 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_122, ff_output_14, hidden_states_282, hidden_states_283, hidden_states_288, mul_64, norm_hidden_states_60, norm_hidden_states_61], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf506, arg311_1, buf6, arg11_1, arg331_1, buf498, arg327_1, buf488, arg332_1, buf510, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg311_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg327_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg331_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf511 = buf498; del buf498 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg334_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf510, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg333_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf511) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg333_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg334_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf512 = reinterpret_tensor(buf488, (8192, 1152), (1152, 1), 0); del buf488 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg336_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf510, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg335_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf512) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg335_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg336_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf513 = buf479; del buf479 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg338_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf510, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg337_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf513) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg337_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg338_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf510 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_289], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf514 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf511, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf512, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf513, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf511 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf515 = buf514[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf514 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf519 = reinterpret_tensor(buf513, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf513 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_290], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf515, buf519, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf521 = reinterpret_tensor(buf515, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf515 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_15, hidden_states_294, hidden_states_295], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf519, arg339_1, arg332_1, buf6, arg11_1, arg340_1, buf506, buf521, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg339_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg340_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf522 = reinterpret_tensor(buf519, (8192, 1152), (1152, 1), 0); del buf519 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg342_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf521, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg341_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf522) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg341_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg342_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf523 = buf491; del buf491 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_62], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg344_1, buf27, arg343_1, buf523, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg343_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg344_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf524 = buf490; del buf490 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_62], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg346_1, buf27, arg345_1, buf524, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg345_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg346_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf525 = buf492; del buf492 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf558 = buf459; del buf459 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf591 = buf426; del buf426 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_296, hidden_states_315, hidden_states_334], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf525, buf558, buf591, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_296], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf526 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf522, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf523, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf524, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf525, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf527 = buf526[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf526 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf531 = buf522; del buf522 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf527, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg347_1, (1152, 1152), (1, 1152), 0), out=buf531) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg347_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf535 = reinterpret_tensor(buf527, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf527 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_126, hidden_states_301, hidden_states_302, mul_66, norm_hidden_states_62, norm_hidden_states_63], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf531, arg348_1, buf521, arg332_1, buf6, arg11_1, buf535, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf536 = reinterpret_tensor(buf504, (8192, 4608), (4608, 1), 0); del buf504 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf535, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg349_1, (1152, 4608), (1, 1152), 0), out=buf536) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg349_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf537 = reinterpret_tensor(buf536, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf536 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_304], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf537, arg350_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg350_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf538 = reinterpret_tensor(buf535, (8192, 1152), (1152, 1), 0); del buf535 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf537, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg351_1, (4608, 1152), (1, 4608), 0), out=buf538) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg351_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf539 = reinterpret_tensor(buf538, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf538 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf543 = buf506; del buf506 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_130, ff_output_15, hidden_states_301, hidden_states_302, hidden_states_307, mul_68, norm_hidden_states_64, norm_hidden_states_65], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf539, arg332_1, buf6, arg11_1, arg352_1, buf531, arg348_1, buf521, arg353_1, buf543, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg332_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg348_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg352_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf544 = buf531; del buf531 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg355_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf543, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg354_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf544) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg354_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg355_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf545 = reinterpret_tensor(buf521, (8192, 1152), (1152, 1), 0); del buf521 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg357_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf543, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg356_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf545) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg356_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg357_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf546 = buf512; del buf512 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg359_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf543, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg358_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf546) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg358_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg359_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf543 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_308], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf547 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf544, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf545, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf546, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf544 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf548 = buf547[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf547 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf552 = reinterpret_tensor(buf546, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf546 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_309], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf548, buf552, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf554 = reinterpret_tensor(buf548, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf548 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_16, hidden_states_313, hidden_states_314], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf552, arg360_1, arg353_1, buf6, arg11_1, arg361_1, buf539, buf554, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg360_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg361_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf555 = reinterpret_tensor(buf552, (8192, 1152), (1152, 1), 0); del buf552 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg363_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf554, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg362_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf555) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg362_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg363_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf556 = buf524; del buf524 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_66], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg365_1, buf27, arg364_1, buf556, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg364_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg365_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf557 = buf523; del buf523 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_66], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg367_1, buf27, arg366_1, buf557, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg366_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg367_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_315], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf559 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf555, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf556, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf557, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf558, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf560 = buf559[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf559 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf564 = buf555; del buf555 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf560, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg368_1, (1152, 1152), (1, 1152), 0), out=buf564) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg368_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf568 = reinterpret_tensor(buf560, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf560 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_134, hidden_states_320, hidden_states_321, mul_70, norm_hidden_states_66, norm_hidden_states_67], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf564, arg369_1, buf554, arg353_1, buf6, arg11_1, buf568, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf569 = reinterpret_tensor(buf537, (8192, 4608), (4608, 1), 0); del buf537 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf568, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg370_1, (1152, 4608), (1, 1152), 0), out=buf569) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg370_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf570 = reinterpret_tensor(buf569, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf569 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_323], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf570, arg371_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg371_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf571 = reinterpret_tensor(buf568, (8192, 1152), (1152, 1), 0); del buf568 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf570, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg372_1, (4608, 1152), (1, 4608), 0), out=buf571) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg372_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf572 = reinterpret_tensor(buf571, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf571 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf576 = buf539; del buf539 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_138, ff_output_16, hidden_states_320, hidden_states_321, hidden_states_326, mul_72, norm_hidden_states_68, norm_hidden_states_69], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf572, arg353_1, buf6, arg11_1, arg373_1, buf564, arg369_1, buf554, arg374_1, buf576, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg353_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg369_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg373_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf577 = buf564; del buf564 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg376_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf576, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg375_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf577) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg375_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg376_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf578 = reinterpret_tensor(buf554, (8192, 1152), (1152, 1), 0); del buf554 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg378_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf576, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg377_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf578) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg377_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg378_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf579 = buf545; del buf545 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg380_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf576, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg379_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf579) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg379_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg380_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf576 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_327], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf580 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf577, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf578, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf579, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf577 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf581 = buf580[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf580 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf585 = reinterpret_tensor(buf579, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf579 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_328], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf581, buf585, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf587 = reinterpret_tensor(buf581, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf581 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_17, hidden_states_332, hidden_states_333], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf585, arg381_1, arg374_1, buf6, arg11_1, arg382_1, buf572, buf587, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg381_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg382_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf588 = reinterpret_tensor(buf585, (8192, 1152), (1152, 1), 0); del buf585 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg384_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf587, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg383_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf588) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg383_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg384_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf589 = buf557; del buf557 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_70], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg386_1, buf27, arg385_1, buf589, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg385_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg386_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf590 = buf556; del buf556 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_70], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg388_1, buf27, arg387_1, buf590, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg387_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg388_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_334], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf592 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf588, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf589, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf590, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf591, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf593 = buf592[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf592 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf597 = buf588; del buf588 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf593, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg389_1, (1152, 1152), (1, 1152), 0), out=buf597) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg389_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf601 = reinterpret_tensor(buf593, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf593 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_142, hidden_states_339, hidden_states_340, mul_74, norm_hidden_states_70, norm_hidden_states_71], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf597, arg390_1, buf587, arg374_1, buf6, arg11_1, buf601, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf602 = reinterpret_tensor(buf570, (8192, 4608), (4608, 1), 0); del buf570 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf601, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg391_1, (1152, 4608), (1, 1152), 0), out=buf602) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg391_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf603 = reinterpret_tensor(buf602, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf602 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_342], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf603, arg392_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg392_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf604 = reinterpret_tensor(buf601, (8192, 1152), (1152, 1), 0); del buf601 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf603, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg393_1, (4608, 1152), (1, 4608), 0), out=buf604) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg393_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf605 = reinterpret_tensor(buf604, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf604 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf609 = buf572; del buf572 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_146, ff_output_17, hidden_states_339, hidden_states_340, hidden_states_345, mul_76, norm_hidden_states_72, norm_hidden_states_73], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf605, arg374_1, buf6, arg11_1, arg394_1, buf597, arg390_1, buf587, arg395_1, buf609, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg374_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg390_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg394_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf610 = buf597; del buf597 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg397_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf609, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg396_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf610) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg396_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg397_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf611 = reinterpret_tensor(buf587, (8192, 1152), (1152, 1), 0); del buf587 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg399_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf609, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg398_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf611) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg398_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg399_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf612 = buf578; del buf578 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg401_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf609, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg400_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf612) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg400_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg401_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf609 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_346], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf613 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf610, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf611, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf612, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf610 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf614 = buf613[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf613 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf618 = reinterpret_tensor(buf612, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf612 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_347], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf614, buf618, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf620 = reinterpret_tensor(buf614, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf614 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_18, hidden_states_351, hidden_states_352], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf618, arg402_1, arg395_1, buf6, arg11_1, arg403_1, buf605, buf620, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg402_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg403_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf621 = reinterpret_tensor(buf618, (8192, 1152), (1152, 1), 0); del buf618 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg405_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf620, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg404_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf621) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg404_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg405_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf622 = buf590; del buf590 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_74], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg407_1, buf27, arg406_1, buf622, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg406_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg407_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf623 = buf589; del buf589 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_74], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg409_1, buf27, arg408_1, buf623, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg408_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg409_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf624 = buf591; del buf591 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf657 = buf558; del buf558 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf690 = buf525; del buf525 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_353, hidden_states_372, hidden_states_391], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf624, buf657, buf690, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_353], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf625 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf621, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf622, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf623, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf624, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf626 = buf625[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf625 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf630 = buf621; del buf621 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf626, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg410_1, (1152, 1152), (1, 1152), 0), out=buf630) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg410_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf634 = reinterpret_tensor(buf626, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf626 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_150, hidden_states_358, hidden_states_359, mul_78, norm_hidden_states_74, norm_hidden_states_75], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf630, arg411_1, buf620, arg395_1, buf6, arg11_1, buf634, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf635 = reinterpret_tensor(buf603, (8192, 4608), (4608, 1), 0); del buf603 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf634, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg412_1, (1152, 4608), (1, 1152), 0), out=buf635) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg412_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf636 = reinterpret_tensor(buf635, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf635 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_361], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf636, arg413_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg413_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf637 = reinterpret_tensor(buf634, (8192, 1152), (1152, 1), 0); del buf634 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf636, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg414_1, (4608, 1152), (1, 4608), 0), out=buf637) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg414_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf638 = reinterpret_tensor(buf637, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf637 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf642 = buf605; del buf605 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_154, ff_output_18, hidden_states_358, hidden_states_359, hidden_states_364, mul_80, norm_hidden_states_76, norm_hidden_states_77], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf638, arg395_1, buf6, arg11_1, arg415_1, buf630, arg411_1, buf620, arg416_1, buf642, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg395_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg411_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg415_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf643 = buf630; del buf630 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg418_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf642, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg417_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf643) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg417_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg418_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf644 = reinterpret_tensor(buf620, (8192, 1152), (1152, 1), 0); del buf620 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg420_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf642, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg419_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf644) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg419_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg420_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf645 = buf611; del buf611 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg422_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf642, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg421_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf645) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg421_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg422_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf642 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_365], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf646 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf643, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf644, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf645, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf643 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf647 = buf646[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf646 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf651 = reinterpret_tensor(buf645, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf645 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_366], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf647, buf651, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf653 = reinterpret_tensor(buf647, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf647 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_19, hidden_states_370, hidden_states_371], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf651, arg423_1, arg416_1, buf6, arg11_1, arg424_1, buf638, buf653, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg423_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg424_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf654 = reinterpret_tensor(buf651, (8192, 1152), (1152, 1), 0); del buf651 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg426_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf653, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg425_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf654) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg425_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg426_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf655 = buf623; del buf623 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_78], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg428_1, buf27, arg427_1, buf655, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg427_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg428_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf656 = buf622; del buf622 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_78], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg430_1, buf27, arg429_1, buf656, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg429_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg430_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_372], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf658 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf654, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf655, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf656, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf657, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf659 = buf658[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf658 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf663 = buf654; del buf654 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf659, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg431_1, (1152, 1152), (1, 1152), 0), out=buf663) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg431_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf667 = reinterpret_tensor(buf659, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf659 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_158, hidden_states_377, hidden_states_378, mul_82, norm_hidden_states_78, norm_hidden_states_79], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf663, arg432_1, buf653, arg416_1, buf6, arg11_1, buf667, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf668 = reinterpret_tensor(buf636, (8192, 4608), (4608, 1), 0); del buf636 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf667, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg433_1, (1152, 4608), (1, 1152), 0), out=buf668) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg433_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf669 = reinterpret_tensor(buf668, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf668 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_380], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf669, arg434_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg434_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf670 = reinterpret_tensor(buf667, (8192, 1152), (1152, 1), 0); del buf667 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf669, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg435_1, (4608, 1152), (1, 4608), 0), out=buf670) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg435_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf671 = reinterpret_tensor(buf670, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf670 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf675 = buf638; del buf638 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_162, ff_output_19, hidden_states_377, hidden_states_378, hidden_states_383, mul_84, norm_hidden_states_80, norm_hidden_states_81], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf671, arg416_1, buf6, arg11_1, arg436_1, buf663, arg432_1, buf653, arg437_1, buf675, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg416_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg432_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg436_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf676 = buf663; del buf663 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg439_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf675, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg438_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf676) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg438_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg439_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf677 = reinterpret_tensor(buf653, (8192, 1152), (1152, 1), 0); del buf653 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg441_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf675, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg440_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf677) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg440_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg441_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf678 = buf644; del buf644 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg443_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf675, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg442_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf678) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg442_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg443_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf675 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_384], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf679 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf676, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf677, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf678, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf676 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf680 = buf679[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf679 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf684 = reinterpret_tensor(buf678, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf678 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_385], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf680, buf684, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf686 = reinterpret_tensor(buf680, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf680 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_20, hidden_states_389, hidden_states_390], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf684, arg444_1, arg437_1, buf6, arg11_1, arg445_1, buf671, buf686, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg444_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg445_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf687 = reinterpret_tensor(buf684, (8192, 1152), (1152, 1), 0); del buf684 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg447_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf686, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg446_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf687) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg446_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg447_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf688 = buf656; del buf656 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_82], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg449_1, buf27, arg448_1, buf688, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg448_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg449_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf689 = buf655; del buf655 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_82], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg451_1, buf27, arg450_1, buf689, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg450_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg451_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_391], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf691 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf687, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf688, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf689, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf690, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf692 = buf691[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf691 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf696 = buf687; del buf687 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf692, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg452_1, (1152, 1152), (1, 1152), 0), out=buf696) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg452_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf700 = reinterpret_tensor(buf692, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf692 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_166, hidden_states_396, hidden_states_397, mul_86, norm_hidden_states_82, norm_hidden_states_83], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf696, arg453_1, buf686, arg437_1, buf6, arg11_1, buf700, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf701 = reinterpret_tensor(buf669, (8192, 4608), (4608, 1), 0); del buf669 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf700, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg454_1, (1152, 4608), (1, 1152), 0), out=buf701) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg454_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf702 = reinterpret_tensor(buf701, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf701 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_399], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf702, arg455_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg455_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf703 = reinterpret_tensor(buf700, (8192, 1152), (1152, 1), 0); del buf700 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf702, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg456_1, (4608, 1152), (1, 4608), 0), out=buf703) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg456_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf704 = reinterpret_tensor(buf703, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf703 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf708 = buf671; del buf671 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_170, ff_output_20, hidden_states_396, hidden_states_397, hidden_states_402, mul_88, norm_hidden_states_84, norm_hidden_states_85], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf704, arg437_1, buf6, arg11_1, arg457_1, buf696, arg453_1, buf686, arg458_1, buf708, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg437_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg453_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg457_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf709 = buf696; del buf696 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg460_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf708, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg459_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf709) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg459_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg460_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf710 = reinterpret_tensor(buf686, (8192, 1152), (1152, 1), 0); del buf686 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg462_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf708, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg461_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf710) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg461_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg462_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf711 = buf677; del buf677 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg464_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf708, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg463_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf711) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg463_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg464_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf708 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_403], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf712 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf709, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf710, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf711, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf709 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf713 = buf712[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf712 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf717 = reinterpret_tensor(buf711, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf711 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_404], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf713, buf717, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf719 = reinterpret_tensor(buf713, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf713 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_21, hidden_states_408, hidden_states_409], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf717, arg465_1, arg458_1, buf6, arg11_1, arg466_1, buf704, buf719, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg465_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg466_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf720 = reinterpret_tensor(buf717, (8192, 1152), (1152, 1), 0); del buf717 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg468_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf719, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg467_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf720) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg467_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg468_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf721 = buf689; del buf689 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_86], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg470_1, buf27, arg469_1, buf721, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg469_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg470_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf722 = buf688; del buf688 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_86], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg472_1, buf27, arg471_1, buf722, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg471_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg472_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf723 = buf690; del buf690 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf756 = buf657; del buf657 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf789 = buf624; del buf624 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_410, hidden_states_429, hidden_states_448], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf723, buf756, buf789, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_410], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf724 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf720, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf721, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf722, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf723, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf725 = buf724[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf724 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf729 = buf720; del buf720 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf725, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg473_1, (1152, 1152), (1, 1152), 0), out=buf729) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg473_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf733 = reinterpret_tensor(buf725, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf725 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_174, hidden_states_415, hidden_states_416, mul_90, norm_hidden_states_86, norm_hidden_states_87], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf729, arg474_1, buf719, arg458_1, buf6, arg11_1, buf733, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf734 = reinterpret_tensor(buf702, (8192, 4608), (4608, 1), 0); del buf702 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf733, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg475_1, (1152, 4608), (1, 1152), 0), out=buf734) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg475_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf735 = reinterpret_tensor(buf734, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf734 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_418], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf735, arg476_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg476_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf736 = reinterpret_tensor(buf733, (8192, 1152), (1152, 1), 0); del buf733 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf735, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg477_1, (4608, 1152), (1, 4608), 0), out=buf736) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg477_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf737 = reinterpret_tensor(buf736, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf736 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf741 = buf704; del buf704 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_178, ff_output_21, hidden_states_415, hidden_states_416, hidden_states_421, mul_92, norm_hidden_states_88, norm_hidden_states_89], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf737, arg458_1, buf6, arg11_1, arg478_1, buf729, arg474_1, buf719, arg479_1, buf741, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg458_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg474_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg478_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf742 = buf729; del buf729 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg481_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf741, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg480_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf742) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg480_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg481_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf743 = reinterpret_tensor(buf719, (8192, 1152), (1152, 1), 0); del buf719 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg483_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf741, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg482_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf743) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg482_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg483_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf744 = buf710; del buf710 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg485_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf741, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg484_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf744) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg484_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg485_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf741 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_422], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf745 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf742, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf743, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf744, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf742 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf746 = buf745[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf745 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf750 = reinterpret_tensor(buf744, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf744 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_423], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf746, buf750, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf752 = reinterpret_tensor(buf746, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf746 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_22, hidden_states_427, hidden_states_428], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf750, arg486_1, arg479_1, buf6, arg11_1, arg487_1, buf737, buf752, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg486_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg487_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf753 = reinterpret_tensor(buf750, (8192, 1152), (1152, 1), 0); del buf750 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg489_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf752, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg488_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf753) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg488_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg489_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf754 = buf722; del buf722 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_90], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg491_1, buf27, arg490_1, buf754, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg490_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg491_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf755 = buf721; del buf721 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_90], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg493_1, buf27, arg492_1, buf755, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg492_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg493_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_429], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf757 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf753, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf754, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf755, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf756, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf758 = buf757[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf757 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf762 = buf753; del buf753 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf758, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg494_1, (1152, 1152), (1, 1152), 0), out=buf762) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg494_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf766 = reinterpret_tensor(buf758, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf758 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_182, hidden_states_434, hidden_states_435, mul_94, norm_hidden_states_90, norm_hidden_states_91], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf762, arg495_1, buf752, arg479_1, buf6, arg11_1, buf766, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf767 = reinterpret_tensor(buf735, (8192, 4608), (4608, 1), 0); del buf735 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf766, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg496_1, (1152, 4608), (1, 1152), 0), out=buf767) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg496_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf768 = reinterpret_tensor(buf767, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf767 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_437], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf768, arg497_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg497_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf769 = reinterpret_tensor(buf766, (8192, 1152), (1152, 1), 0); del buf766 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf768, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg498_1, (4608, 1152), (1, 4608), 0), out=buf769) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg498_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf770 = reinterpret_tensor(buf769, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf769 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf774 = buf737; del buf737 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_186, ff_output_22, hidden_states_434, hidden_states_435, hidden_states_440, mul_96, norm_hidden_states_92, norm_hidden_states_93], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf770, arg479_1, buf6, arg11_1, arg499_1, buf762, arg495_1, buf752, arg500_1, buf774, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg479_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg495_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg499_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf775 = buf762; del buf762 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg502_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf774, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg501_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf775) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg501_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg502_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf776 = reinterpret_tensor(buf752, (8192, 1152), (1152, 1), 0); del buf752 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg504_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf774, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg503_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf776) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg503_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg504_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf777 = buf743; del buf743 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg506_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf774, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg505_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf777) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg505_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg506_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf774 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_441], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf778 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf775, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf776, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf777, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf775 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf779 = buf778[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf778 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf783 = reinterpret_tensor(buf777, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf777 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_442], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf779, buf783, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf785 = reinterpret_tensor(buf779, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf779 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_23, hidden_states_446, hidden_states_447], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf783, arg507_1, arg500_1, buf6, arg11_1, arg508_1, buf770, buf785, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg507_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg508_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf786 = reinterpret_tensor(buf783, (8192, 1152), (1152, 1), 0); del buf783 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg510_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf785, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg509_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf786) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg509_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg510_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf787 = buf755; del buf755 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_94], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg512_1, buf27, arg511_1, buf787, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg511_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg512_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf788 = buf754; del buf754 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_94], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg514_1, buf27, arg513_1, buf788, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg513_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg514_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_448], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf790 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf786, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf787, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf788, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf789, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf791 = buf790[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf790 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf795 = buf786; del buf786 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf791, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg515_1, (1152, 1152), (1, 1152), 0), out=buf795) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg515_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf799 = reinterpret_tensor(buf791, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf791 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_190, hidden_states_453, hidden_states_454, mul_98, norm_hidden_states_94, norm_hidden_states_95], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf795, arg516_1, buf785, arg500_1, buf6, arg11_1, buf799, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf800 = reinterpret_tensor(buf768, (8192, 4608), (4608, 1), 0); del buf768 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf799, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg517_1, (1152, 4608), (1, 1152), 0), out=buf800) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg517_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf801 = reinterpret_tensor(buf800, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf800 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_456], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf801, arg518_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg518_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf802 = reinterpret_tensor(buf799, (8192, 1152), (1152, 1), 0); del buf799 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf801, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg519_1, (4608, 1152), (1, 4608), 0), out=buf802) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg519_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf803 = reinterpret_tensor(buf802, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf802 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf807 = buf770; del buf770 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_194, ff_output_23, hidden_states_453, hidden_states_454, hidden_states_459, mul_100, norm_hidden_states_96, norm_hidden_states_97], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf803, arg500_1, buf6, arg11_1, arg520_1, buf795, arg516_1, buf785, arg521_1, buf807, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg500_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg516_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg520_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf808 = buf795; del buf795 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg523_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf807, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg522_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf808) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg522_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg523_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf809 = reinterpret_tensor(buf785, (8192, 1152), (1152, 1), 0); del buf785 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg525_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf807, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg524_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf809) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg524_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg525_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf810 = buf776; del buf776 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg527_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf807, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg526_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf810) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg526_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg527_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf807 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_460], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf811 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf808, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf809, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf810, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf808 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf812 = buf811[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf811 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf816 = reinterpret_tensor(buf810, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf810 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_461], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf812, buf816, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf818 = reinterpret_tensor(buf812, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf812 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_24, hidden_states_465, hidden_states_466], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf816, arg528_1, arg521_1, buf6, arg11_1, arg529_1, buf803, buf818, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg528_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg529_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf819 = reinterpret_tensor(buf816, (8192, 1152), (1152, 1), 0); del buf816 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg531_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf818, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg530_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf819) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg530_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg531_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf820 = buf788; del buf788 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_98], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg533_1, buf27, arg532_1, buf820, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg532_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg533_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf821 = buf787; del buf787 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_98], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg535_1, buf27, arg534_1, buf821, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg534_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg535_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf822 = buf789; del buf789 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf855 = buf756; del buf756 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf888 = buf723; del buf723 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_467, hidden_states_486, hidden_states_505], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_12.run(arg0_1, buf822, buf855, buf888, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_467], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf823 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf819, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf820, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf821, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf822, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf822 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf824 = buf823[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf823 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf828 = buf819; del buf819 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf824, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg536_1, (1152, 1152), (1, 1152), 0), out=buf828) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg536_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf832 = reinterpret_tensor(buf824, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf824 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_198, hidden_states_472, hidden_states_473, mul_102, norm_hidden_states_98, norm_hidden_states_99], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf828, arg537_1, buf818, arg521_1, buf6, arg11_1, buf832, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf833 = reinterpret_tensor(buf801, (8192, 4608), (4608, 1), 0); del buf801 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf832, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg538_1, (1152, 4608), (1, 1152), 0), out=buf833) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg538_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf834 = reinterpret_tensor(buf833, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf833 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_475], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf834, arg539_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg539_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf835 = reinterpret_tensor(buf832, (8192, 1152), (1152, 1), 0); del buf832 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf834, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg540_1, (4608, 1152), (1, 4608), 0), out=buf835) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg540_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf836 = reinterpret_tensor(buf835, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf835 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf840 = buf803; del buf803 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_202, ff_output_24, hidden_states_472, hidden_states_473, hidden_states_478, mul_104, norm_hidden_states_100, norm_hidden_states_101], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf836, arg521_1, buf6, arg11_1, arg541_1, buf828, arg537_1, buf818, arg542_1, buf840, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg521_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg537_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg541_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf841 = buf828; del buf828 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg544_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf840, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg543_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf841) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg543_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg544_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf842 = reinterpret_tensor(buf818, (8192, 1152), (1152, 1), 0); del buf818 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg546_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf840, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg545_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf842) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg545_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg546_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf843 = buf809; del buf809 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg548_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf840, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg547_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf843) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg547_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg548_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf840 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_479], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf844 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf841, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf842, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf843, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf841 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf845 = buf844[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf844 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf849 = reinterpret_tensor(buf843, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf843 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_480], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf845, buf849, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf851 = reinterpret_tensor(buf845, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf845 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_25, hidden_states_484, hidden_states_485], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf849, arg549_1, arg542_1, buf6, arg11_1, arg550_1, buf836, buf851, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg549_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg550_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf852 = reinterpret_tensor(buf849, (8192, 1152), (1152, 1), 0); del buf849 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg552_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf851, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg551_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf852) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg551_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg552_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf853 = buf821; del buf821 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_102], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg554_1, buf27, arg553_1, buf853, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg553_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg554_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf854 = buf820; del buf820 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_102], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg556_1, buf27, arg555_1, buf854, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg555_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg556_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_486], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf856 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf852, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf853, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf854, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf855, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf855 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf857 = buf856[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf856 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf861 = buf852; del buf852 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf857, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg557_1, (1152, 1152), (1, 1152), 0), out=buf861) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg557_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf865 = reinterpret_tensor(buf857, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf857 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_206, hidden_states_491, hidden_states_492, mul_106, norm_hidden_states_102, norm_hidden_states_103], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf861, arg558_1, buf851, arg542_1, buf6, arg11_1, buf865, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf866 = reinterpret_tensor(buf834, (8192, 4608), (4608, 1), 0); del buf834 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf865, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg559_1, (1152, 4608), (1, 1152), 0), out=buf866) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg559_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf867 = reinterpret_tensor(buf866, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf866 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_494], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf867, arg560_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg560_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf868 = reinterpret_tensor(buf865, (8192, 1152), (1152, 1), 0); del buf865 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf867, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg561_1, (4608, 1152), (1, 4608), 0), out=buf868) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg561_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf869 = reinterpret_tensor(buf868, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf868 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf873 = buf836; del buf836 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_210, ff_output_25, hidden_states_491, hidden_states_492, hidden_states_497, mul_108, norm_hidden_states_104, norm_hidden_states_105], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf869, arg542_1, buf6, arg11_1, arg562_1, buf861, arg558_1, buf851, arg563_1, buf873, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg542_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg558_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg562_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf874 = buf861; del buf861 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg565_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf873, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg564_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf874) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg564_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg565_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf875 = reinterpret_tensor(buf851, (8192, 1152), (1152, 1), 0); del buf851 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg567_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf873, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg566_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf875) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg566_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg567_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf876 = buf842; del buf842 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg569_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf873, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg568_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf876) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg568_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg569_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf873 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_498], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf877 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf874, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf875, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf876, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf874 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf878 = buf877[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf877 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf882 = reinterpret_tensor(buf876, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf876 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_499], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf878, buf882, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf884 = reinterpret_tensor(buf878, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf878 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_26, hidden_states_503, hidden_states_504], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf882, arg570_1, arg563_1, buf6, arg11_1, arg571_1, buf869, buf884, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg570_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg571_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf885 = reinterpret_tensor(buf882, (8192, 1152), (1152, 1), 0); del buf882 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg573_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf884, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg572_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf885) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg572_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg573_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf886 = buf854; del buf854 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_106], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg575_1, buf27, arg574_1, buf886, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg574_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg575_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf887 = buf853; del buf853 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_106], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg577_1, buf27, arg576_1, buf887, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg576_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg577_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_505], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf889 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf885, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf886, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf887, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf888, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf890 = buf889[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf889 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf894 = buf885; del buf885 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf890, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg578_1, (1152, 1152), (1, 1152), 0), out=buf894) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg578_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf898 = reinterpret_tensor(buf890, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf890 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_214, hidden_states_510, hidden_states_511, mul_110, norm_hidden_states_106, norm_hidden_states_107], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf894, arg579_1, buf884, arg563_1, buf6, arg11_1, buf898, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf899 = reinterpret_tensor(buf867, (8192, 4608), (4608, 1), 0); del buf867 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf898, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg580_1, (1152, 4608), (1, 1152), 0), out=buf899) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg580_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf900 = reinterpret_tensor(buf899, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf899 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_513], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf900, arg581_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg581_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf901 = reinterpret_tensor(buf898, (8192, 1152), (1152, 1), 0); del buf898 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf900, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg582_1, (4608, 1152), (1, 4608), 0), out=buf901) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg582_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf902 = reinterpret_tensor(buf901, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf901 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf906 = buf869; del buf869 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_218, ff_output_26, hidden_states_510, hidden_states_511, hidden_states_516, mul_112, norm_hidden_states_108, norm_hidden_states_109], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_15.run(buf902, arg563_1, buf6, arg11_1, arg583_1, buf894, arg579_1, buf884, arg584_1, buf906, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg563_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg579_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg583_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf907 = buf894; del buf894 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg586_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf906, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg585_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf907) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg585_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg586_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf908 = reinterpret_tensor(buf884, (8192, 1152), (1152, 1), 0); del buf884 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg588_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf906, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg587_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf908) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg587_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg588_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf909 = buf875; del buf875 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg590_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf906, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg589_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf909) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg589_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg590_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf906 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_517], Original ATen: [aten._scaled_dot_product_cudnn_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf910 = torch.ops.aten._scaled_dot_product_cudnn_attention.default(reinterpret_tensor(buf907, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf908, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf909, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), None, False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf907 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf908 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf911 = buf910[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf910 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf915 = reinterpret_tensor(buf909, (2, 4096, 16, 72), (4718592, 1152, 72, 1), 0); del buf909 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_518], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_8.run(buf911, buf915, 9437184, grid=grid(9437184), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf917 = reinterpret_tensor(buf911, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf911 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [attn_output_27, hidden_states_522, hidden_states_523], Original ATen: [aten.add, aten.div, aten.mul] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_add_div_mul_16.run(buf915, arg591_1, arg584_1, buf6, arg11_1, arg592_1, buf902, buf917, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 1152, meta6), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg591_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg592_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf918 = reinterpret_tensor(buf915, (8192, 1152), (1152, 1), 0); del buf915 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.bias_addmm(reinterpret_tensor(arg594_1, (8192, 1152), (0, 1), 0), reinterpret_tensor(buf917, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg593_1, (1152, 1152), (1, 1152), 0), alpha=1, beta=1, out=buf918) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg593_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg594_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf919 = buf887; del buf887 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [key_110], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg596_1, buf27, arg595_1, buf919, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg595_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg596_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf920 = buf886; del buf886 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [value_110], Original ATen: [aten.addmm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_addmm_11.run(arg598_1, buf27, arg597_1, buf920, grid=torch._inductor.kernel.mm_common.mm_grid(600, 1152, meta5), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg597_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg598_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf27 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf921 = buf888; del buf888 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_524], Original ATen: [aten.constant_pad_nd] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_constant_pad_nd_17.run(arg0_1, buf921, 9728, grid=grid(9728), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg0_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_524], Original ATen: [aten._scaled_dot_product_efficient_attention] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf922 = torch.ops.aten._scaled_dot_product_efficient_attention.default(reinterpret_tensor(buf918, (2, 16, 4096, 72), (4718592, 72, 1152, 1), 0), reinterpret_tensor(buf919, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf920, (2, 16, 300, 72), (345600, 72, 1152, 1), 0), reinterpret_tensor(buf921, (2, 16, 4096, 300), (4864, 304, 0, 1), 0), False) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf919 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf920 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf921 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf923 = buf922[0] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf922 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf927 = buf918; del buf918 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf923, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg599_1, (1152, 1152), (1, 1152), 0), out=buf927) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg599_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf931 = reinterpret_tensor(buf923, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf923 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_222, hidden_states_529, hidden_states_530, mul_114, norm_hidden_states_110, norm_hidden_states_111], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_13.run(buf927, arg600_1, buf917, arg584_1, buf6, arg11_1, buf931, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf932 = reinterpret_tensor(buf900, (8192, 4608), (4608, 1), 0); del buf900 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf931, (8192, 1152), (1152, 1), 0), reinterpret_tensor(arg601_1, (1152, 4608), (1, 1152), 0), out=buf932) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg601_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf933 = reinterpret_tensor(buf932, (2, 4096, 4608), (18874368, 4608, 1), 0); del buf932 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [hidden_states_532], Original ATen: [aten.gelu] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_gelu_14.run(buf933, arg602_1, 37748736, grid=grid(37748736), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg602_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf934 = reinterpret_tensor(buf931, (8192, 1152), (1152, 1), 0); del buf931 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] extern_kernels.mm(reinterpret_tensor(buf933, (8192, 4608), (4608, 1), 0), reinterpret_tensor(arg603_1, (4608, 1152), (1, 4608), 0), out=buf934) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg603_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf933 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf935 = reinterpret_tensor(buf934, (2, 4096, 1152), (4718592, 1152, 1), 0); del buf934 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf939 = buf902; del buf902 # reuse +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [add_226, ff_output_27, hidden_states_529, hidden_states_530, hidden_states_535, hidden_states_536, hidden_states_537, mul_116], Original ATen: [aten.add, aten.div, aten.mul, aten.native_layer_norm] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_red_fused_add_div_mul_native_layer_norm_18.run(buf935, arg584_1, buf6, arg11_1, arg604_1, buf927, arg600_1, buf917, arg605_1, buf4, arg9_1, buf939, 8192, 1152, grid=grid(8192), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg11_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg584_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg600_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg604_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg605_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg9_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf4 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf6 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf917 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf927 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf935 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf940 = empty_strided_cuda((8192, 32), (32, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [], Original ATen: [] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_tem_fused_19.run(buf939, arg606_1, buf940, grid=torch._inductor.kernel.mm_common.mm_grid(8192, 32, meta7), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg606_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf939 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] buf941 = empty_strided_cuda((2, 8, 64, 2, 64, 2), (131072, 16384, 256, 128, 2, 1), torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] # Source Nodes: [output], Original ATen: [aten.clone] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] triton_poi_fused_clone_20.run(buf940, arg607_1, buf941, 16, 16384, grid=grid(16, 16384), stream=stream0) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del arg607_1 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] del buf940 +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return (reinterpret_tensor(buf941, (2, 8, 128, 128), (131072, 16384, 128, 1), 0), ) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] def benchmark_compiled_module(times=10, repeat=10): +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._dynamo.testing import rand_strided +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.utils import print_performance +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg0_1 = rand_strided((2, 300), (300, 1), device='cuda:0', dtype=torch.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg1_1 = rand_strided((2, 4, 128, 128), (65536, 16384, 128, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg2_1 = rand_strided((1152, 4, 2, 2), (16, 1, 8, 4), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg3_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg4_1 = rand_strided((1, 4096, 1152), (4718592, 1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg5_1 = rand_strided((2, ), (0, ), device='cuda:0', dtype=torch.int64) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg6_1 = rand_strided((1152, 256), (256, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg7_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg8_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg9_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg10_1 = rand_strided((6912, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg11_1 = rand_strided((6912, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg12_1 = rand_strided((1152, 4096), (4096, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg13_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg14_1 = rand_strided((2, 300, 4096), (1228800, 4096, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg15_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg16_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg17_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg18_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg19_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg20_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg21_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg22_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg23_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg24_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg25_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg26_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg27_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg28_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg29_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg30_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg31_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg32_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg33_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg34_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg35_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg36_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg37_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg38_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg39_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg40_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg41_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg42_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg43_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg44_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg45_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg46_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg47_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg48_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg49_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg50_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg51_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg52_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg53_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg54_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg55_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg56_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg57_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg58_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg59_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg60_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg61_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg62_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg63_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg64_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg65_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg66_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg67_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg68_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg69_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg70_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg71_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg72_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg73_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg74_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg75_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg76_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg77_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg78_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg79_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg80_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg81_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg82_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg83_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg84_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg85_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg86_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg87_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg88_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg89_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg90_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg91_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg92_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg93_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg94_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg95_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg96_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg97_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg98_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg99_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg100_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg101_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg102_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg103_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg104_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg105_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg106_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg107_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg108_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg109_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg110_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg111_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg112_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg113_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg114_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg115_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg116_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg117_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg118_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg119_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg120_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg121_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg122_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg123_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg124_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg125_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg126_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg127_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg128_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg129_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg130_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg131_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg132_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg133_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg134_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg135_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg136_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg137_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg138_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg139_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg140_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg141_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg142_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg143_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg144_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg145_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg146_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg147_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg148_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg149_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg150_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg151_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg152_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg153_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg154_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg155_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg156_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg157_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg158_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg159_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg160_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg161_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg162_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg163_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg164_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg165_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg166_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg167_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg168_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg169_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg170_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg171_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg172_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg173_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg174_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg175_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg176_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg177_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg178_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg179_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg180_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg181_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg182_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg183_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg184_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg185_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg186_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg187_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg188_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg189_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg190_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg191_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg192_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg193_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg194_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg195_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg196_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg197_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg198_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg199_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg200_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg201_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg202_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg203_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg204_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg205_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg206_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg207_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg208_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg209_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg210_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg211_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg212_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg213_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg214_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg215_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg216_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg217_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg218_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg219_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg220_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg221_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg222_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg223_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg224_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg225_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg226_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg227_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg228_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg229_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg230_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg231_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg232_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg233_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg234_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg235_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg236_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg237_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg238_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg239_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg240_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg241_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg242_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg243_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg244_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg245_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg246_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg247_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg248_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg249_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg250_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg251_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg252_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg253_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg254_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg255_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg256_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg257_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg258_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg259_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg260_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg261_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg262_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg263_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg264_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg265_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg266_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg267_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg268_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg269_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg270_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg271_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg272_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg273_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg274_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg275_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg276_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg277_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg278_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg279_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg280_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg281_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg282_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg283_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg284_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg285_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg286_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg287_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg288_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg289_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg290_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg291_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg292_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg293_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg294_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg295_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg296_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg297_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg298_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg299_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg300_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg301_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg302_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg303_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg304_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg305_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg306_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg307_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg308_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg309_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg310_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg311_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg312_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg313_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg314_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg315_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg316_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg317_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg318_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg319_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg320_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg321_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg322_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg323_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg324_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg325_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg326_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg327_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg328_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg329_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg330_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg331_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg332_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg333_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg334_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg335_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg336_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg337_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg338_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg339_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg340_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg341_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg342_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg343_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg344_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg345_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg346_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg347_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg348_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg349_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg350_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg351_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg352_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg353_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg354_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg355_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg356_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg357_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg358_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg359_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg360_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg361_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg362_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg363_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg364_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg365_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg366_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg367_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg368_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg369_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg370_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg371_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg372_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg373_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg374_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg375_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg376_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg377_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg378_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg379_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg380_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg381_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg382_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg383_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg384_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg385_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg386_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg387_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg388_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg389_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg390_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg391_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg392_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg393_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg394_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg395_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg396_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg397_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg398_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg399_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg400_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg401_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg402_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg403_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg404_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg405_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg406_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg407_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg408_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg409_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg410_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg411_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg412_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg413_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg414_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg415_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg416_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg417_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg418_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg419_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg420_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg421_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg422_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg423_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg424_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg425_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg426_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg427_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg428_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg429_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg430_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg431_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg432_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg433_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg434_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg435_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg436_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg437_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg438_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg439_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg440_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg441_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg442_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg443_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg444_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg445_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg446_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg447_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg448_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg449_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg450_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg451_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg452_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg453_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg454_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg455_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg456_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg457_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg458_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg459_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg460_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg461_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg462_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg463_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg464_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg465_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg466_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg467_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg468_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg469_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg470_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg471_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg472_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg473_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg474_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg475_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg476_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg477_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg478_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg479_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg480_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg481_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg482_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg483_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg484_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg485_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg486_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg487_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg488_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg489_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg490_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg491_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg492_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg493_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg494_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg495_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg496_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg497_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg498_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg499_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg500_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg501_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg502_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg503_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg504_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg505_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg506_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg507_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg508_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg509_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg510_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg511_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg512_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg513_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg514_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg515_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg516_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg517_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg518_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg519_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg520_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg521_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg522_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg523_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg524_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg525_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg526_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg527_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg528_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg529_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg530_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg531_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg532_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg533_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg534_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg535_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg536_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg537_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg538_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg539_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg540_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg541_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg542_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg543_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg544_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg545_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg546_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg547_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg548_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg549_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg550_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg551_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg552_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg553_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg554_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg555_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg556_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg557_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg558_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg559_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg560_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg561_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg562_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg563_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg564_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg565_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg566_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg567_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg568_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg569_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg570_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg571_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg572_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg573_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg574_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg575_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg576_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg577_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg578_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg579_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg580_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg581_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg582_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg583_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg584_1 = rand_strided((6, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg585_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg586_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg587_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg588_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg589_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg590_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg591_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg592_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg593_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg594_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg595_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg596_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg597_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg598_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg599_1 = rand_strided((1152, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg600_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg601_1 = rand_strided((4608, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg602_1 = rand_strided((4608, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg603_1 = rand_strided((1152, 4608), (4608, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg604_1 = rand_strided((1152, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg605_1 = rand_strided((2, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg606_1 = rand_strided((32, 1152), (1152, 1), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] arg607_1 = rand_strided((32, ), (1, ), device='cuda:0', dtype=torch.bfloat16) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] fn = lambda: call([arg0_1, arg1_1, arg2_1, arg3_1, arg4_1, arg5_1, arg6_1, arg7_1, arg8_1, arg9_1, arg10_1, arg11_1, arg12_1, arg13_1, arg14_1, arg15_1, arg16_1, arg17_1, arg18_1, arg19_1, arg20_1, arg21_1, arg22_1, arg23_1, arg24_1, arg25_1, arg26_1, arg27_1, arg28_1, arg29_1, arg30_1, arg31_1, arg32_1, arg33_1, arg34_1, arg35_1, arg36_1, arg37_1, arg38_1, arg39_1, arg40_1, arg41_1, arg42_1, arg43_1, arg44_1, arg45_1, arg46_1, arg47_1, arg48_1, arg49_1, arg50_1, arg51_1, arg52_1, arg53_1, arg54_1, arg55_1, arg56_1, arg57_1, arg58_1, arg59_1, arg60_1, arg61_1, arg62_1, arg63_1, arg64_1, arg65_1, arg66_1, arg67_1, arg68_1, arg69_1, arg70_1, arg71_1, arg72_1, arg73_1, arg74_1, arg75_1, arg76_1, arg77_1, arg78_1, arg79_1, arg80_1, arg81_1, arg82_1, arg83_1, arg84_1, arg85_1, arg86_1, arg87_1, arg88_1, arg89_1, arg90_1, arg91_1, arg92_1, arg93_1, arg94_1, arg95_1, arg96_1, arg97_1, arg98_1, arg99_1, arg100_1, arg101_1, arg102_1, arg103_1, arg104_1, arg105_1, arg106_1, arg107_1, arg108_1, arg109_1, arg110_1, arg111_1, arg112_1, arg113_1, arg114_1, arg115_1, arg116_1, arg117_1, arg118_1, arg119_1, arg120_1, arg121_1, arg122_1, arg123_1, arg124_1, arg125_1, arg126_1, arg127_1, arg128_1, arg129_1, arg130_1, arg131_1, arg132_1, arg133_1, arg134_1, arg135_1, arg136_1, arg137_1, arg138_1, arg139_1, arg140_1, arg141_1, arg142_1, arg143_1, arg144_1, arg145_1, arg146_1, arg147_1, arg148_1, arg149_1, arg150_1, arg151_1, arg152_1, arg153_1, arg154_1, arg155_1, arg156_1, arg157_1, arg158_1, arg159_1, arg160_1, arg161_1, arg162_1, arg163_1, arg164_1, arg165_1, arg166_1, arg167_1, arg168_1, arg169_1, arg170_1, arg171_1, arg172_1, arg173_1, arg174_1, arg175_1, arg176_1, arg177_1, arg178_1, arg179_1, arg180_1, arg181_1, arg182_1, arg183_1, arg184_1, arg185_1, arg186_1, arg187_1, arg188_1, arg189_1, arg190_1, arg191_1, arg192_1, arg193_1, arg194_1, arg195_1, arg196_1, arg197_1, arg198_1, arg199_1, arg200_1, arg201_1, arg202_1, arg203_1, arg204_1, arg205_1, arg206_1, arg207_1, arg208_1, arg209_1, arg210_1, arg211_1, arg212_1, arg213_1, arg214_1, arg215_1, arg216_1, arg217_1, arg218_1, arg219_1, arg220_1, arg221_1, arg222_1, arg223_1, arg224_1, arg225_1, arg226_1, arg227_1, arg228_1, arg229_1, arg230_1, arg231_1, arg232_1, arg233_1, arg234_1, arg235_1, arg236_1, arg237_1, arg238_1, arg239_1, arg240_1, arg241_1, arg242_1, arg243_1, arg244_1, arg245_1, arg246_1, arg247_1, arg248_1, arg249_1, arg250_1, arg251_1, arg252_1, arg253_1, arg254_1, arg255_1, arg256_1, arg257_1, arg258_1, arg259_1, arg260_1, arg261_1, arg262_1, arg263_1, arg264_1, arg265_1, arg266_1, arg267_1, arg268_1, arg269_1, arg270_1, arg271_1, arg272_1, arg273_1, arg274_1, arg275_1, arg276_1, arg277_1, arg278_1, arg279_1, arg280_1, arg281_1, arg282_1, arg283_1, arg284_1, arg285_1, arg286_1, arg287_1, arg288_1, arg289_1, arg290_1, arg291_1, arg292_1, arg293_1, arg294_1, arg295_1, arg296_1, arg297_1, arg298_1, arg299_1, arg300_1, arg301_1, arg302_1, arg303_1, arg304_1, arg305_1, arg306_1, arg307_1, arg308_1, arg309_1, arg310_1, arg311_1, arg312_1, arg313_1, arg314_1, arg315_1, arg316_1, arg317_1, arg318_1, arg319_1, arg320_1, arg321_1, arg322_1, arg323_1, arg324_1, arg325_1, arg326_1, arg327_1, arg328_1, arg329_1, arg330_1, arg331_1, arg332_1, arg333_1, arg334_1, arg335_1, arg336_1, arg337_1, arg338_1, arg339_1, arg340_1, arg341_1, arg342_1, arg343_1, arg344_1, arg345_1, arg346_1, arg347_1, arg348_1, arg349_1, arg350_1, arg351_1, arg352_1, arg353_1, arg354_1, arg355_1, arg356_1, arg357_1, arg358_1, arg359_1, arg360_1, arg361_1, arg362_1, arg363_1, arg364_1, arg365_1, arg366_1, arg367_1, arg368_1, arg369_1, arg370_1, arg371_1, arg372_1, arg373_1, arg374_1, arg375_1, arg376_1, arg377_1, arg378_1, arg379_1, arg380_1, arg381_1, arg382_1, arg383_1, arg384_1, arg385_1, arg386_1, arg387_1, arg388_1, arg389_1, arg390_1, arg391_1, arg392_1, arg393_1, arg394_1, arg395_1, arg396_1, arg397_1, arg398_1, arg399_1, arg400_1, arg401_1, arg402_1, arg403_1, arg404_1, arg405_1, arg406_1, arg407_1, arg408_1, arg409_1, arg410_1, arg411_1, arg412_1, arg413_1, arg414_1, arg415_1, arg416_1, arg417_1, arg418_1, arg419_1, arg420_1, arg421_1, arg422_1, arg423_1, arg424_1, arg425_1, arg426_1, arg427_1, arg428_1, arg429_1, arg430_1, arg431_1, arg432_1, arg433_1, arg434_1, arg435_1, arg436_1, arg437_1, arg438_1, arg439_1, arg440_1, arg441_1, arg442_1, arg443_1, arg444_1, arg445_1, arg446_1, arg447_1, arg448_1, arg449_1, arg450_1, arg451_1, arg452_1, arg453_1, arg454_1, arg455_1, arg456_1, arg457_1, arg458_1, arg459_1, arg460_1, arg461_1, arg462_1, arg463_1, arg464_1, arg465_1, arg466_1, arg467_1, arg468_1, arg469_1, arg470_1, arg471_1, arg472_1, arg473_1, arg474_1, arg475_1, arg476_1, arg477_1, arg478_1, arg479_1, arg480_1, arg481_1, arg482_1, arg483_1, arg484_1, arg485_1, arg486_1, arg487_1, arg488_1, arg489_1, arg490_1, arg491_1, arg492_1, arg493_1, arg494_1, arg495_1, arg496_1, arg497_1, arg498_1, arg499_1, arg500_1, arg501_1, arg502_1, arg503_1, arg504_1, arg505_1, arg506_1, arg507_1, arg508_1, arg509_1, arg510_1, arg511_1, arg512_1, arg513_1, arg514_1, arg515_1, arg516_1, arg517_1, arg518_1, arg519_1, arg520_1, arg521_1, arg522_1, arg523_1, arg524_1, arg525_1, arg526_1, arg527_1, arg528_1, arg529_1, arg530_1, arg531_1, arg532_1, arg533_1, arg534_1, arg535_1, arg536_1, arg537_1, arg538_1, arg539_1, arg540_1, arg541_1, arg542_1, arg543_1, arg544_1, arg545_1, arg546_1, arg547_1, arg548_1, arg549_1, arg550_1, arg551_1, arg552_1, arg553_1, arg554_1, arg555_1, arg556_1, arg557_1, arg558_1, arg559_1, arg560_1, arg561_1, arg562_1, arg563_1, arg564_1, arg565_1, arg566_1, arg567_1, arg568_1, arg569_1, arg570_1, arg571_1, arg572_1, arg573_1, arg574_1, arg575_1, arg576_1, arg577_1, arg578_1, arg579_1, arg580_1, arg581_1, arg582_1, arg583_1, arg584_1, arg585_1, arg586_1, arg587_1, arg588_1, arg589_1, arg590_1, arg591_1, arg592_1, arg593_1, arg594_1, arg595_1, arg596_1, arg597_1, arg598_1, arg599_1, arg600_1, arg601_1, arg602_1, arg603_1, arg604_1, arg605_1, arg606_1, arg607_1]) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] return print_performance(fn, times=times, repeat=repeat) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] if __name__ == "__main__": +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] from torch._inductor.wrapper_benchmark import compiled_module_main +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] compiled_module_main('None', benchmark_compiled_module) +V0808 07:22:57.587218 1260637 torch/_inductor/graph.py:1780] [0/0] [__output_code] +I0808 07:23:14.171741 1260637 torch/_inductor/graph.py:1814] [0/0] [__output_code] Output code written to: /tmp/torchinductor_sayak/67/c67lqfgzdp5sk2m7yauch3ri2zntwq5epdl6a47463236u4r7hoh.py