upload auto_gptq format

Signed-off-by: n1ck-guo <[email protected]>

config.json

@@ -139,11 +139,12 @@
   "quantization_config": {
     "amp": true,
     "autoround_version": "0.4.3",
-    "backend": "auto_round:gptq:exllamav2",
     "batch_size": 8,
     "bits": 4,
+    "block_name_to_quantize": "language.model.layers",
+    "damp_percent": 0.01,
     "data_type": "int",
-    "dataset": "NeelNanda/pile-10k",
+    "desc_act": false,
     "enable_minmax_tuning": true,
     "enable_norm_bias_tuning": false,
     "enable_quanted_input": true,
@@ -154,11 +155,11 @@
     "lr": 0.001,
     "minmax_lr": 0.001,
     "nsamples": 512,
-    "quant_method": "intel/auto-round",
+    "quant_method": "gptq",
     "scale_dtype": "torch.float16",
     "seqlen": 2048,
     "sym": true,
-    "to_quant_block_names": "language.model.layers"
+    "true_sequential": false
   },
   "tile_tag": "2D",
   "torch_dtype": "bfloat16",

quantize_config.json

@@ -0,0 +1,25 @@
+{
+  "bits": 4,
+  "group_size": 128,
+  "sym": true,
+  "data_type": "int",
+  "enable_quanted_input": true,
+  "enable_minmax_tuning": true,
+  "seqlen": 2048,
+  "batch_size": 8,
+  "scale_dtype": "torch.float16",
+  "lr": 0.001,
+  "minmax_lr": 0.001,
+  "gradient_accumulate_steps": 1,
+  "iters": 1000,
+  "amp": true,
+  "nsamples": 512,
+  "low_gpu_mem_usage": true,
+  "enable_norm_bias_tuning": false,
+  "autoround_version": "0.4.3",
+  "block_name_to_quantize": "language.model.layers",
+  "quant_method": "gptq",
+  "desc_act": false,
+  "true_sequential": false,
+  "damp_percent": 0.01
+}