Hugging Face's logo

Vykyan
/
FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview-ao-int4wo-gs128

PyTorch
qwen2
torchao-my-repo
torchao
Model card Files Community
Vykyan commited on
Commit
ed12f93
·
verified ·
1 Parent(s): 2a9bc81

Upload folder using huggingface_hub

Browse files
Files changed (11) hide show
  1. .gitattributes +1 -0
  2. README.md +304 -0
  3. config.json +53 -0
  4. pytorch_model-00001-of-00004.bin +3 -0
  5. pytorch_model-00002-of-00004.bin +3 -0
  6. pytorch_model-00003-of-00004.bin +3 -0
  7. pytorch_model-00004-of-00004.bin +3 -0
  8. pytorch_model.bin.index.json +777 -0
  9. special_tokens_map.json +23 -0
  10. tokenizer.json +3 -0
  11. tokenizer_config.json +195 -0
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,304 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview
4
+ license: apache-2.0
5
+ tags:
6
+ - torchao-my-repo
7
+ ---
8
+ # FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview (Quantized)
9
+
10
+ ## Description
11
+ This model is a quantized version of the original model [`FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview`](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview).
12
+
13
+ It's quantized using the TorchAO library using the [torchao-my-repo](https://huggingface.co/spaces/pytorch/torchao-my-repo) space.
14
+
15
+ ## Quantization Details
16
+ - **Quantization Type**: Int4WeightOnly
17
+ - **Group Size**: 128
18
+
19
+
20
+
21
+ # 📄 Original Model Information
22
+
23
+
24
+ <p align="center" width="100%">
25
+ </p>
26
+
27
+ <div id="top" align="center">
28
+
29
+ FuseO1-Preview: System-II Reasoning Fusion of LLMs
30
+ -----------------------------
31
+
32
+ <h4> |<a href="https://arxiv.org/abs/2408.07990"> 📑 Paper </a> |
33
+ <a href="https://github.com/fanqiwan/FuseAI"> 🐱 GitHub Repo </a> |
34
+ <a href="https://huggingface.co/FuseAI"> 🤗 Hugging Face </a> |
35
+ <a href="https://huggingface.co/blog/Wanfq/fuseo1-preview"> 🌐 Blog </a> |
36
+ </h4>
37
+
38
+ <!-- **Authors:** -->
39
+
40
+ _Fanqi Wan, Longguang Zhong, Ziyi Yang, Weizhou Shen, Xinting Huang_
41
+
42
+
43
+ <!-- **Affiliations:** -->
44
+
45
+ _FuseAI Team_
46
+
47
+ </div>
48
+
49
+ <p align="center">
50
+ <img src="./assets/fuseo1-preview.jpg" width="100%"> <br>
51
+ </p>
52
+
53
+
54
+ ## Overview
55
+
56
+ [FuseO1-Preview](https://huggingface.co/collections/FuseAI/fuseo1-preview-678eb56093649b2688bc9977) is our initial endeavor to enhance the System-II reasoning capabilities of large language models (LLMs) through innovative model fusion techniques. By employing our advanced [SCE](https://arxiv.org/abs/2408.07990) merging methodologies, we integrate multiple open-source o1-like LLMs into a unified model. Our goal is to incorporate the distinct knowledge and strengths from different reasoning LLMs into a single, unified model with strong System-II reasoning abilities, particularly in mathematics, coding, and science domains.
57
+
58
+ <p align="center">
59
+ <img src="./assets/sce.jpg" width="70%"> <br>
60
+ </p>
61
+
62
+ To achieve this, we conduct two types of model merging:
63
+
64
+ - **Long-Long Reasoning Merging**: This approach involves model fusion across LLMs that utilize long-CoT reasoning, with the goal of enhancing long-CoT reasoning capabilities. The resulted [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview) achieves a Pass@1 accuracy of **74.0 on AIME24**, demonstrating significant performance improvements compared to the OpenAI o1-preview (44.6) and OpenAI o1-mini (63.4), even approaching OpenAI o1 (79.2).
65
+ - **Long-Short Reasoning Merging**: This approach involves model fusion between long-CoT and short-CoT LLMs, aiming to improve reasoning capabilities in both long and short reasoning processes. The resulted [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview) and [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview) is capable of utilizing both long and short reasoning processes and demonstrates relatively strong performance in long reasoning tasks.
66
+
67
+ | Model | Merge Type | Source Models | HF Link |
68
+ |:----- | ---- | ---- | ---- |
69
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview) | Long-Long Reasoning Merge | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B), [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview), [NovaSky-AI/Sky-T1-32B-Preview](https://huggingface.co/NovaSky-AI/Sky-T1-32B-Preview) | [🤗 Hugging Face](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview), [GGUF](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview-GGUF) |
70
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview) | Long-Long Reasoning Merge | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B), [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview) | [🤗 Hugging Face](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview) |
71
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview) | Long-Short Reasoning Merge | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B), [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview), [NovaSky-AI/Sky-T1-32B-Flash](https://huggingface.co/NovaSky-AI/Sky-T1-32B-Flash) | [🤗 Hugging Face](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview) |
72
+ | [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview) | Long-Short Reasoning Merge | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B), [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) | [🤗 Hugging Face](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview) |
73
+ | [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview) | Long-Short Reasoning Merge | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B), [Qwen/Qwen2.5-32B-Coder](https://huggingface.co/Qwen/Qwen2.5-32B-Coder) | [🤗 Hugging Face](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview) |
74
+
75
+
76
+ ## Long-Long Reasoning Merging
77
+
78
+ We conduct experiments on these folloing long-cot LLMs.
79
+
80
+ - [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B)
81
+ - [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview)
82
+ - [NovaSky-AI/Sky-T1-32B-Preview](https://huggingface.co/NovaSky-AI/Sky-T1-32B-Preview)
83
+
84
+ To reproduce the merged [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview) model, using the script below.
85
+
86
+ ```sh
87
+ cd FuseAI/FuseO1-Preview/mergekit
88
+ pip3 install -e .
89
+ model_save_dir=xx # your path to save the merged models
90
+ mergekit-yaml fuseo1_configs/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview.yaml ${model_save_dir}/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview --cudas
91
+ ```
92
+
93
+ To reproduce the merged [FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview) model, using the script below.
94
+
95
+ ```sh
96
+ cd FuseAI/FuseO1-Preview/mergekit
97
+ pip3 install -e .
98
+ model_save_dir=xxx # your path to save the merged models
99
+ mergekit-yaml fuseo1_configs/FuseO1-DeepSeekR1-QwQ-32B-Preview.yaml ${model_save_dir}/FuseO1-DeepSeekR1-QwQ-32B-Preview --cuda
100
+ ```
101
+
102
+ We provide the example code to use FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview.
103
+
104
+ ```python3
105
+ from vllm import LLM, SamplingParams
106
+
107
+ llm = LLM(model="FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview", tensor_parallel_size=8)
108
+ sampling_params = SamplingParams(max_tokens=32768, temperature=0.7, stop=["<|im_end|>", "<｜end▁of▁sentence｜>"], stop_token_ids=[151645, 151643])
109
+
110
+ conversations = [
111
+ [
112
+ {"role": "system", "content": "Please reason step by step, and put your final answer within \\boxed{{}}."},
113
+ {"role": "user", "content": "Quadratic polynomials $P(x)$ and $Q(x)$ have leading coefficients $2$ and $-2,$ respectively. The graphs of both polynomials pass through the two points $(16,54)$ and $(20,53).$ Find $P(0) + Q(0).$."},
114
+ ],
115
+ ]
116
+
117
+ responses = llm.chat(messages=conversations, sampling_params=sampling_params, use_tqdm=True)
118
+
119
+ for response in responses:
120
+ print(response.outputs[0].text.strip())
121
+ ```
122
+
123
+ ## Long-Short Reasoning Merging
124
+
125
+ We conduct experiments on these folloing long-cot and short-cot LLMs.
126
+
127
+ - [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B)
128
+ - [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct)
129
+ - [Qwen/Qwen2.5-32B-Coder](https://huggingface.co/Qwen/Qwen2.5-32B-Coder)
130
+
131
+ To reproduce the merged [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview) model, using the script below.
132
+
133
+ ```sh
134
+ cd FuseAI/FuseO1-Preview/mergekit
135
+ pip3 install -e .
136
+ model_save_dir=xxx # your path to save the merged models
137
+ mergekit-yaml fuseo1_configs/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview.yaml ${model_save_dir}/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview --cuda
138
+ ```
139
+
140
+ To reproduce the merged [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview) model, using the script below.
141
+
142
+ ```sh
143
+ cd FuseAI/FuseO1-Preview/mergekit
144
+ pip3 install -e .
145
+ model_save_dir=xxx # your path to save the merged models
146
+ mergekit-yaml fuseo1_configs/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview.yaml ${model_save_dir}/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview --cuda
147
+ ```
148
+
149
+ To reproduce the merged [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview) model, using the script below.
150
+
151
+ ```sh
152
+ cd FuseAI/FuseO1-Preview/mergekit
153
+ pip3 install -e .
154
+ model_save_dir=xxx # your path to save the merged models
155
+ mergekit-yaml fuseo1_configs/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview.yaml ${model_save_dir}/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview --cuda
156
+ ```
157
+
158
+ We provide the code to use FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview.
159
+
160
+ ```python3
161
+ from vllm import LLM, SamplingParams
162
+
163
+ llm = LLM(model="FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview", tensor_parallel_size=8)
164
+ sampling_params = SamplingParams(max_tokens=32768, temperature=0.7, stop=["<|im_end|>", "<｜end▁of▁sentence｜>"], stop_token_ids=[151645, 151643])
165
+
166
+ conversations = [
167
+ [
168
+ {"role": "system", "content": "Please reason step by step, and put your final answer within \\boxed{{}}."},
169
+ {"role": "user", "content": "Quadratic polynomials $P(x)$ and $Q(x)$ have leading coefficients $2$ and $-2,$ respectively. The graphs of both polynomials pass through the two points $(16,54)$ and $(20,53).$ Find $P(0) + Q(0).$."},
170
+ ],
171
+ ]
172
+
173
+ responses = llm.chat(messages=conversations, sampling_params=sampling_params, use_tqdm=True)
174
+
175
+ for response in responses:
176
+ print(response.outputs[0].text.strip())
177
+ ```
178
+
179
+ ## Evaluation Results
180
+
181
+ We test the resulted models on three kinds of benchmarks, including **Math Reasoning**, **Code Reasoning** , and **Scientific Reasoning**.
182
+
183
+ Math Reasoning
184
+ - AIME24
185
+ - MATH500
186
+ - OlympiadBench
187
+
188
+ Scientific Reasoning
189
+ - GPQA-Diamond
190
+ - MMLU-Pro
191
+ - MMLU
192
+
193
+
194
+ Code Reasoning
195
+ - LiveCodeBench (2408-2502)
196
+
197
+ > Important Note: We manully set `"add_bos_token": false` in `tokenizer_config.json` for all the evaluated LLMs to prevent the bos_token to be added twice for each prompt. Please download and modify to ensure consistency.
198
+
199
+ ### Math Reasoning
200
+
201
+ The evaluation code is modified from [Qwen2.5-Math](https://github.com/QwenLM/Qwen2.5-Math). In our evaluation, we set the temperature to 0.6, the top-p to 0.95 and the max_tokens to 32768. We provide the example to reproduce our results in [math_evaluation](https://github.com/fanqiwan/FuseAI/tree/main/FuseO1-Preview/math_evaluation).
202
+
203
+ The system prompt for evaluation is set to:
204
+
205
+ ```sh
206
+ Please reason step by step, and put your final answer within \\boxed{{}}.
207
+ ```
208
+
209
+ The evaluation results are shown in the table below:
210
+
211
+ In our evaluation of AIME24, we follow the method from DeepSeek-R1, wherein Pass@1 is computed by averaging the results across 32 sampled responses per prompt, while Cons@32 is determined through self-consistency analysis of the same 32 sampled responses for each prompt. For other benchmarks, we only sample 1 response and report the Pass@1.
212
+
213
+ | Models | AIME24 Pass@1 | AIME24 Cons@32 | MATH500 | OlympiadBench |
214
+ |:------ | --------------| ------------------- | ------------ | -------------- |
215
+ | OpenAI o1 | 79.2 | - | 96.4 | - |
216
+ | OpenAI o1-preview | 44.6 | - | 85.5 | - |
217
+ | OpenAI o1-mini | 63.6 | - | 90.0 | - |
218
+ | DeepSeek R1 | 79.8 | - | 97.3 | - |
219
+ | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) | 69.2 | 83.3 | 93.6 | 64.3 |
220
+ | [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview) | 43.8 | 56.7 | 88.4 | 60.3 |
221
+ | [NovaSky-AI/Sky-T1-32B-Preview](https://huggingface.co/NovaSky-AI/Sky-T1-32B-Preview) | 37.7 | 50.0 | 88.0 | 55.1 |
222
+ | [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) | 17.0 | 20.0 | 81.8 | 48.1 |
223
+ | [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview) | 68.6 | 83.3 | 94.6 | 64.9 |
224
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview) | 69.7 | 83.3 | 94.6 | 64.0 |
225
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview) | 72.9 | 86.7 | - | - |
226
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview) | 74.0 | 86.7 | 94.8 | 65.0 |
227
+
228
+ We show that our merged FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview demonstrate superior performance improvements comparet to DeepSeek-R1-Distill-Qwen-32B, QwQ-32B-Preview, and Sky-T1-32B-Preview on math reasoning. Specifically, our model achieves an accuracy of **74.0 Pass@1 and 86.7 Cons@32 on AIME24**, demonstrating significant performance improvements compared to DeepSeek-R1-Distill-Qwen-32B (69.2 Pass@1 and 83.3 Cons@32), OpenAI o1-preview (44.6 Pass@1) and OpenAI o1-mini (63.4 Pass@1), even approaching OpenAI o1 (79.2 Pass@1).
229
+
230
+ ### Scientific Reasoning
231
+
232
+ The evaluation code is modified from [SkyThought](https://github.com/NovaSky-AI/SkyThought). In our evaluation, we set the temperature to 0.7 and the max_tokens to 32768. We provide the example to reproduce our results in [evaluation](https://github.com/fanqiwan/FuseAI/tree/main/FuseO1-Preview/evaluation).
233
+
234
+ The system prompt for evaluation is set to:
235
+
236
+ ```sh
237
+ You are a helpful and harmless assistant. You should think step-by-step.
238
+ ```
239
+
240
+ The evaluation results are shown in the table below:
241
+
242
+ | Models | GPQA-Diamond| MMLU-Pro | MMLU |
243
+ |:------ | --------------| ------------ | -------------- |
244
+ | OpenAI o1 | 75.7 | - | 91.8 |
245
+ | OpenAI o1-preview | 73.3 | - | 90.8 |
246
+ | OpenAI o1-mini | 60.0 | 80.3 | 85.2 |
247
+ | DeepSeek R1 | 71.5 | 84.0 | 90.8 |
248
+ | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) | 57.6 | 68.7 | 82.2 |
249
+ | [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview) | 49.5 | 63.5 | 85.2 |
250
+ | [NovaSky-AI/Sky-T1-32B-Preview](https://huggingface.co/NovaSky-AI/Sky-T1-32B-Preview) | 50.5 | 65.8 | 82.7 |
251
+ | [Qwen/Qwen2.5-32B-Instruct](https://huggingface.co/Qwen/Qwen2.5-32B-Instruct) | 46.5 | 56.3 | 79.6 |
252
+ | [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Instruct-32B-Preview) | 55.1 | 68.6 | 82.0 |
253
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview) | 62.1 | 68.9 | 82.7 |
254
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview) | 54.6 | 70.6 | 84.0 |
255
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview) | 62.1 | 70.8 | 83.6 |
256
+
257
+ We show that our merged FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview demonstrate superior performance improvements comparet to DeepSeek-R1-Distill-Qwen-32B, QwQ-32B-Preview, and Sky-T1-32B-Preview on scientific reasoning. Specifically, our model achieves an accuracy of **62.1 on GPQA-Diamond and 70.8 on MMLU-Pro**, demonstrating significant performance improvements compared to DeepSeek-R1-Distill-Qwen-32B (57.6 on GPQA-Diamond and 68.7 on MMLU-Pro).
258
+
259
+
260
+ ## Code Reasoning
261
+
262
+ The evaluation code is modified from [Qwen2.5-Coder](https://github.com/QwenLM/Qwen2.5-Coder/tree/main/qwencoder-eval/reasoning/livecode_bench_cot). In our evaluation, we set the temperature to 0.6, the top-p to 0.95 and the max_tokens to 32768. We provide the example to reproduce our results in [code_evaluation](https://github.com/fanqiwan/FuseAI/tree/main/FuseO1-Preview/code_evaluation).
263
+
264
+ The system prompt for evaluation is set to:
265
+
266
+ ```sh
267
+ A conversation between User and Assistant. The user asks a question, and the Assistant solves it. The assistant first thinks about the reasoning process in the mind and then provides the user with the answer. The reasoning process and answer are enclosed within <think> </think> and <answer> </answer> tags, respectively, i.e., <think> reasoning process here </think> <answer> answer here </answer>.
268
+ ```
269
+
270
+ In our evaluation of LiveCodeBench, we follow the method from DeepSeek-R1 and make a slight modification. The Pass@1 is computed by averaging the results across 16 sampled responses per prompt.
271
+
272
+ The evaluation results are shown in the table below:
273
+
274
+ | Models | LiveCodeBench | LiveCodeBench-Easy | LiveCodeBench-Medium | LiveCodeBench-Hard |
275
+ |:------ | --------------| ------------------- | ------------ | -------------- |
276
+ | OpenAI o1 | 63.4 | 98.5 | 80.9 | 31.7 |
277
+ | OpenAI o1-preview | 42.7 | 97.0 | 47.2 | 9.8 |
278
+ | OpenAI o1-mini | 52.00 | 91.0 | 67.4 | 19.5 |
279
+ | DeepSeek R1 | 62.8 | 98.4 | 78.3 | 32.2 |
280
+ | [deepseek-ai/DeepSeek-R1-Distill-Qwen-32B](https://huggingface.co/deepseek-ai/DeepSeek-R1-Distill-Qwen-32B) | 56.1 | 93.6 | 73.1 | 23.4 |
281
+ | [Qwen/QwQ-32B-Preview](https://huggingface.co/Qwen/QwQ-32B-Preview) | 44.4 | 94.9 | 53.8 | 10.0 |
282
+ | [NovaSky-AI/Sky-T1-32B-Preview](https://huggingface.co/NovaSky-AI/Sky-T1-32B-Preview) | 37.3 | 89.7 | 40.4 | 6.6 |
283
+ | [FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-Qwen2.5-Coder-32B-Preview) | 56.4 | 92.9 | 73.5 | 24.2 |
284
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-32B-Preview) | 54.8| 93.9 | 71.7 | 21.3 |
285
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-Flash-32B-Preview) | 58.2 | 94.3 | 77.1 | 25.0 |
286
+ | [FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview](https://huggingface.co/FuseAI/FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview) | 57.9 | 93.6 | 76.0 | 25.5 |
287
+
288
+ We show that our merged FuseO1-DeepSeekR1-QwQ-SkyT1-32B-Preview demonstrate superior performance improvements comparet to DeepSeek-R1-Distill-Qwen-32B, QwQ-32B-Preview, and Sky-T1-32B-Preview on scientific reasoning. Specifically, our model achieves an accuracy of **57.9 on LiveCodeBench and 25.5 on LiveCodeBench-Hard**, demonstrating significant performance improvements compared to DeepSeek-R1-Distill-Qwen-32B (56.1 on LiveCodeBench and 23.4 on LiveCodeBench-Hard), OpenAI o1-preview (42.7 on LiveCodeBench and 9.8 on LiveCodeBench-Hard) and OpenAI o1-mini (52.0 on LiveCodeBench and 19.5 on LiveCodeBench-Hard Pass@1).
289
+
290
+ ## Future Works
291
+
292
+ This work is our first attempt effort to achieve knowledge fusion of System-II reasoning LLMs through a model merging approach, which is limited to LLMs with identical scale and architecture. In future work, we plan to employ our [explicit model fusion](https://arxiv.org/abs/2401.10491) method, based on multi-teacher knowledge distillation, and our [implici model fusion](https://arxiv.org/abs/2412.03187) method, which utilizes weighted-reward preference optimization for LLMs with different scales and architectures.
293
+ Furthermore, we intend to explore the combination of knowledge fusion with reinforcement learning (RL) methods, which have been demonstrated as the most effective approach for enhancing reasoning abilities. Stay tuned for the next version of FuseO1!
294
+
295
+ ## Citations
296
+
297
+ ```
298
+ @article{wan2024fusechat,
299
+ title={Fusechat: Knowledge fusion of chat models},
300
+ author={Wan, Fanqi and Zhong, Longguang and Yang, Ziyi and Chen, Ruijun and Quan, Xiaojun},
301
+ journal={arXiv preprint arXiv:2408.07990},
302
+ year={2024}
303
+ }
304
+ ```
config.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen2Model"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 151643,
7
+ "eos_token_id": 151643,
8
+ "hidden_act": "silu",
9
+ "hidden_size": 5120,
10
+ "initializer_range": 0.02,
11
+ "intermediate_size": 27648,
12
+ "max_position_embeddings": 131072,
13
+ "max_window_layers": 64,
14
+ "model_type": "qwen2",
15
+ "num_attention_heads": 40,
16
+ "num_hidden_layers": 64,
17
+ "num_key_value_heads": 8,
18
+ "quantization_config": {
19
+ "modules_to_not_convert": null,
20
+ "quant_method": "torchao",
21
+ "quant_type": {
22
+ "default": {
23
+ "_data": {
24
+ "group_size": 128,
25
+ "layout": {
26
+ "_data": {},
27
+ "_type": "Int4CPULayout",
28
+ "_version": 1
29
+ },
30
+ "set_inductor_config": true,
31
+ "use_hqq": false,
32
+ "zero_point_domain": {
33
+ "_data": "NONE",
34
+ "_type": "ZeroPointDomain"
35
+ }
36
+ },
37
+ "_type": "Int4WeightOnlyConfig",
38
+ "_version": 1
39
+ }
40
+ },
41
+ "quant_type_kwargs": {}
42
+ },
43
+ "rms_norm_eps": 1e-05,
44
+ "rope_scaling": null,
45
+ "rope_theta": 1000000.0,
46
+ "sliding_window": 131072,
47
+ "tie_word_embeddings": false,
48
+ "torch_dtype": "bfloat16",
49
+ "transformers_version": "4.51.3",
50
+ "use_cache": true,
51
+ "use_sliding_window": false,
52
+ "vocab_size": 152064
53
+ }
pytorch_model-00001-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:abfb40acd09f6fe383c3b265b7d72c836b0d6b9b25fb80f56c8fd0684b329d9b
3
+ size 4958534211
pytorch_model-00002-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7a6ec3db581974e433a160f09819c73a214577638f1aa8ac6cdfd218b42c4be7
3
+ size 4997600767
pytorch_model-00003-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9b5eae67558b37cd9b4acd3e41ca94999d2d51046ae90e33862649b2ac4af9a8
3
+ size 4997600791
pytorch_model-00004-of-00004.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50316138c2e1ebd1ed5419edea9a7990d65b2f07e75d8dca74699acbf0254c36
3
+ size 3184117585
pytorch_model.bin.index.json ADDED
@@ -0,0 +1,777 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 18137360384
4
+ },
5
+ "weight_map": {
6
+ "embed_tokens.weight": "pytorch_model-00001-of-00004.bin",
7
+ "layers.0.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
8
+ "layers.0.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
9
+ "layers.0.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
10
+ "layers.0.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
11
+ "layers.0.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
12
+ "layers.0.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
13
+ "layers.0.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
14
+ "layers.0.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
15
+ "layers.0.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
16
+ "layers.0.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
17
+ "layers.0.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
18
+ "layers.0.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
19
+ "layers.1.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
20
+ "layers.1.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
21
+ "layers.1.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
22
+ "layers.1.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
23
+ "layers.1.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
24
+ "layers.1.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
25
+ "layers.1.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
26
+ "layers.1.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
27
+ "layers.1.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
28
+ "layers.1.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
29
+ "layers.1.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
30
+ "layers.1.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
31
+ "layers.10.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
32
+ "layers.10.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
33
+ "layers.10.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
34
+ "layers.10.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
35
+ "layers.10.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
36
+ "layers.10.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
37
+ "layers.10.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
38
+ "layers.10.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
39
+ "layers.10.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
40
+ "layers.10.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
41
+ "layers.10.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
42
+ "layers.10.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
43
+ "layers.11.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
44
+ "layers.11.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
45
+ "layers.11.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
46
+ "layers.11.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
47
+ "layers.11.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
48
+ "layers.11.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
49
+ "layers.11.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
50
+ "layers.11.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
51
+ "layers.11.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
52
+ "layers.11.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
53
+ "layers.11.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
54
+ "layers.11.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
55
+ "layers.12.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
56
+ "layers.12.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
57
+ "layers.12.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
58
+ "layers.12.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
59
+ "layers.12.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
60
+ "layers.12.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
61
+ "layers.12.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
62
+ "layers.12.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
63
+ "layers.12.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
64
+ "layers.12.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
65
+ "layers.12.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
66
+ "layers.12.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
67
+ "layers.13.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
68
+ "layers.13.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
69
+ "layers.13.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
70
+ "layers.13.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
71
+ "layers.13.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
72
+ "layers.13.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
73
+ "layers.13.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
74
+ "layers.13.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
75
+ "layers.13.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
76
+ "layers.13.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
77
+ "layers.13.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
78
+ "layers.13.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
79
+ "layers.14.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
80
+ "layers.14.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
81
+ "layers.14.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
82
+ "layers.14.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
83
+ "layers.14.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
84
+ "layers.14.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
85
+ "layers.14.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
86
+ "layers.14.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
87
+ "layers.14.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
88
+ "layers.14.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
89
+ "layers.14.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
90
+ "layers.14.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
91
+ "layers.15.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
92
+ "layers.15.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
93
+ "layers.15.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
94
+ "layers.15.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
95
+ "layers.15.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
96
+ "layers.15.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
97
+ "layers.15.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
98
+ "layers.15.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
99
+ "layers.15.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
100
+ "layers.15.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
101
+ "layers.15.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
102
+ "layers.15.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
103
+ "layers.16.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
104
+ "layers.16.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
105
+ "layers.16.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
106
+ "layers.16.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
107
+ "layers.16.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
108
+ "layers.16.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
109
+ "layers.16.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
110
+ "layers.16.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
111
+ "layers.16.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
112
+ "layers.16.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
113
+ "layers.16.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
114
+ "layers.16.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
115
+ "layers.17.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
116
+ "layers.17.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
117
+ "layers.17.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
118
+ "layers.17.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
119
+ "layers.17.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
120
+ "layers.17.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
121
+ "layers.17.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
122
+ "layers.17.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
123
+ "layers.17.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
124
+ "layers.17.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
125
+ "layers.17.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
126
+ "layers.17.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
127
+ "layers.18.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
128
+ "layers.18.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
129
+ "layers.18.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
130
+ "layers.18.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
131
+ "layers.18.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
132
+ "layers.18.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
133
+ "layers.18.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
134
+ "layers.18.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
135
+ "layers.18.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
136
+ "layers.18.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
137
+ "layers.18.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
138
+ "layers.18.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
139
+ "layers.19.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
140
+ "layers.19.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
141
+ "layers.19.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
142
+ "layers.19.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
143
+ "layers.19.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
144
+ "layers.19.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
145
+ "layers.19.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
146
+ "layers.19.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
147
+ "layers.19.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
148
+ "layers.19.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
149
+ "layers.19.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
150
+ "layers.19.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
151
+ "layers.2.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
152
+ "layers.2.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
153
+ "layers.2.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
154
+ "layers.2.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
155
+ "layers.2.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
156
+ "layers.2.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
157
+ "layers.2.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
158
+ "layers.2.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
159
+ "layers.2.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
160
+ "layers.2.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
161
+ "layers.2.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
162
+ "layers.2.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
163
+ "layers.20.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
164
+ "layers.20.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
165
+ "layers.20.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
166
+ "layers.20.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
167
+ "layers.20.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
168
+ "layers.20.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
169
+ "layers.20.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
170
+ "layers.20.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
171
+ "layers.20.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
172
+ "layers.20.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
173
+ "layers.20.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
174
+ "layers.20.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
175
+ "layers.21.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
176
+ "layers.21.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
177
+ "layers.21.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
178
+ "layers.21.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
179
+ "layers.21.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
180
+ "layers.21.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
181
+ "layers.21.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
182
+ "layers.21.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
183
+ "layers.21.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
184
+ "layers.21.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
185
+ "layers.21.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
186
+ "layers.21.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
187
+ "layers.22.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
188
+ "layers.22.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
189
+ "layers.22.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
190
+ "layers.22.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
191
+ "layers.22.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
192
+ "layers.22.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
193
+ "layers.22.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
194
+ "layers.22.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
195
+ "layers.22.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
196
+ "layers.22.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
197
+ "layers.22.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
198
+ "layers.22.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
199
+ "layers.23.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
200
+ "layers.23.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
201
+ "layers.23.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
202
+ "layers.23.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
203
+ "layers.23.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
204
+ "layers.23.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
205
+ "layers.23.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
206
+ "layers.23.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
207
+ "layers.23.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
208
+ "layers.23.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
209
+ "layers.23.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
210
+ "layers.23.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
211
+ "layers.24.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
212
+ "layers.24.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
213
+ "layers.24.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
214
+ "layers.24.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
215
+ "layers.24.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
216
+ "layers.24.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
217
+ "layers.24.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
218
+ "layers.24.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
219
+ "layers.24.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
220
+ "layers.24.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
221
+ "layers.24.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
222
+ "layers.24.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
223
+ "layers.25.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
224
+ "layers.25.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
225
+ "layers.25.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
226
+ "layers.25.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
227
+ "layers.25.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
228
+ "layers.25.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
229
+ "layers.25.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
230
+ "layers.25.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
231
+ "layers.25.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
232
+ "layers.25.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
233
+ "layers.25.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
234
+ "layers.25.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
235
+ "layers.26.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
236
+ "layers.26.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
237
+ "layers.26.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
238
+ "layers.26.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
239
+ "layers.26.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
240
+ "layers.26.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
241
+ "layers.26.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
242
+ "layers.26.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
243
+ "layers.26.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
244
+ "layers.26.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
245
+ "layers.26.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
246
+ "layers.26.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
247
+ "layers.27.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
248
+ "layers.27.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
249
+ "layers.27.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
250
+ "layers.27.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
251
+ "layers.27.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
252
+ "layers.27.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
253
+ "layers.27.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
254
+ "layers.27.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
255
+ "layers.27.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
256
+ "layers.27.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
257
+ "layers.27.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
258
+ "layers.27.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
259
+ "layers.28.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
260
+ "layers.28.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
261
+ "layers.28.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
262
+ "layers.28.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
263
+ "layers.28.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
264
+ "layers.28.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
265
+ "layers.28.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
266
+ "layers.28.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
267
+ "layers.28.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
268
+ "layers.28.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
269
+ "layers.28.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
270
+ "layers.28.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
271
+ "layers.29.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
272
+ "layers.29.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
273
+ "layers.29.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
274
+ "layers.29.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
275
+ "layers.29.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
276
+ "layers.29.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
277
+ "layers.29.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
278
+ "layers.29.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
279
+ "layers.29.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
280
+ "layers.29.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
281
+ "layers.29.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
282
+ "layers.29.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
283
+ "layers.3.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
284
+ "layers.3.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
285
+ "layers.3.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
286
+ "layers.3.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
287
+ "layers.3.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
288
+ "layers.3.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
289
+ "layers.3.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
290
+ "layers.3.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
291
+ "layers.3.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
292
+ "layers.3.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
293
+ "layers.3.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
294
+ "layers.3.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
295
+ "layers.30.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
296
+ "layers.30.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
297
+ "layers.30.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
298
+ "layers.30.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
299
+ "layers.30.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
300
+ "layers.30.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
301
+ "layers.30.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
302
+ "layers.30.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
303
+ "layers.30.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
304
+ "layers.30.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
305
+ "layers.30.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
306
+ "layers.30.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
307
+ "layers.31.input_layernorm.weight": "pytorch_model-00002-of-00004.bin",
308
+ "layers.31.mlp.down_proj.weight": "pytorch_model-00002-of-00004.bin",
309
+ "layers.31.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
310
+ "layers.31.mlp.up_proj.weight": "pytorch_model-00002-of-00004.bin",
311
+ "layers.31.post_attention_layernorm.weight": "pytorch_model-00002-of-00004.bin",
312
+ "layers.31.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
313
+ "layers.31.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
314
+ "layers.31.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
315
+ "layers.31.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
316
+ "layers.31.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
317
+ "layers.31.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
318
+ "layers.31.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
319
+ "layers.32.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
320
+ "layers.32.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
321
+ "layers.32.mlp.gate_proj.weight": "pytorch_model-00002-of-00004.bin",
322
+ "layers.32.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
323
+ "layers.32.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
324
+ "layers.32.self_attn.k_proj.bias": "pytorch_model-00002-of-00004.bin",
325
+ "layers.32.self_attn.k_proj.weight": "pytorch_model-00002-of-00004.bin",
326
+ "layers.32.self_attn.o_proj.weight": "pytorch_model-00002-of-00004.bin",
327
+ "layers.32.self_attn.q_proj.bias": "pytorch_model-00002-of-00004.bin",
328
+ "layers.32.self_attn.q_proj.weight": "pytorch_model-00002-of-00004.bin",
329
+ "layers.32.self_attn.v_proj.bias": "pytorch_model-00002-of-00004.bin",
330
+ "layers.32.self_attn.v_proj.weight": "pytorch_model-00002-of-00004.bin",
331
+ "layers.33.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
332
+ "layers.33.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
333
+ "layers.33.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
334
+ "layers.33.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
335
+ "layers.33.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
336
+ "layers.33.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
337
+ "layers.33.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
338
+ "layers.33.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
339
+ "layers.33.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
340
+ "layers.33.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
341
+ "layers.33.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
342
+ "layers.33.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
343
+ "layers.34.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
344
+ "layers.34.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
345
+ "layers.34.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
346
+ "layers.34.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
347
+ "layers.34.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
348
+ "layers.34.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
349
+ "layers.34.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
350
+ "layers.34.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
351
+ "layers.34.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
352
+ "layers.34.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
353
+ "layers.34.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
354
+ "layers.34.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
355
+ "layers.35.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
356
+ "layers.35.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
357
+ "layers.35.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
358
+ "layers.35.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
359
+ "layers.35.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
360
+ "layers.35.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
361
+ "layers.35.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
362
+ "layers.35.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
363
+ "layers.35.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
364
+ "layers.35.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
365
+ "layers.35.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
366
+ "layers.35.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
367
+ "layers.36.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
368
+ "layers.36.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
369
+ "layers.36.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
370
+ "layers.36.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
371
+ "layers.36.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
372
+ "layers.36.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
373
+ "layers.36.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
374
+ "layers.36.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
375
+ "layers.36.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
376
+ "layers.36.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
377
+ "layers.36.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
378
+ "layers.36.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
379
+ "layers.37.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
380
+ "layers.37.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
381
+ "layers.37.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
382
+ "layers.37.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
383
+ "layers.37.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
384
+ "layers.37.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
385
+ "layers.37.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
386
+ "layers.37.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
387
+ "layers.37.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
388
+ "layers.37.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
389
+ "layers.37.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
390
+ "layers.37.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
391
+ "layers.38.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
392
+ "layers.38.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
393
+ "layers.38.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
394
+ "layers.38.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
395
+ "layers.38.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
396
+ "layers.38.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
397
+ "layers.38.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
398
+ "layers.38.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
399
+ "layers.38.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
400
+ "layers.38.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
401
+ "layers.38.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
402
+ "layers.38.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
403
+ "layers.39.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
404
+ "layers.39.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
405
+ "layers.39.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
406
+ "layers.39.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
407
+ "layers.39.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
408
+ "layers.39.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
409
+ "layers.39.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
410
+ "layers.39.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
411
+ "layers.39.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
412
+ "layers.39.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
413
+ "layers.39.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
414
+ "layers.39.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
415
+ "layers.4.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
416
+ "layers.4.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
417
+ "layers.4.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
418
+ "layers.4.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
419
+ "layers.4.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
420
+ "layers.4.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
421
+ "layers.4.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
422
+ "layers.4.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
423
+ "layers.4.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
424
+ "layers.4.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
425
+ "layers.4.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
426
+ "layers.4.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
427
+ "layers.40.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
428
+ "layers.40.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
429
+ "layers.40.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
430
+ "layers.40.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
431
+ "layers.40.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
432
+ "layers.40.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
433
+ "layers.40.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
434
+ "layers.40.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
435
+ "layers.40.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
436
+ "layers.40.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
437
+ "layers.40.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
438
+ "layers.40.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
439
+ "layers.41.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
440
+ "layers.41.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
441
+ "layers.41.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
442
+ "layers.41.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
443
+ "layers.41.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
444
+ "layers.41.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
445
+ "layers.41.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
446
+ "layers.41.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
447
+ "layers.41.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
448
+ "layers.41.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
449
+ "layers.41.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
450
+ "layers.41.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
451
+ "layers.42.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
452
+ "layers.42.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
453
+ "layers.42.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
454
+ "layers.42.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
455
+ "layers.42.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
456
+ "layers.42.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
457
+ "layers.42.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
458
+ "layers.42.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
459
+ "layers.42.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
460
+ "layers.42.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
461
+ "layers.42.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
462
+ "layers.42.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
463
+ "layers.43.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
464
+ "layers.43.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
465
+ "layers.43.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
466
+ "layers.43.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
467
+ "layers.43.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
468
+ "layers.43.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
469
+ "layers.43.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
470
+ "layers.43.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
471
+ "layers.43.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
472
+ "layers.43.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
473
+ "layers.43.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
474
+ "layers.43.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
475
+ "layers.44.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
476
+ "layers.44.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
477
+ "layers.44.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
478
+ "layers.44.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
479
+ "layers.44.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
480
+ "layers.44.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
481
+ "layers.44.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
482
+ "layers.44.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
483
+ "layers.44.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
484
+ "layers.44.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
485
+ "layers.44.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
486
+ "layers.44.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
487
+ "layers.45.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
488
+ "layers.45.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
489
+ "layers.45.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
490
+ "layers.45.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
491
+ "layers.45.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
492
+ "layers.45.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
493
+ "layers.45.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
494
+ "layers.45.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
495
+ "layers.45.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
496
+ "layers.45.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
497
+ "layers.45.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
498
+ "layers.45.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
499
+ "layers.46.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
500
+ "layers.46.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
501
+ "layers.46.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
502
+ "layers.46.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
503
+ "layers.46.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
504
+ "layers.46.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
505
+ "layers.46.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
506
+ "layers.46.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
507
+ "layers.46.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
508
+ "layers.46.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
509
+ "layers.46.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
510
+ "layers.46.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
511
+ "layers.47.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
512
+ "layers.47.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
513
+ "layers.47.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
514
+ "layers.47.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
515
+ "layers.47.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
516
+ "layers.47.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
517
+ "layers.47.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
518
+ "layers.47.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
519
+ "layers.47.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
520
+ "layers.47.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
521
+ "layers.47.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
522
+ "layers.47.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
523
+ "layers.48.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
524
+ "layers.48.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
525
+ "layers.48.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
526
+ "layers.48.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
527
+ "layers.48.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
528
+ "layers.48.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
529
+ "layers.48.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
530
+ "layers.48.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
531
+ "layers.48.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
532
+ "layers.48.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
533
+ "layers.48.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
534
+ "layers.48.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
535
+ "layers.49.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
536
+ "layers.49.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
537
+ "layers.49.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
538
+ "layers.49.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
539
+ "layers.49.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
540
+ "layers.49.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
541
+ "layers.49.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
542
+ "layers.49.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
543
+ "layers.49.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
544
+ "layers.49.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
545
+ "layers.49.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
546
+ "layers.49.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
547
+ "layers.5.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
548
+ "layers.5.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
549
+ "layers.5.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
550
+ "layers.5.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
551
+ "layers.5.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
552
+ "layers.5.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
553
+ "layers.5.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
554
+ "layers.5.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
555
+ "layers.5.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
556
+ "layers.5.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
557
+ "layers.5.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
558
+ "layers.5.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
559
+ "layers.50.input_layernorm.weight": "pytorch_model-00003-of-00004.bin",
560
+ "layers.50.mlp.down_proj.weight": "pytorch_model-00003-of-00004.bin",
561
+ "layers.50.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
562
+ "layers.50.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
563
+ "layers.50.post_attention_layernorm.weight": "pytorch_model-00003-of-00004.bin",
564
+ "layers.50.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
565
+ "layers.50.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
566
+ "layers.50.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
567
+ "layers.50.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
568
+ "layers.50.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
569
+ "layers.50.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
570
+ "layers.50.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
571
+ "layers.51.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
572
+ "layers.51.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
573
+ "layers.51.mlp.gate_proj.weight": "pytorch_model-00003-of-00004.bin",
574
+ "layers.51.mlp.up_proj.weight": "pytorch_model-00003-of-00004.bin",
575
+ "layers.51.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
576
+ "layers.51.self_attn.k_proj.bias": "pytorch_model-00003-of-00004.bin",
577
+ "layers.51.self_attn.k_proj.weight": "pytorch_model-00003-of-00004.bin",
578
+ "layers.51.self_attn.o_proj.weight": "pytorch_model-00003-of-00004.bin",
579
+ "layers.51.self_attn.q_proj.bias": "pytorch_model-00003-of-00004.bin",
580
+ "layers.51.self_attn.q_proj.weight": "pytorch_model-00003-of-00004.bin",
581
+ "layers.51.self_attn.v_proj.bias": "pytorch_model-00003-of-00004.bin",
582
+ "layers.51.self_attn.v_proj.weight": "pytorch_model-00003-of-00004.bin",
583
+ "layers.52.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
584
+ "layers.52.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
585
+ "layers.52.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
586
+ "layers.52.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
587
+ "layers.52.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
588
+ "layers.52.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
589
+ "layers.52.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
590
+ "layers.52.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
591
+ "layers.52.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
592
+ "layers.52.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
593
+ "layers.52.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
594
+ "layers.52.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
595
+ "layers.53.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
596
+ "layers.53.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
597
+ "layers.53.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
598
+ "layers.53.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
599
+ "layers.53.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
600
+ "layers.53.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
601
+ "layers.53.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
602
+ "layers.53.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
603
+ "layers.53.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
604
+ "layers.53.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
605
+ "layers.53.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
606
+ "layers.53.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
607
+ "layers.54.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
608
+ "layers.54.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
609
+ "layers.54.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
610
+ "layers.54.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
611
+ "layers.54.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
612
+ "layers.54.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
613
+ "layers.54.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
614
+ "layers.54.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
615
+ "layers.54.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
616
+ "layers.54.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
617
+ "layers.54.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
618
+ "layers.54.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
619
+ "layers.55.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
620
+ "layers.55.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
621
+ "layers.55.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
622
+ "layers.55.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
623
+ "layers.55.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
624
+ "layers.55.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
625
+ "layers.55.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
626
+ "layers.55.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
627
+ "layers.55.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
628
+ "layers.55.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
629
+ "layers.55.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
630
+ "layers.55.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
631
+ "layers.56.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
632
+ "layers.56.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
633
+ "layers.56.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
634
+ "layers.56.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
635
+ "layers.56.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
636
+ "layers.56.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
637
+ "layers.56.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
638
+ "layers.56.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
639
+ "layers.56.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
640
+ "layers.56.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
641
+ "layers.56.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
642
+ "layers.56.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
643
+ "layers.57.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
644
+ "layers.57.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
645
+ "layers.57.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
646
+ "layers.57.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
647
+ "layers.57.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
648
+ "layers.57.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
649
+ "layers.57.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
650
+ "layers.57.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
651
+ "layers.57.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
652
+ "layers.57.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
653
+ "layers.57.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
654
+ "layers.57.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
655
+ "layers.58.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
656
+ "layers.58.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
657
+ "layers.58.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
658
+ "layers.58.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
659
+ "layers.58.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
660
+ "layers.58.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
661
+ "layers.58.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
662
+ "layers.58.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
663
+ "layers.58.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
664
+ "layers.58.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
665
+ "layers.58.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
666
+ "layers.58.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
667
+ "layers.59.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
668
+ "layers.59.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
669
+ "layers.59.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
670
+ "layers.59.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
671
+ "layers.59.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
672
+ "layers.59.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
673
+ "layers.59.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
674
+ "layers.59.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
675
+ "layers.59.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
676
+ "layers.59.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
677
+ "layers.59.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
678
+ "layers.59.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
679
+ "layers.6.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
680
+ "layers.6.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
681
+ "layers.6.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
682
+ "layers.6.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
683
+ "layers.6.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
684
+ "layers.6.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
685
+ "layers.6.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
686
+ "layers.6.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
687
+ "layers.6.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
688
+ "layers.6.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
689
+ "layers.6.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
690
+ "layers.6.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
691
+ "layers.60.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
692
+ "layers.60.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
693
+ "layers.60.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
694
+ "layers.60.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
695
+ "layers.60.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
696
+ "layers.60.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
697
+ "layers.60.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
698
+ "layers.60.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
699
+ "layers.60.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
700
+ "layers.60.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
701
+ "layers.60.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
702
+ "layers.60.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
703
+ "layers.61.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
704
+ "layers.61.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
705
+ "layers.61.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
706
+ "layers.61.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
707
+ "layers.61.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
708
+ "layers.61.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
709
+ "layers.61.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
710
+ "layers.61.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
711
+ "layers.61.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
712
+ "layers.61.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
713
+ "layers.61.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
714
+ "layers.61.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
715
+ "layers.62.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
716
+ "layers.62.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
717
+ "layers.62.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
718
+ "layers.62.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
719
+ "layers.62.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
720
+ "layers.62.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
721
+ "layers.62.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
722
+ "layers.62.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
723
+ "layers.62.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
724
+ "layers.62.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
725
+ "layers.62.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
726
+ "layers.62.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
727
+ "layers.63.input_layernorm.weight": "pytorch_model-00004-of-00004.bin",
728
+ "layers.63.mlp.down_proj.weight": "pytorch_model-00004-of-00004.bin",
729
+ "layers.63.mlp.gate_proj.weight": "pytorch_model-00004-of-00004.bin",
730
+ "layers.63.mlp.up_proj.weight": "pytorch_model-00004-of-00004.bin",
731
+ "layers.63.post_attention_layernorm.weight": "pytorch_model-00004-of-00004.bin",
732
+ "layers.63.self_attn.k_proj.bias": "pytorch_model-00004-of-00004.bin",
733
+ "layers.63.self_attn.k_proj.weight": "pytorch_model-00004-of-00004.bin",
734
+ "layers.63.self_attn.o_proj.weight": "pytorch_model-00004-of-00004.bin",
735
+ "layers.63.self_attn.q_proj.bias": "pytorch_model-00004-of-00004.bin",
736
+ "layers.63.self_attn.q_proj.weight": "pytorch_model-00004-of-00004.bin",
737
+ "layers.63.self_attn.v_proj.bias": "pytorch_model-00004-of-00004.bin",
738
+ "layers.63.self_attn.v_proj.weight": "pytorch_model-00004-of-00004.bin",
739
+ "layers.7.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
740
+ "layers.7.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
741
+ "layers.7.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
742
+ "layers.7.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
743
+ "layers.7.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
744
+ "layers.7.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
745
+ "layers.7.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
746
+ "layers.7.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
747
+ "layers.7.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
748
+ "layers.7.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
749
+ "layers.7.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
750
+ "layers.7.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
751
+ "layers.8.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
752
+ "layers.8.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
753
+ "layers.8.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
754
+ "layers.8.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
755
+ "layers.8.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
756
+ "layers.8.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
757
+ "layers.8.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
758
+ "layers.8.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
759
+ "layers.8.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
760
+ "layers.8.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
761
+ "layers.8.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
762
+ "layers.8.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
763
+ "layers.9.input_layernorm.weight": "pytorch_model-00001-of-00004.bin",
764
+ "layers.9.mlp.down_proj.weight": "pytorch_model-00001-of-00004.bin",
765
+ "layers.9.mlp.gate_proj.weight": "pytorch_model-00001-of-00004.bin",
766
+ "layers.9.mlp.up_proj.weight": "pytorch_model-00001-of-00004.bin",
767
+ "layers.9.post_attention_layernorm.weight": "pytorch_model-00001-of-00004.bin",
768
+ "layers.9.self_attn.k_proj.bias": "pytorch_model-00001-of-00004.bin",
769
+ "layers.9.self_attn.k_proj.weight": "pytorch_model-00001-of-00004.bin",
770
+ "layers.9.self_attn.o_proj.weight": "pytorch_model-00001-of-00004.bin",
771
+ "layers.9.self_attn.q_proj.bias": "pytorch_model-00001-of-00004.bin",
772
+ "layers.9.self_attn.q_proj.weight": "pytorch_model-00001-of-00004.bin",
773
+ "layers.9.self_attn.v_proj.bias": "pytorch_model-00001-of-00004.bin",
774
+ "layers.9.self_attn.v_proj.weight": "pytorch_model-00001-of-00004.bin",
775
+ "norm.weight": "pytorch_model-00004-of-00004.bin"
776
+ }
777
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<｜begin▁of▁sentence｜>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<｜end▁of▁sentence｜>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<｜end▁of▁sentence｜>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ }
23
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:df4e7ca41f3f7f64a5b6945b3bf69d8b620334fdde07a1e8932f522775798602
3
+ size 11422185
tokenizer_config.json ADDED
@@ -0,0 +1,195 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": null,
5
+ "added_tokens_decoder": {
6
+ "151643": {
7
+ "content": "<｜end▁of▁sentence｜>",
8
+ "lstrip": false,
9
+ "normalized": false,
10
+ "rstrip": false,
11
+ "single_word": false,
12
+ "special": true
13
+ },
14
+ "151644": {
15
+ "content": "<｜User｜>",
16
+ "lstrip": false,
17
+ "normalized": false,
18
+ "rstrip": false,
19
+ "single_word": false,
20
+ "special": false
21
+ },
22
+ "151645": {
23
+ "content": "<｜Assistant｜>",
24
+ "lstrip": false,
25
+ "normalized": false,
26
+ "rstrip": false,
27
+ "single_word": false,
28
+ "special": false
29
+ },
30
+ "151646": {
31
+ "content": "<｜begin▁of▁sentence｜>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ },
38
+ "151647": {
39
+ "content": "<|EOT|>",
40
+ "lstrip": false,
41
+ "normalized": false,
42
+ "rstrip": false,
43
+ "single_word": false,
44
+ "special": false
45
+ },
46
+ "151648": {
47
+ "content": "<think>",
48
+ "lstrip": false,
49
+ "normalized": false,
50
+ "rstrip": false,
51
+ "single_word": false,
52
+ "special": false
53
+ },
54
+ "151649": {
55
+ "content": "</think>",
56
+ "lstrip": false,
57
+ "normalized": false,
58
+ "rstrip": false,
59
+ "single_word": false,
60
+ "special": false
61
+ },
62
+ "151650": {
63
+ "content": "<|quad_start|>",
64
+ "lstrip": false,
65
+ "normalized": false,
66
+ "rstrip": false,
67
+ "single_word": false,
68
+ "special": true
69
+ },
70
+ "151651": {
71
+ "content": "<|quad_end|>",
72
+ "lstrip": false,
73
+ "normalized": false,
74
+ "rstrip": false,
75
+ "single_word": false,
76
+ "special": true
77
+ },
78
+ "151652": {
79
+ "content": "<|vision_start|>",
80
+ "lstrip": false,
81
+ "normalized": false,
82
+ "rstrip": false,
83
+ "single_word": false,
84
+ "special": true
85
+ },
86
+ "151653": {
87
+ "content": "<|vision_end|>",
88
+ "lstrip": false,
89
+ "normalized": false,
90
+ "rstrip": false,
91
+ "single_word": false,
92
+ "special": true
93
+ },
94
+ "151654": {
95
+ "content": "<|vision_pad|>",
96
+ "lstrip": false,
97
+ "normalized": false,
98
+ "rstrip": false,
99
+ "single_word": false,
100
+ "special": true
101
+ },
102
+ "151655": {
103
+ "content": "<|image_pad|>",
104
+ "lstrip": false,
105
+ "normalized": false,
106
+ "rstrip": false,
107
+ "single_word": false,
108
+ "special": true
109
+ },
110
+ "151656": {
111
+ "content": "<|video_pad|>",
112
+ "lstrip": false,
113
+ "normalized": false,
114
+ "rstrip": false,
115
+ "single_word": false,
116
+ "special": true
117
+ },
118
+ "151657": {
119
+ "content": "<tool_call>",
120
+ "lstrip": false,
121
+ "normalized": false,
122
+ "rstrip": false,
123
+ "single_word": false,
124
+ "special": false
125
+ },
126
+ "151658": {
127
+ "content": "</tool_call>",
128
+ "lstrip": false,
129
+ "normalized": false,
130
+ "rstrip": false,
131
+ "single_word": false,
132
+ "special": false
133
+ },
134
+ "151659": {
135
+ "content": "<|fim_prefix|>",
136
+ "lstrip": false,
137
+ "normalized": false,
138
+ "rstrip": false,
139
+ "single_word": false,
140
+ "special": false
141
+ },
142
+ "151660": {
143
+ "content": "<|fim_middle|>",
144
+ "lstrip": false,
145
+ "normalized": false,
146
+ "rstrip": false,
147
+ "single_word": false,
148
+ "special": false
149
+ },
150
+ "151661": {
151
+ "content": "<|fim_suffix|>",
152
+ "lstrip": false,
153
+ "normalized": false,
154
+ "rstrip": false,
155
+ "single_word": false,
156
+ "special": false
157
+ },
158
+ "151662": {
159
+ "content": "<|fim_pad|>",
160
+ "lstrip": false,
161
+ "normalized": false,
162
+ "rstrip": false,
163
+ "single_word": false,
164
+ "special": false
165
+ },
166
+ "151663": {
167
+ "content": "<|repo_name|>",
168
+ "lstrip": false,
169
+ "normalized": false,
170
+ "rstrip": false,
171
+ "single_word": false,
172
+ "special": false
173
+ },
174
+ "151664": {
175
+ "content": "<|file_sep|>",
176
+ "lstrip": false,
177
+ "normalized": false,
178
+ "rstrip": false,
179
+ "single_word": false,
180
+ "special": false
181
+ }
182
+ },
183
+ "bos_token": "<｜begin▁of▁sentence｜>",
184
+ "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set ns = namespace(is_first=false, is_tool=false, is_output_first=true, system_prompt='') %}{%- for message in messages %}{%- if message['role'] == 'system' %}{% set ns.system_prompt = message['content'] %}{%- endif %}{%- endfor %}{{bos_token}}{{ns.system_prompt}}{%- for message in messages %}{%- if message['role'] == 'user' %}{%- set ns.is_tool = false -%}{{'<｜User｜>' + message['content']}}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is none %}{%- set ns.is_tool = false -%}{%- for tool in message['tool_calls']%}{%- if not ns.is_first %}{{'<｜Assistant｜><｜tool▁calls▁begin｜><｜tool▁call▁begin���>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{%- set ns.is_first = true -%}{%- else %}{{'\\n' + '<｜tool▁call▁begin｜>' + tool['type'] + '<｜tool▁sep｜>' + tool['function']['name'] + '\\n' + '```json' + '\\n' + tool['function']['arguments'] + '\\n' + '```' + '<｜tool▁call▁end｜>'}}{{'<｜tool▁calls▁end｜><｜end▁of▁sentence｜>'}}{%- endif %}{%- endfor %}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is not none %}{%- if ns.is_tool %}{{'<｜tool▁outputs▁end｜>' + message['content'] + '<｜end▁of▁sentence｜>'}}{%- set ns.is_tool = false -%}{%- else %}{% set content = message['content'] %}{% if '</think>' in content %}{% set content = content.split('</think>')[-1] %}{% endif %}{{'<｜Assistant｜>' + content + '<｜end▁of▁sentence｜>'}}{%- endif %}{%- endif %}{%- if message['role'] == 'tool' %}{%- set ns.is_tool = true -%}{%- if ns.is_output_first %}{{'<｜tool▁outputs▁begin｜><｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- set ns.is_output_first = false %}{%- else %}{{'\\n<｜tool▁output▁begin｜>' + message['content'] + '<｜tool▁output▁end｜>'}}{%- endif %}{%- endif %}{%- endfor -%}{% if ns.is_tool %}{{'<｜tool▁outputs▁end｜>'}}{% endif %}{% if add_generation_prompt and not ns.is_tool %}{{'<｜Assistant｜>'}}{% endif %}",
185
+ "clean_up_tokenization_spaces": false,
186
+ "eos_token": "<｜end▁of▁sentence｜>",
187
+ "extra_special_tokens": {},
188
+ "legacy": true,
189
+ "model_max_length": 16384,
190
+ "pad_token": "<｜end▁of▁sentence｜>",
191
+ "sp_model_kwargs": {},
192
+ "tokenizer_class": "LlamaTokenizerFast",
193
+ "unk_token": null,
194
+ "use_default_system_prompt": false
195
+ }