quantized_model (#1)

- quanitze jais-13b-chat to w_fp8_a_fp8 (2685f14274138a184129e4e679b1601b4f6eecee)


Co-authored-by: hang yang <[email protected]>

README.md

@@ -1,3 +1,78 @@
----
-license: apache-2.0
----
+---
+license: apache-2.0
+base_model: inceptionai/jais-13b-chat
+---
+
+
+# amd/jais-13b-chat-FP8
+- ## Introduction
+  This model was created by applying [Quark](https://quark.docs.amd.com/latest/index.html) with calibration samples from Pile dataset.
+- ## Quantization Stragegy
+  - ***Quantized Layers***: All linear layers excluding "lm_head"
+  - ***Weight***: FP8 symmetric per-tensor
+  - ***Activation***: FP8 symmetric per-tensor
+  - ***KV Cache***: FP8 symmetric  per-tensor
+- ## Quick Start
+1. [Download and install Quark](https://quark.docs.amd.com/latest/install.html)
+2. Run the quantization script in the example folder using the following command line:
+```sh
+export MODEL_DIR = [local model checkpoint folder] or inceptionai/jais-13b-chat 
+# single GPU
+python3 quantize_quark.py \ 
+        --model_dir $MODEL_DIR \
+        --output_dir jais-13b-chat \                           
+        --quant_scheme w_fp8_a_fp8 \
+        --kv_cache_dtype fp8 \
+        --num_calib_data 128 \
+        --model_export quark_safetensors \
+        --no_weight_matrix_merge
+# If model size is too large for single GPU, please use multi GPU instead.
+python3 quantize_quark.py \ 
+        --model_dir $MODEL_DIR \
+        --output_dir jais-13b-chat \                           
+        --quant_scheme w_fp8_a_fp8 \
+        --kv_cache_dtype fp8 \
+        --num_calib_data 128 \
+        --model_export quark_safetensors \
+        --no_weight_matrix_merge \
+        --multi_gpu
+```
+## Deployment
+Quark has its own export format and allows FP8 quantized models to be efficiently deployed using the vLLM backend(vLLM-compatible).
+## Evaluation
+Quark currently uses perplexity(PPL) as the evaluation metric for accuracy loss before and after quantization.The specific PPL algorithm can be referenced in the quantize_quark.py.
+The quantization evaluation results are conducted in pseudo-quantization mode, which may slightly differ from the actual quantized inference accuracy. These results are provided for reference only.
+#### Evaluation scores
+<table>
+  <tr>
+   <td><strong>Benchmark</strong>
+   </td>
+   <td><strong>jais-13b-chat</strong>
+   </td>
+   <td><strong>jais-13b-chat(this model)</strong>
+   </td>
+  </tr>
+  <tr>
+   <td>Perplexity-wikitext2
+   </td>
+   <td> 11.6331
+   </td>
+   <td> 13.6648
+   </td>
+  </tr>
+</table>
+
+#### License
+Modifications copyright(c) 2024 Advanced Micro Devices,Inc. All rights reserved.
+
+Licensed under the Apache License, Version 2.0 (the "License");
+you may not use this file except in compliance with the License.
+You may obtain a copy of the License at
+
+    http://www.apache.org/licenses/LICENSE-2.0
+
+Unless required by applicable law or agreed to in writing, software
+distributed under the License is distributed on an "AS IS" BASIS,
+WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+See the License for the specific language governing permissions and
+limitations under the License.

config.json

@@ -0,0 +1,48 @@
+{
+  "_name_or_path": "/root/project/Quark/examples/torch/language_modeling/jais-13b-chat",
+  "activation_function": "swiglu",
+  "architectures": [
+    "JAISLMHeadModel"
+  ],
+  "attn_pdrop": 0.0,
+  "auto_map": {
+    "AutoConfig": "configuration_jais.JAISConfig",
+    "AutoModel": "modeling_jais.JAISModel",
+    "AutoModelForCausalLM": "modeling_jais.JAISLMHeadModel",
+    "AutoModelForQuestionAnswering": "modeling_jais.JAISForQuestionAnswering",
+    "AutoModelForSequenceClassification": "modeling_jais.JAISForSequenceClassification",
+    "AutoModelForTokenClassification": "modeling_jais.JAISForTokenClassification"
+  },
+  "bos_token_id": 0,
+  "embd_pdrop": 0.0,
+  "embeddings_scale": 14.6,
+  "eos_token_id": 0,
+  "initializer_range": 0.02,
+  "layer_norm_epsilon": 1e-05,
+  "model_type": "jais",
+  "n_embd": 5120,
+  "n_head": 40,
+  "n_inner": 13653,
+  "n_layer": 40,
+  "n_positions": 2048,
+  "pad_token_id": 0,
+  "position_embedding_type": "alibi",
+  "quantization_config": {
+    "activation_scheme": "static",
+    "ignored_layers": [
+      "lm_head"
+    ],
+    "kv_cache_scheme": "static",
+    "quant_method": "fp8"
+  },
+  "reorder_and_upcast_attn": false,
+  "resid_pdrop": 0.0,
+  "scale_attn_by_inverse_layer_idx": false,
+  "scale_attn_weights": true,
+  "scale_qk_dot_by_d": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.45.2",
+  "use_cache": true,
+  "vocab_size": 84992,
+  "width_scale": 0.11100000000000002
+}

configuration_jais.py

@@ -0,0 +1,143 @@
+# coding=utf-8
+# Copyright 2023 The OpenAI Team Authors and HuggingFace Inc. team.
+# Copyright (c) 2018, NVIDIA CORPORATION.  All rights reserved.
+# Copyright 2023 Cerebras Systems.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+""" JAIS configuration"""
+
+from transformers.configuration_utils import PretrainedConfig
+from transformers.utils import logging
+
+
+logger = logging.get_logger(__name__)
+
+class JAISConfig(PretrainedConfig):
+    """
+    This is the configuration class to store the configuration of a [`JAISModel`]. It is used to
+    instantiate a JAIS model according to the specified arguments, defining the model architecture.
+
+    Configuration objects inherit from [`PretrainedConfig`] and can be used to control the model outputs. Read the
+    documentation from [`PretrainedConfig`] for more information.
+
+
+    Args:
+        vocab_size (`int`, *optional*, defaults to 50257):
+            Vocabulary size of the JAIS model. Defines the number of different tokens that can be represented by the
+            `inputs_ids` passed when calling [`JAISModel`].
+        n_positions (`int`, *optional*, defaults to 1024):
+            The maximum sequence length that this model might ever be used with. Typically set this to something large
+            just in case (e.g., 512 or 1024 or 2048).
+        n_embd (`int`, *optional*, defaults to 768):
+            Dimensionality of the embeddings and hidden states.
+        n_layer (`int`, *optional*, defaults to 12):
+            Number of hidden layers in the Transformer encoder.
+        n_head (`int`, *optional*, defaults to 12):
+            Number of attention heads for each attention layer in the Transformer encoder.
+        n_inner (`int`, *optional*, defaults to None):
+            Dimensionality of the inner feed-forward layers. `None` will set it to 4 times n_embd
+        activation_function (`str`, *optional*, defaults to `"gelu"`):
+            Activation function, to be selected in the list `["relu", "silu", "gelu", "tanh", "gelu_new", "swiglu"]`.
+        resid_pdrop (`float`, *optional*, defaults to 0.1):
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
+        embd_pdrop (`float`, *optional*, defaults to 0.1):
+            The dropout ratio for the embeddings.
+        attn_pdrop (`float`, *optional*, defaults to 0.1):
+            The dropout ratio for the attention.
+        layer_norm_epsilon (`float`, *optional*, defaults to 1e-5):
+            The epsilon to use in the layer normalization layers.
+        initializer_range (`float`, *optional*, defaults to 0.02):
+            The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
+        scale_attn_weights (`bool`, *optional*, defaults to `True`):
+            Scale attention weights by dividing by sqrt(hidden_size)..
+        use_cache (`bool`, *optional*, defaults to `True`):
+            Whether or not the model should return the last key/values attentions (not used by all models).
+        scale_attn_by_inverse_layer_idx (`bool`, *optional*, defaults to `False`):
+            Whether to additionally scale attention weights by `1 / layer_idx + 1`.
+        reorder_and_upcast_attn (`bool`, *optional*, defaults to `False`):
+            Whether to scale keys (K) prior to computing attention (dot-product) and upcast attention
+            dot-product/softmax to float() when training with mixed precision.
+        position_embedding_type (`str`, *optional*, defaults to `"learned"`):
+            Positional embedding can be either `"alibi"` or `"learned"`.
+        width_scale (`float`, *optional*, defaults to 1.0):
+            muP parameter to scale output logits and initializers. Calculated as (`d_model,0 / d_model`),
+            where `d_model` is the model's width and `d_model,0` is the proxy model's width.
+        embeddings_scale (`float`, *optional*, defaults to 1.0):
+            muP parameter to scale token and position embeddings.
+        scale_qk_dot_by_d (`bool`, *optional*, defaults to `False`):
+            Scale attention weights by dividing by hidden_size instead of sqrt(hidden_size).
+            Need to set scale_attn_weights to `True` as well.
+
+    """
+
+    model_type = "jais"
+    keys_to_ignore_at_inference = ["past_key_values"]
+    attribute_map = {
+        "hidden_size": "n_embd",
+        "max_position_embeddings": "n_positions",
+        "num_attention_heads": "n_head",
+        "num_hidden_layers": "n_layer",
+    }
+
+    def __init__(
+        self,
+        vocab_size=50257,
+        n_positions=1024,
+        n_embd=768,
+        n_layer=12,
+        n_head=12,
+        n_inner=None,
+        activation_function="gelu_new",
+        resid_pdrop=0.1,
+        embd_pdrop=0.1,
+        attn_pdrop=0.1,
+        layer_norm_epsilon=1e-5,
+        initializer_range=0.02,
+        scale_attn_weights=True,
+        use_cache=True,
+        bos_token_id=50256,
+        eos_token_id=50256,
+        scale_attn_by_inverse_layer_idx=False,
+        reorder_and_upcast_attn=False,
+        position_embedding_type="learned",
+        width_scale=1.0,
+        embeddings_scale=1.0,
+        scale_qk_dot_by_d=False,
+        **kwargs,
+    ):
+        self.vocab_size = vocab_size
+        self.n_positions = n_positions
+        self.n_embd = n_embd
+        self.n_layer = n_layer
+        self.n_head = n_head
+        self.n_inner = n_inner
+        self.activation_function = activation_function
+        self.resid_pdrop = resid_pdrop
+        self.embd_pdrop = embd_pdrop
+        self.attn_pdrop = attn_pdrop
+        self.layer_norm_epsilon = layer_norm_epsilon
+        self.initializer_range = initializer_range
+        self.scale_attn_weights = scale_attn_weights
+        self.use_cache = use_cache
+        self.scale_attn_by_inverse_layer_idx = scale_attn_by_inverse_layer_idx
+        self.reorder_and_upcast_attn = reorder_and_upcast_attn
+
+        self.bos_token_id = bos_token_id
+        self.eos_token_id = eos_token_id
+
+        self.position_embedding_type = position_embedding_type
+        self.width_scale = width_scale
+        self.embeddings_scale = embeddings_scale
+        self.scale_qk_dot_by_d = scale_qk_dot_by_d
+
+        super().__init__(bos_token_id=bos_token_id, eos_token_id=eos_token_id, **kwargs)

generation_config.json

@@ -0,0 +1,7 @@
+{
+  "_from_model_config": true,
+  "bos_token_id": 0,
+  "eos_token_id": 0,
+  "pad_token_id": 0,
+  "transformers_version": "4.45.2"
+}

model-00001-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:35fe8a885b7d64d0d2493cfb7dcb27d3e2fc6b2fb726f2ee577e330060a35763
+size 4994297084

model-00002-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:54cf788ebab72c5869e905cae79bb94bf467907c1ec0270cc187c6820a18f1e2
+size 4932839768

model-00003-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:730c918b3bc29d006129f3a947f5dafef0f10216d07757b9b336eb4e17de64dd
+size 4408093764

model-00004-of-00004.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c3b1a6451ac2d36c3d06500ac6aff22870d6d329e18807feb69701403f2bf4f2
+size 1740636288

model.safetensors.index.json

@@ -0,0 +1,1012 @@
+{
+  "metadata": {
+    "total_size": 16075764160
+  },
+  "weight_map": {
+    "lm_head.weight": "model-00004-of-00004.safetensors",
+    "transformer.h.0.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.lh.0.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.0.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.0.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.0.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.1.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.10.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.10.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.10.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.10.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.10.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.10.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.11.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.12.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.13.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.14.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.15.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.16.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.17.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.18.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.19.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.2.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.2.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.20.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.20.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.21.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.22.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.23.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.24.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_attn.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_attn.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_attn.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_attn.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.ln_2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.ln_2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc2.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc2.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc2.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_fc2.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_proj.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_proj.input_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_proj.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.25.mlp.c_proj.weight_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.26.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.attn.kv_scale": "model-00002-of-00004.safetensors",
+    "transformer.h.26.ln_1.bias": "model-00002-of-00004.safetensors",
+    "transformer.h.26.ln_1.weight": "model-00002-of-00004.safetensors",
+    "transformer.h.26.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.26.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.26.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.27.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.28.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.29.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.3.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.3.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.30.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.30.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.31.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.32.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.33.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.34.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.35.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.36.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.37.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.38.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_attn.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_attn.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_attn.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_attn.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.attn.kv_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.ln_1.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.ln_1.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.ln_2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.ln_2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc2.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc2.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc2.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_fc2.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_proj.bias": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_proj.input_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_proj.weight": "model-00003-of-00004.safetensors",
+    "transformer.h.39.mlp.c_proj.weight_scale": "model-00003-of-00004.safetensors",
+    "transformer.h.4.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.4.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.5.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.6.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.7.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.8.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_attn.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_attn.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_attn.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_attn.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.attn.kv_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.ln_1.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.ln_1.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.ln_2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.ln_2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc2.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc2.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc2.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_fc2.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_proj.bias": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_proj.input_scale": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_proj.weight": "model-00001-of-00004.safetensors",
+    "transformer.h.9.mlp.c_proj.weight_scale": "model-00001-of-00004.safetensors",
+    "transformer.ln_f.bias": "model-00003-of-00004.safetensors",
+    "transformer.ln_f.weight": "model-00003-of-00004.safetensors",
+    "transformer.relative_pe.slopes": "model-00003-of-00004.safetensors",
+    "transformer.wte.weight": "model-00001-of-00004.safetensors"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<|endoftext|>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,19 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "<|endoftext|>",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "bos_token": "<|endoftext|>",
+  "clean_up_tokenization_spaces": true,
+  "eos_token": "<|endoftext|>",
+  "model_max_length": 2048,
+  "pad_token": "<|endoftext|>",
+  "tokenizer_class": "PreTrainedTokenizerFast",
+  "unk_token": "<|endoftext|>"
+}