ADOHAHA123 commited on
Commit
87040f9
·
verified ·
1 Parent(s): 5ecc41f

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -1,35 +1,12 @@
1
- *.7z filter=lfs diff=lfs merge=lfs -text
2
- *.arrow filter=lfs diff=lfs merge=lfs -text
3
  *.bin filter=lfs diff=lfs merge=lfs -text
4
- *.bz2 filter=lfs diff=lfs merge=lfs -text
5
- *.ckpt filter=lfs diff=lfs merge=lfs -text
6
- *.ftz filter=lfs diff=lfs merge=lfs -text
7
- *.gz filter=lfs diff=lfs merge=lfs -text
8
  *.h5 filter=lfs diff=lfs merge=lfs -text
9
- *.joblib filter=lfs diff=lfs merge=lfs -text
10
- *.lfs.* filter=lfs diff=lfs merge=lfs -text
11
- *.mlmodel filter=lfs diff=lfs merge=lfs -text
12
- *.model filter=lfs diff=lfs merge=lfs -text
13
  *.msgpack filter=lfs diff=lfs merge=lfs -text
14
- *.npy filter=lfs diff=lfs merge=lfs -text
15
- *.npz filter=lfs diff=lfs merge=lfs -text
16
- *.onnx filter=lfs diff=lfs merge=lfs -text
17
  *.ot filter=lfs diff=lfs merge=lfs -text
18
- *.parquet filter=lfs diff=lfs merge=lfs -text
19
- *.pb filter=lfs diff=lfs merge=lfs -text
20
- *.pickle filter=lfs diff=lfs merge=lfs -text
21
  *.pkl filter=lfs diff=lfs merge=lfs -text
 
22
  *.pt filter=lfs diff=lfs merge=lfs -text
23
  *.pth filter=lfs diff=lfs merge=lfs -text
24
- *.rar filter=lfs diff=lfs merge=lfs -text
25
- *.safetensors filter=lfs diff=lfs merge=lfs -text
26
- saved_model/**/* filter=lfs diff=lfs merge=lfs -text
27
- *.tar.* filter=lfs diff=lfs merge=lfs -text
28
- *.tar filter=lfs diff=lfs merge=lfs -text
29
- *.tflite filter=lfs diff=lfs merge=lfs -text
30
- *.tgz filter=lfs diff=lfs merge=lfs -text
31
- *.wasm filter=lfs diff=lfs merge=lfs -text
32
- *.xz 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
 
1
+ *.safetensors filter=lfs diff=lfs merge=lfs -text
 
2
  *.bin filter=lfs diff=lfs merge=lfs -text
 
 
 
 
3
  *.h5 filter=lfs diff=lfs merge=lfs -text
 
 
 
 
4
  *.msgpack filter=lfs diff=lfs merge=lfs -text
 
 
 
5
  *.ot filter=lfs diff=lfs merge=lfs -text
 
 
 
6
  *.pkl filter=lfs diff=lfs merge=lfs -text
7
+ *.pb filter=lfs diff=lfs merge=lfs -text
8
  *.pt filter=lfs diff=lfs merge=lfs -text
9
  *.pth filter=lfs diff=lfs merge=lfs -text
10
+ *.ckpt filter=lfs diff=lfs merge=lfs -text
11
+ figure2github.png filter=lfs diff=lfs merge=lfs -text
12
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
 
 
 
 
 
 
 
 
 
README.md ADDED
@@ -0,0 +1,286 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - zh
4
+ - en
5
+ license: apache-2.0
6
+ library_name: transformers
7
+ pipeline_tag: text-generation
8
+ tags:
9
+ - roleplay
10
+ - dialogue
11
+ - multi-turn
12
+ - qwen
13
+ - reinforcement-learning
14
+ - chat
15
+ base_model: Qwen/Qwen3-32B
16
+ ---
17
+
18
+ # HER-Qwen-32B
19
+
20
+ <p align="center">
21
+ <a href="https://arxiv.org/abs/xxxx.xxxxx"><img src="https://img.shields.io/badge/Paper-arXiv-red?logo=arxiv" alt="Paper"></a>
22
+ <a href="https://huggingface.co/datasets/ADOHAHA123/HER-Dataset"><img src="https://img.shields.io/badge/🤗%20Dataset-HER--Dataset-yellow" alt="Dataset"></a>
23
+ <a href="https://huggingface.co/ADOHAHA123/HER-Qwen3-32B"><img src="https://img.shields.io/badge/🤗%20Model-HER--RL-blue" alt="HER-RL"></a>
24
+ <a href="https://huggingface.co/ADOHAHA123/HER-SFT-Qwen3-32B"><img src="https://img.shields.io/badge/🤗%20Model-HER--SFT-green" alt="HER-SFT"></a>
25
+ <a href="https://github.com/your-username/HER"><img src="https://img.shields.io/badge/GitHub-Code-black?logo=github" alt="GitHub"></a>
26
+ </p>
27
+
28
+ HER (Human Emulation Reasoning) models are state-of-the-art models for role-playing language agents (RPLAs), built upon Qwen-32B base model. HER is a unified framework that enables cognitive-level persona simulation through structured reasoning and preference-aligned reinforcement learning.
29
+
30
+ HER models excel at role-playing through **Dual-layer Thinking**, which distinguishes between:
31
+ - **System Thinking** (third-person): LLM's meta-level planning on how to portray the character
32
+ - **Role Thinking** (first-person): Character's inner thoughts and cognitive processes
33
+
34
+ This dual-layer approach enables models to produce highly human-like responses that include reasoning traces, inner thoughts, physical actions, and natural dialogue. Extensive experiments demonstrate that HER models achieve competitive role-playing performance on multiple benchmarks, with HER-RL significantly outperforming the Qwen3-32B baseline by 30.26% on CoSER and 14.97% on MiniMax Role-Play Bench.
35
+
36
+ ## Model Variants
37
+
38
+ - **HER-SFT**: Supervised fine-tuned version from Qwen-32B
39
+ - **HER-RL**: Reinforcement learning enhanced version (this model)
40
+
41
+ ## Key Features
42
+
43
+ Our models generate responses with rich, interleaved structure:
44
+
45
+ - `<system_thinking>`: Third-person analysis of how to portray the role
46
+ - `<role_thinking>`: Character's inner thoughts (invisible to others)
47
+ - `<role_action>`: Character's physical actions and expressions
48
+ - Speech: Natural dialogue text
49
+
50
+ This hierarchical approach enables more nuanced and authentic character portrayal.
51
+
52
+ ## How to Use
53
+
54
+ ### Quick Start: Interactive Chat Demo
55
+
56
+ The easiest way to try the model is using our interactive chat demo:
57
+
58
+ ```bash
59
+ cd chat_demo
60
+ python chat_demo.py
61
+ ```
62
+
63
+ This will start an interactive session where you can:
64
+ 1. Choose a scenario from classic literature (Pride and Prejudice, The Great Gatsby, etc.)
65
+ 2. Select which character the AI should play
66
+ 3. Select which character you want to play
67
+ 4. Start chatting with the AI character!
68
+
69
+ **Demo Options:**
70
+
71
+ ```bash
72
+ # Show the model's reasoning process (system thinking)
73
+ python chat_demo.py --show-think
74
+
75
+ # Show character's inner thoughts (role thinking)
76
+ python chat_demo.py --show-rolethink
77
+
78
+ # Directly specify scenario and character
79
+ python chat_demo.py --scenario 0 --character 1
80
+ ```
81
+
82
+ **Chat Commands:**
83
+ - `quit` / `exit` / `q` - Exit the chat
84
+ - `clear` - Clear conversation history
85
+ - `history` - View conversation history
86
+ - `prompt` - View the full prompt
87
+
88
+ See `chat_demo/README.md` for detailed instructions.
89
+
90
+ ### Programmatic Usage
91
+
92
+ ```python
93
+ from transformers import AutoModelForCausalLM, AutoTokenizer
94
+
95
+ model_name = "your-username/her-qwen-32b"
96
+ tokenizer = AutoTokenizer.from_pretrained(model_name)
97
+ model = AutoModelForCausalLM.from_pretrained(
98
+ model_name,
99
+ torch_dtype="auto",
100
+ device_map="auto"
101
+ )
102
+
103
+ # Example: Role-playing as Mr. Bennet from Pride and Prejudice
104
+ system_prompt = """You are Mr Bennet from Pride and Prejudice.
105
+
106
+ ===Mr Bennet's Profile===
107
+ Elizabeth's father, known for his sarcastic wit and detachment. Mr. Bennet is the patriarch of the Bennet family, a genteel country gentleman residing at Longbourn estate in rural England.
108
+
109
+ Background: Father to five daughters (Jane, Elizabeth, Mary, Kitty, and Lydia). Owner of the Longbourn estate, which is entailed away from female inheritance.
110
+
111
+ Personality: Highly intelligent and well-read, preferring the solitude of his library. Known for his biting sarcasm and sardonic humor. Emotionally detached and often passive in family matters.
112
+
113
+ ===Current Scenario===
114
+ The scene is set in Mr. Bennet's private study. Elizabeth has been summoned unexpectedly, and Mr. Bennet holds a letter that seems to spark his characteristic sardonic amusement.
115
+
116
+ ===Output Format===
117
+ Your output should follow this structure:
118
+ 1. System Thinking: Wrapped in <system_thinking></system_thinking> tags - third-person analysis of how to portray the role
119
+ 2. Role-play Response: Including <role_thinking> for inner thoughts, <role_action> for actions, and plain text for speech"""
120
+
121
+ user_input = "[Elizabeth enters the study]"
122
+
123
+ messages = [
124
+ {"role": "system", "content": system_prompt},
125
+ {"role": "user", "content": user_input}
126
+ ]
127
+
128
+ text = tokenizer.apply_chat_template(
129
+ messages,
130
+ tokenize=False,
131
+ add_generation_prompt=True
132
+ )
133
+ model_inputs = tokenizer([text], return_tensors="pt").to(model.device)
134
+
135
+ generated_ids = model.generate(
136
+ **model_inputs,
137
+ max_new_tokens=512,
138
+ temperature=0.8,
139
+ top_p=0.9
140
+ )
141
+ generated_ids = [
142
+ output_ids[len(input_ids):] for input_ids, output_ids in zip(model_inputs.input_ids, generated_ids)
143
+ ]
144
+
145
+ response = tokenizer.batch_decode(generated_ids, skip_special_tokens=True)[0]
146
+ print(response)
147
+ ```
148
+
149
+ ## Framework Overview
150
+
151
+ <p align="center">
152
+ <img src="figure2github.png" alt="HER Framework" width="100%">
153
+ </p>
154
+
155
+ <p align="center">
156
+ <em>HER Framework: Dual-layer Thinking for Cognitive-Level Persona Simulation</em>
157
+ </p>
158
+
159
+ ## Training Methodology
160
+
161
+ HER employs a comprehensive training pipeline:
162
+
163
+ 1. **Dual-layer Thinking**: Separates hidden third-person system thinking (how the LLM plans to portray the character) from first-person role thinking (the character's actual inner thoughts). This dual-layer structure enables more authentic and cognitively grounded character simulation.
164
+
165
+ 2. **Reverse Engineering Data Synthesis**: We curate reasoning-augmented role-playing data through a three-stage reverse synthesis pipeline, constructing high-quality training trajectories with explicit reasoning traces.
166
+
167
+ 3. **Principle-Aligned Reward Model**: We construct human-aligned evaluation principles across 12 dimensions (character consistency, emotional authenticity, narrative quality, etc.) and train a Generative Reward Model (GRM) that provides detailed, case-by-case feedback.
168
+
169
+ 4. **Reinforcement Learning Enhancement** (HER-RL): Building on HER-SFT, we apply RL with the GRM to further align the model with human preferences, significantly improving interaction quality and storyline coherence.
170
+
171
+ ## Performance
172
+
173
+ ### Main Leaderboard Results
174
+
175
+ | Rank | Model | CoSER Avg | CoSER SC | CoSER AN | CoSER CF | CoSER SQ | MiniMax Avg | MiniMax Worlds (50%) | MiniMax Stories (25%) | MiniMax Pref (25%) | 95% CI |
176
+ |------|-------|-----------|----------|----------|----------|----------|-------------|----------------------|----------------------|--------------------|---------|
177
+ | 1 | Claude-4.5-Opus | **62.43** | 63.74 | **64.28** | 58.45 | 63.24 | 76.62 | 67.23 | 82.10 | 89.90 | [75.5, 77.7] |
178
+ | 2 | Gemini-3-Pro | 61.80 | **65.95** | 60.42 | **58.34** | 62.49 | 75.60 | 62.72 | 83.87 | 93.08 | [74.5, 76.7] |
179
+ | 3 | GPT-5.1 | 61.10 | 64.95 | 53.99 | 60.13 | 65.35 | 80.63 | 76.62 | 72.21 | 97.05 | [79.6, 81.6] |
180
+ | 4 | Gemini-2.5-Pro | 60.68 | 61.05 | 60.80 | 57.48 | 63.40 | 68.23 | 52.36 | 82.11 | 86.08 | [67.1, 69.3] |
181
+ | 5 | DeepSeek-v3.2 | 58.68 | 55.85 | 57.07 | 57.44 | 64.35 | 60.27 | 45.81 | 66.64 | 82.83 | [59.2, 61.4] |
182
+ | 6 | MiniMax-M2-RP | 57.30 | 60.03 | 50.11 | 49.30 | **69.77** | **84.65** | **80.55** | 79.97 | **97.51** | [83.6, 85.7] |
183
+ | 7 | DeepSeek-v3.1 | 53.50 | 50.15 | 53.18 | 53.93 | 56.72 | 64.22 | 51.11 | 66.45 | 88.21 | [62.9, 65.5] |
184
+ | **8** | **HER-RL (this model)** | **53.12** | **54.33** | **47.26** | **52.78** | **58.12** | **65.73** | **59.13** | **57.74** | **86.90** | **[63.0, 68.4]** |
185
+ | 9 | HER-SFT | 50.92 | 50.52 | 45.99 | 49.78 | 57.37 | 58.44 | 47.29 | 52.78 | 86.40 | [56.5, 60.4] |
186
+ | 10 | Grok-4.1-Fast | 47.40 | 49.21 | 47.57 | 42.64 | 50.17 | 48.47 | 29.87 | 47.51 | 86.64 | [47.4, 49.5] |
187
+ | 11 | Claude-4.5-Sonnet | 45.21 | 47.18 | 36.02 | 47.55 | 50.09 | 69.35 | 55.72 | 75.66 | 90.28 | [68.2, 70.5] |
188
+ | 12 | Claude-3.7-Think | 39.73 | 44.84 | 31.00 | 42.45 | 40.65 | 61.25 | 50.66 | 59.53 | 84.15 | [58.5, 64.0] |
189
+ | 13 | CoSER-70B | 35.95 | 35.05 | 31.16 | 32.28 | 45.33 | 45.38 | 34.32 | 30.32 | 82.58 | [43.5, 47.2] |
190
+ | 14 | GPT-5-Mini | 32.97 | 38.10 | 24.60 | 27.20 | 42.00 | 57.63 | 43.32 | 50.11 | 93.78 | [55.9, 59.3] |
191
+ | 15 | GPT-4o-240806 | 27.69 | 34.00 | 14.90 | 22.90 | 38.90 | 66.39 | 64.96 | 46.23 | 89.40 | [64.1, 68.7] |
192
+ | 16 | GPT-OSS-120B | 26.12 | 32.80 | 14.80 | 21.50 | 35.40 | 60.72 | 47.27 | 56.65 | 91.71 | [58.0, 63.4] |
193
+ | 17 | Qwen3-32B | 22.86 | 30.56 | 19.61 | 15.52 | 30.56 | 50.76 | 40.38 | 32.82 | 89.48 | [48.4, 53.2] |
194
+
195
+ **CoSER Benchmark**: Evaluates role-playing quality on 0-100 scale across four dimensions:
196
+ - **SC** (Story Consistency): Narrative coherence and plot continuity
197
+ - **AN** (Anthropomorphism): Human-like behavior and natural expression
198
+ - **CF** (Character Fidelity): Consistency with character profile and traits
199
+ - **SQ** (Storyline Quality): Overall narrative quality and engagement
200
+
201
+ **MiniMax Role-Play Bench**: Comprehensive evaluation on 0-100 scale:
202
+ - **Worlds** (50%): Basic text quality and world-building
203
+ - **Stories** (25%): Diversity and content logic
204
+ - **Preferences** (25%): Interaction quality and user satisfaction
205
+
206
+ **HER-RL** achieves **rank 8** overall, demonstrating strong performance particularly in:
207
+ - **MiniMax Preference (86.90)**: Competitive with top-tier models, showing excellent interaction quality
208
+ - **CoSER Storyline Quality (58.12)**: Strong narrative coherence and engagement
209
+ - **Balanced Performance**: Consistent scores across both benchmarks, indicating robust role-playing capabilities
210
+
211
+ The reinforcement learning approach significantly improves upon HER-SFT, particularly in interaction quality (+0.50 preference score) and character consistency (+2.52 average CoSER score).
212
+
213
+ ## Example Use Case
214
+
215
+ ### Mr. Bennet from Pride and Prejudice
216
+
217
+ **System Prompt** (abbreviated):
218
+ ```
219
+ You are Mr Bennet from Pride and Prejudice. Elizabeth's father, known for his sarcastic wit and detachment...
220
+
221
+ ===Current Scenario===
222
+ The scene is set in Mr. Bennet's private study. Elizabeth has been summoned unexpectedly...
223
+ ```
224
+
225
+ **User Input**:
226
+ ```
227
+ [Elizabeth enters the study]
228
+ ```
229
+
230
+ **Model Output**:
231
+ ```
232
+ <system_thinking>
233
+ I need to establish the opening of this scene effectively, grounding Mr Bennet in his private sanctuary while acknowledging the tension of the moment. The scenario places us in his study, a place of refuge, yet he holds a letter that disrupts his usual solitude.
234
+
235
+ Context Analysis: Elizabeth has been summoned. The atmosphere is charged. Mr Bennet has received a letter regarding Elizabeth's engagement to Darcy. He is amused by the drama but aware of the gravity of the situation.
236
+
237
+ Character Motivation: Mr Bennet often uses humor to deflect from serious emotional labor. He wants to gauge Elizabeth's state of mind regarding the recent events without being overly sentimental.
238
+
239
+ Plan:
240
+ - Action: Use the physical prop—the letter. Handling it slowly will emphasize its importance.
241
+ - Internal Thought: Acknowledge the dichotomy—finding the situation amusing yet worrying for her reputation.
242
+ - Speech: Address her directly but with a teasing opener.
243
+ </system_thinking>
244
+
245
+ Mr Bennet: <role_action>picks up the letter, tapping it thoughtfully against the desk</role_action><role_thinking>It is a delicate matter, this business with Darcy, yet I cannot help but find the drama amusing, though I must play the part of the concerned father</role_thinking>Well, my dear Lizzy, I trust you are not too greatly troubled by recent events?
246
+ ```
247
+
248
+ ## Training Data
249
+
250
+ The models are trained on a comprehensive role-playing dataset that includes:
251
+
252
+ - Authentic multi-turn, multi-character dialogues
253
+ - Character profiles with detailed background information
254
+ - Scenario descriptions and conversation contexts
255
+ - Rich annotations including system thinking, role thinking, and role actions
256
+ - Preference data collected through multi-dimensional evaluation
257
+
258
+ ## Ethical Considerations
259
+
260
+ We have conducted safety checks on the training dataset and implemented safeguards. However, users should be aware that:
261
+
262
+ - The models may generate content that reflects biases present in the training data
263
+ - Role-playing as certain characters might involve generating content with specific personality traits or behaviors
264
+ - Users should implement appropriate content filtering when deploying these models in production applications
265
+ - The models include safety evaluation dimensions to minimize harmful outputs
266
+
267
+ ## Citation
268
+
269
+ If you use HER models in your research, please cite our paper:
270
+
271
+ ```bibtex
272
+ @article{her2025,
273
+ title={HER: Human Emulation Reasoning for Cognitive-Level Role-Playing Language Models},
274
+ author={[Your Author Names]},
275
+ journal={[Conference/Journal Name]},
276
+ year={2025}
277
+ }
278
+ ```
279
+
280
+ ## License
281
+
282
+ Apache-2.0
283
+
284
+ ## Acknowledgments
285
+
286
+ This model is based on Qwen-32B developed by Alibaba Cloud. We thank the Qwen team for their excellent base model.
added_tokens.json ADDED
@@ -0,0 +1,28 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "</think>": 151668,
3
+ "</tool_call>": 151658,
4
+ "</tool_response>": 151666,
5
+ "<think>": 151667,
6
+ "<tool_call>": 151657,
7
+ "<tool_response>": 151665,
8
+ "<|box_end|>": 151649,
9
+ "<|box_start|>": 151648,
10
+ "<|endoftext|>": 151643,
11
+ "<|file_sep|>": 151664,
12
+ "<|fim_middle|>": 151660,
13
+ "<|fim_pad|>": 151662,
14
+ "<|fim_prefix|>": 151659,
15
+ "<|fim_suffix|>": 151661,
16
+ "<|im_end|>": 151645,
17
+ "<|im_start|>": 151644,
18
+ "<|image_pad|>": 151655,
19
+ "<|object_ref_end|>": 151647,
20
+ "<|object_ref_start|>": 151646,
21
+ "<|quad_end|>": 151651,
22
+ "<|quad_start|>": 151650,
23
+ "<|repo_name|>": 151663,
24
+ "<|video_pad|>": 151656,
25
+ "<|vision_end|>": 151653,
26
+ "<|vision_pad|>": 151654,
27
+ "<|vision_start|>": 151652
28
+ }
chat_demo/README.md ADDED
@@ -0,0 +1,169 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Interactive Chat Demo
2
+
3
+ This directory contains an interactive chat tool to test the HER model with character role-playing scenarios.
4
+
5
+ ## Quick Start
6
+
7
+ ```bash
8
+ # Basic chat (uses 200 CoSER scenarios from classic books)
9
+ python chat_demo.py
10
+
11
+ # Show system thinking process
12
+ python chat_demo.py --show-think
13
+
14
+ # Show role thinking
15
+ python chat_demo.py --show-rolethink
16
+
17
+ # Use simple built-in scenarios (2 scenarios: Pride and Prejudice, The Great Gatsby)
18
+ python chat_demo.py --simple
19
+ ```
20
+
21
+ ## Features
22
+
23
+ - **Character Role-Playing**: Chat with AI as characters from classic literature
24
+ - **Multi-turn Dialogue**: Maintains conversation history and context
25
+ - **Dual-layer Thinking Display**: Optional display of system thinking and role thinking
26
+ - **Format Transformation**: Converts XML tags to readable format:
27
+ - `<role_thinking>` → `[inner thought]`
28
+ - `<role_action>` → `(physical action)`
29
+ - **Auto-save**: Saves conversation logs on exit
30
+
31
+ ## Usage
32
+
33
+ ### Interactive Mode
34
+
35
+ ```bash
36
+ python chat_demo.py
37
+ ```
38
+
39
+ The script will prompt you to:
40
+ 1. Choose a scenario (book and scene)
41
+ 2. Select which character the AI should play
42
+ 3. Select which character you want to play
43
+ 4. Start chatting!
44
+
45
+ ### Commands During Chat
46
+
47
+ | Command | Function |
48
+ |---------|----------|
49
+ | `quit` / `exit` / `q` | Exit chat |
50
+ | `clear` | Clear conversation history |
51
+ | `history` | View current conversation history |
52
+ | `prompt` | View full prompt |
53
+
54
+ ## Example Scenarios
55
+
56
+ ### CoSER Dataset (200 Scenarios)
57
+
58
+ By default, the demo uses the **CoSER test dataset** (`coser_scenarios.json`) with 200 rich scenarios from classic literature:
59
+
60
+ - **Pride and Prejudice** (Elizabeth Bennet, Mr. Darcy, Mr. Bennet, etc.)
61
+ - **A Game of Thrones** (Jon Snow, Tyrion Lannister, Daenerys Targaryen, etc.)
62
+ - **The Great Gatsby** (Jay Gatsby, Nick Carraway, Daisy Buchanan, etc.)
63
+ - **To Kill a Mockingbird** (Atticus Finch, Scout Finch, etc.)
64
+ - **1984** (Winston Smith, Julia, O'Brien, etc.)
65
+ - **Harry Potter** (Harry Potter, Hermione Granger, Ron Weasley, etc.)
66
+ - **The Lord of the Rings** (Frodo, Gandalf, Aragorn, etc.)
67
+ - And 150+ more scenarios from renowned novels!
68
+
69
+ Each scenario includes:
70
+ - **Book title** and author context
71
+ - **Scene description** with detailed setting
72
+ - **Character profiles** for all participants
73
+ - **Initial character thoughts** and motivations
74
+ - **Topic/situation** summary
75
+
76
+ ### Built-in Scenarios (2 Simple Examples)
77
+
78
+ If you prefer simpler scenarios or want to test without the full dataset, use `--simple` flag to load 2 basic scenarios:
79
+ - Pride and Prejudice (Mr. Bennet and Elizabeth)
80
+ - The Great Gatsby (Gatsby and Nick Carraway)
81
+
82
+ ## Options
83
+
84
+ | Option | Description | Default |
85
+ |--------|-------------|---------|
86
+ | `--model-path` | Path to HER model directory | `.` (current dir) |
87
+ | `--show-think` | Show `<system_thinking>` | False |
88
+ | `--show-rolethink` | Show `<role_thinking>` | False |
89
+ | `--scenario` | Scenario index | Interactive |
90
+ | `--character` | Character index | Interactive |
91
+ | `--simple` | Use 2 built-in scenarios instead of 200 CoSER scenarios | False |
92
+
93
+ ## Output Format
94
+
95
+ The model generates responses with:
96
+
97
+ 1. **System Thinking** (optional display):
98
+ - Third-person analysis of how to portray the character
99
+ - Planning and reasoning about the response
100
+
101
+ 2. **Role Response**:
102
+ - **Role Thinking** `[...]`: Character's inner thoughts (invisible to others)
103
+ - **Role Action** `(...)`: Physical actions and expressions
104
+ - **Speech**: Natural dialogue
105
+
106
+ ### Example Output
107
+
108
+ ```
109
+ ════════════════════════════════════════════════════════════════════════════════
110
+ 🎭 【Elizabeth Bennet's Response】
111
+ ════════════════════════════════════════════════════════════════════════════════
112
+ [His tone is light, but the air feels heavy. I cannot let him see how much
113
+ Lady Catherine's intrusion still stings.]
114
+ (takes a steadying breath, smoothing the folds of her dress)
115
+ I believe I can manage, Father. Though I must admit, I am curious about
116
+ what this letter contains.
117
+ ════════════════════════════════════════════════════════════════════════════════
118
+ ```
119
+
120
+ ## Requirements
121
+
122
+ - Python 3.8+
123
+ - transformers
124
+ - torch
125
+
126
+ Install dependencies:
127
+ ```bash
128
+ pip install transformers torch
129
+ ```
130
+
131
+ ## File Structure
132
+
133
+ ```
134
+ chat_demo/
135
+ ├── README.md # This file
136
+ ├── chat_demo.py # Main chat script
137
+ ├── coser_scenarios.json # 200 CoSER test scenarios (2.4MB)
138
+ ├── scenarios.json # Simple built-in scenarios (auto-created if using --simple)
139
+ └── chat_logs/ # Saved chat logs (auto-created)
140
+ └── {book}_{character}_{timestamp}.txt
141
+ ```
142
+
143
+ ## Notes
144
+
145
+ 1. **System Thinking**: Used for training and analysis, not included in conversation history
146
+ 2. **Role Thinking**: Character's inner thoughts, invisible to other characters
147
+ 3. **Role Action**: Physical behaviors visible to others
148
+ 4. **Speech**: What the character says out loud
149
+
150
+ ## Tips for Best Results
151
+
152
+ - Stay in character when chatting
153
+ - Provide context in your messages
154
+ - Use the character's background knowledge
155
+ - Be patient - the model generates thoughtful responses with reasoning
156
+
157
+ ## Troubleshooting
158
+
159
+ **Model not loading?**
160
+ - Ensure the model files are in the correct directory
161
+ - Check that you have enough GPU memory
162
+
163
+ **Empty responses?**
164
+ - Try adjusting temperature (default: 0.7)
165
+ - Check the prompt format
166
+
167
+ **Inconsistent character behavior?**
168
+ - Review the character profile
169
+ - Ensure your messages align with the scenario context
chat_demo/chat_demo.py ADDED
@@ -0,0 +1,536 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ #!/usr/bin/env python3
2
+ """
3
+ Interactive Chat Demo for HER Model
4
+ Chat with AI characters from classic literature using role-playing scenarios.
5
+
6
+ Usage:
7
+ python chat_demo.py
8
+ python chat_demo.py --show-think
9
+ python chat_demo.py --show-rolethink
10
+ """
11
+
12
+ import re
13
+ import json
14
+ import argparse
15
+ from pathlib import Path
16
+ from datetime import datetime
17
+ from transformers import AutoModelForCausalLM, AutoTokenizer
18
+
19
+ # Colors for terminal output
20
+ class Colors:
21
+ HEADER = '\033[95m'
22
+ BLUE = '\033[94m'
23
+ CYAN = '\033[96m'
24
+ GREEN = '\033[92m'
25
+ YELLOW = '\033[93m'
26
+ RED = '\033[91m'
27
+ BOLD = '\033[1m'
28
+ UNDERLINE = '\033[4m'
29
+ END = '\033[0m'
30
+ GRAY = '\033[90m'
31
+ MAGENTA = '\033[35m'
32
+
33
+
34
+ def remove_system_thinking(text: str) -> str:
35
+ """Remove <system_thinking>...</system_thinking> tags and content"""
36
+ if not text:
37
+ return text
38
+ pattern = r'<system_thinking>.*?</system_thinking>\s*'
39
+ cleaned = re.sub(pattern, '', text, flags=re.DOTALL)
40
+ return cleaned.strip()
41
+
42
+
43
+ def extract_system_thinking(text: str) -> str:
44
+ """Extract system_thinking content (without role tags inside)"""
45
+ if not text:
46
+ return ""
47
+ match = re.search(r'<system_thinking>(.*?)</system_thinking>', text, flags=re.DOTALL)
48
+ if match:
49
+ content = match.group(1).strip()
50
+ # Remove any role tags that might have leaked in
51
+ content = re.sub(r'</?role_\w+>', '', content)
52
+ return content
53
+ return ""
54
+
55
+
56
+ def format_for_display(text: str, show_rolethink: bool = True) -> str:
57
+ """Format for display: replace role_thinking with [], role_action with ()"""
58
+ if not text:
59
+ return text
60
+
61
+ result = text
62
+
63
+ # Handle role_thinking
64
+ if show_rolethink:
65
+ result = result.replace('<role_thinking>', '[').replace('</role_thinking>', ']')
66
+ else:
67
+ result = re.sub(r'<role_thinking>.*?</role_thinking>', '', result, flags=re.DOTALL)
68
+
69
+ # Replace role_action with ()
70
+ result = result.replace('<role_action>', '(').replace('</role_action>', ')')
71
+ result = result.replace('<role_speech>', '').replace('</role_speech>', '')
72
+
73
+ return result.strip()
74
+
75
+
76
+ def load_sample_scenarios(use_coser: bool = True):
77
+ """Load or create sample scenarios
78
+
79
+ Args:
80
+ use_coser: If True, try to load CoSER scenarios first (200 scenarios from classic books)
81
+ """
82
+ # Try to load CoSER scenarios if available
83
+ if use_coser:
84
+ coser_file = Path(__file__).parent / "coser_scenarios.json"
85
+ if coser_file.exists():
86
+ print(f"{Colors.CYAN}📚 Loading CoSER scenarios (200 book scenes)...{Colors.END}")
87
+ with open(coser_file, 'r', encoding='utf-8') as f:
88
+ return json.load(f)
89
+
90
+ # Otherwise, use built-in scenarios
91
+ scenarios_file = Path(__file__).parent / "scenarios.json"
92
+
93
+ # If scenarios file exists, load it
94
+ if scenarios_file.exists():
95
+ with open(scenarios_file, 'r', encoding='utf-8') as f:
96
+ return json.load(f)
97
+
98
+ # Otherwise create sample scenarios
99
+ scenarios = [
100
+ {
101
+ "book": "Pride and Prejudice",
102
+ "topic": "Mr. Bennet confronts Elizabeth about Mr. Darcy's proposal",
103
+ "scenario": "The scene is set in Mr. Bennet's private study, a sanctuary of leather-bound books and quiet contemplation. Elizabeth has been summoned unexpectedly, and Mr. Bennet holds a letter that seems to spark his characteristic sardonic amusement.",
104
+ "character_profiles": {
105
+ "Mr Bennet": "Elizabeth's father, known for his sarcastic wit and detachment. Highly intelligent and well-read, preferring the solitude of his library. Known for his biting sarcasm and sardonic humor.",
106
+ "Elizabeth Bennet": "The protagonist, intelligent and strong-willed. Quick-witted with a playful sense of humor. Values honesty and integrity. Maintains composure under pressure."
107
+ },
108
+ "key_characters": [
109
+ {
110
+ "name": "Mr Bennet",
111
+ "thought": "It is a delicate matter, this business with Darcy. I must gauge Elizabeth's true feelings without being overly sentimental."
112
+ },
113
+ {
114
+ "name": "Elizabeth Bennet",
115
+ "thought": "Father's summoning me at this hour is unusual. I hope this isn't about Lady Catherine's visit."
116
+ }
117
+ ]
118
+ },
119
+ {
120
+ "book": "The Great Gatsby",
121
+ "topic": "Nick Carraway encounters Gatsby at one of his lavish parties",
122
+ "scenario": "The party is in full swing at Gatsby's mansion. Jazz music fills the air, champagne flows freely, and well-dressed guests mingle on the lawn. Nick has been wandering alone, observing the spectacle, when he encounters a mysterious man by the library.",
123
+ "character_profiles": {
124
+ "Jay Gatsby": "The enigmatic millionaire who throws lavish parties. Behind his elegant facade lies a romantic dreamer obsessed with recapturing the past. Charming yet deeply lonely.",
125
+ "Nick Carraway": "The story's narrator, a Yale graduate from the Midwest. Honest, tolerant, and inclined to reserve judgment. Both drawn to and repelled by the excess around him."
126
+ },
127
+ "key_characters": [
128
+ {
129
+ "name": "Jay Gatsby",
130
+ "thought": "Another party, another night of waiting. Perhaps tonight she'll come. I must maintain appearances."
131
+ },
132
+ {
133
+ "name": "Nick Carraway",
134
+ "thought": "I've never met my host. These parties are magnificent, yet there's something hollow about all this revelry."
135
+ }
136
+ ]
137
+ }
138
+ ]
139
+
140
+ # Save scenarios for future use
141
+ with open(scenarios_file, 'w', encoding='utf-8') as f:
142
+ json.dump(scenarios, f, indent=2, ensure_ascii=False)
143
+
144
+ return scenarios
145
+
146
+
147
+ def print_scenarios(scenarios: list):
148
+ """Print available scenarios"""
149
+ print(f"\n{Colors.HEADER}{'='*80}{Colors.END}")
150
+ print(f"{Colors.HEADER}📚 Available Scenarios{Colors.END}")
151
+ print(f"{Colors.HEADER}{'='*80}{Colors.END}\n")
152
+
153
+ for i, s in enumerate(scenarios):
154
+ print(f"{Colors.CYAN}[{i}]{Colors.END} {Colors.BOLD}📖 {s['book']}{Colors.END}")
155
+ print(f" {Colors.GRAY}{s['topic']}{Colors.END}")
156
+ chars = list(s['character_profiles'].keys())
157
+ print(f" {Colors.MAGENTA}👥 Characters: {', '.join(chars)}{Colors.END}")
158
+ print()
159
+
160
+
161
+ def print_characters(scenario: dict):
162
+ """Print available characters in the scenario"""
163
+ print(f"\n{Colors.HEADER}{'='*80}{Colors.END}")
164
+ print(f"{Colors.HEADER}👥 Available Characters - {scenario['book']}{Colors.END}")
165
+ print(f"{Colors.HEADER}{'='*80}{Colors.END}\n")
166
+
167
+ for i, (name, profile) in enumerate(scenario['character_profiles'].items()):
168
+ print(f"{Colors.CYAN}[{i}]{Colors.END} {Colors.BOLD}{name}{Colors.END}")
169
+ preview = profile[:150] + "..." if len(profile) > 150 else profile
170
+ print(f" {Colors.GRAY}{preview}{Colors.END}")
171
+ print()
172
+
173
+
174
+ def build_system_prompt(scenario: dict, character_name: str, user_character_name: str) -> str:
175
+ """Build system prompt for the character"""
176
+ book = scenario['book']
177
+ scene = scenario['scenario']
178
+ profiles = scenario['character_profiles']
179
+
180
+ char_profile = profiles.get(character_name, "")
181
+ user_profile = profiles.get(user_character_name, "A person interacting with the character.")
182
+
183
+ # Find character's initial thought
184
+ char_thought = ""
185
+ for kc in scenario['key_characters']:
186
+ if kc['name'] == character_name:
187
+ char_thought = kc.get('thought', '')
188
+ break
189
+
190
+ prompt = f"""You are role-playing as {character_name} from the book "{book}".
191
+
192
+ ==={character_name}'s Profile===
193
+ {char_profile}
194
+
195
+ ===Current Scene===
196
+ {scene}
197
+
198
+ ===Your Current Thoughts===
199
+ {char_thought}
200
+
201
+ ===The Person You Are Interacting With===
202
+ {user_character_name}: {user_profile}
203
+
204
+ ===Instructions===
205
+ - Stay in character as {character_name} at all times
206
+ - Keep responses natural and engaging, consistent with the book's style
207
+ - Respond from {character_name}'s perspective
208
+ - **IMPORTANT: Speak DIRECTLY to "{user_character_name}" using "you" (second person). Do NOT use third person.**
209
+
210
+ ===Output Format===
211
+ Your output should include thought, speech, and action in this two-part structure:
212
+
213
+ 1. System Thinking: A single block at the very beginning, wrapped in <system_thinking> and </system_thinking>. This is third-person analysis of how to portray the character.
214
+
215
+ 2. Role-play Response: The character's actual response including:
216
+ - <role_thinking>inner thoughts</role_thinking> (invisible to others)
217
+ - <role_action>physical actions</role_action> (visible to others)
218
+ - Speech (plain text, what the character says out loud)"""
219
+
220
+ return prompt
221
+
222
+
223
+ def save_chat_log(messages: list, scenario: dict, character_name: str, user_character: str):
224
+ """Save chat log to file"""
225
+ log_dir = Path(__file__).parent / "chat_logs"
226
+ log_dir.mkdir(exist_ok=True)
227
+
228
+ safe_book = re.sub(r'[^\w\-]', '_', scenario['book'])[:30]
229
+ safe_char = re.sub(r'[^\w\-]', '_', character_name)[:20]
230
+ timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
231
+ filename = f"{safe_book}_{safe_char}_{timestamp}.txt"
232
+ filepath = log_dir / filename
233
+
234
+ lines = [
235
+ "=" * 80,
236
+ "HER Chat Demo - Conversation Log",
237
+ "=" * 80,
238
+ f"Time: {datetime.now().strftime('%Y-%m-%d %H:%M:%S')}",
239
+ f"Book: {scenario['book']}",
240
+ f"AI Character: {character_name}",
241
+ f"User Character: {user_character}",
242
+ "=" * 80,
243
+ "",
244
+ "【Scene】",
245
+ scenario['scenario'][:500],
246
+ "",
247
+ "=" * 80,
248
+ "【Conversation】",
249
+ "=" * 80,
250
+ ]
251
+
252
+ for msg in messages:
253
+ role = msg['role']
254
+ content = msg['content']
255
+ if role == 'system':
256
+ continue
257
+ elif role == 'user':
258
+ if "===Conversation Start===" not in content:
259
+ lines.append(f"\n【{user_character}】")
260
+ lines.append(content)
261
+ elif role == 'assistant':
262
+ lines.append(f"\n【{character_name}】")
263
+ lines.append(content)
264
+
265
+ lines.extend(["\n" + "=" * 80, "--- End of Conversation ---"])
266
+
267
+ with open(filepath, 'w', encoding='utf-8') as f:
268
+ f.write('\n'.join(lines))
269
+
270
+ return filepath
271
+
272
+
273
+ def chat_loop(model, tokenizer, scenario: dict, character_name: str, user_character: str,
274
+ show_think: bool = False, show_rolethink: bool = True):
275
+ """Main chat loop"""
276
+ book = scenario['book']
277
+
278
+ print(f"\n{Colors.HEADER}{'='*80}{Colors.END}")
279
+ print(f"{Colors.HEADER}🎭 Starting Conversation - {book}{Colors.END}")
280
+ print(f"{Colors.HEADER}{'='*80}{Colors.END}")
281
+ print(f"{Colors.GREEN}You play: {user_character}{Colors.END}")
282
+ print(f"{Colors.MAGENTA}AI plays: {character_name}{Colors.END}")
283
+ print(f"{Colors.GRAY}Show system_thinking: {'Yes' if show_think else 'No'}{Colors.END}")
284
+ print(f"{Colors.GRAY}Show role_thinking: {'Yes' if show_rolethink else 'No'}{Colors.END}")
285
+ print(f"{Colors.GRAY}Commands: 'quit' to exit, 'clear' to reset, 'history' to view{Colors.END}")
286
+ print(f"{Colors.HEADER}{'='*80}{Colors.END}\n")
287
+
288
+ # Display scene
289
+ print(f"{Colors.CYAN}📍 Scene:{Colors.END}")
290
+ print(f"{Colors.GRAY}{scenario['scenario'][:300]}...{Colors.END}\n")
291
+
292
+ # Build messages
293
+ system_prompt = build_system_prompt(scenario, character_name, user_character)
294
+
295
+ # Initial greeting
296
+ greeting = f"*{character_name} looks at you*"
297
+ for kc in scenario.get('key_characters', []):
298
+ if kc['name'] == character_name:
299
+ greeting = f"*enters the scene* Hello, {user_character}."
300
+ break
301
+
302
+ messages = [
303
+ {"role": "system", "content": system_prompt},
304
+ {"role": "user", "content": "===Conversation Start==="},
305
+ {"role": "assistant", "content": greeting}
306
+ ]
307
+
308
+ print(f"{Colors.GREEN}{character_name}:{Colors.END} {greeting}\n")
309
+
310
+ while True:
311
+ try:
312
+ user_input = input(f"{Colors.BLUE}{user_character}:{Colors.END} ").strip()
313
+
314
+ if not user_input:
315
+ continue
316
+
317
+ if user_input.lower() in ['quit', 'exit', 'q']:
318
+ print(f"\n{Colors.YELLOW}👋 Goodbye!{Colors.END}")
319
+ break
320
+
321
+ if user_input.lower() == 'clear':
322
+ messages = [
323
+ {"role": "system", "content": system_prompt},
324
+ {"role": "user", "content": "===Conversation Start==="},
325
+ {"role": "assistant", "content": greeting}
326
+ ]
327
+ print(f"{Colors.YELLOW}🔄 Conversation history cleared{Colors.END}\n")
328
+ print(f"{Colors.GREEN}{character_name}:{Colors.END} {greeting}\n")
329
+ continue
330
+
331
+ if user_input.lower() == 'history':
332
+ print(f"\n{Colors.CYAN}📜 Conversation History ({len(messages)} messages):{Colors.END}")
333
+ for i, msg in enumerate(messages[1:], 1):
334
+ content = msg['content'][:80] + '...' if len(msg['content']) > 80 else msg['content']
335
+ print(f" [{i}] {msg['role']}: {content}")
336
+ print()
337
+ continue
338
+
339
+ # Add user message
340
+ messages.append({"role": "user", "content": user_input})
341
+
342
+ # Generate response
343
+ print(f"{Colors.GRAY}⏳ Thinking...{Colors.END}", end='\r')
344
+
345
+ # Format messages for model
346
+ text = tokenizer.apply_chat_template(
347
+ messages + [{"role": "assistant", "content": "<system_thinking>"}],
348
+ tokenize=False,
349
+ add_generation_prompt=False
350
+ )
351
+
352
+ inputs = tokenizer([text], return_tensors="pt").to(model.device)
353
+
354
+ try:
355
+ outputs = model.generate(
356
+ **inputs,
357
+ max_new_tokens=1024,
358
+ temperature=0.7,
359
+ top_p=0.9,
360
+ do_sample=True,
361
+ pad_token_id=tokenizer.eos_token_id
362
+ )
363
+
364
+ response = tokenizer.decode(outputs[0][len(inputs[0]):], skip_special_tokens=False)
365
+
366
+ # Clean up response
367
+ response = response.replace("<|im_end|>", "").replace("<|im_start|>", "").strip()
368
+
369
+ full_response = "<system_thinking>" + response
370
+ clean_response = remove_system_thinking(full_response)
371
+
372
+ except Exception as e:
373
+ print(f"{Colors.RED}❌ Generation failed: {e}{Colors.END}")
374
+ messages.pop()
375
+ continue
376
+
377
+ print(" " * 50, end='\r')
378
+
379
+ # Display system thinking if requested
380
+ if show_think:
381
+ think_content = extract_system_thinking(full_response)
382
+ if think_content:
383
+ print(f"\n{Colors.GRAY}{'─'*80}{Colors.END}")
384
+ print(f"{Colors.GRAY}📝 【System Thinking】{Colors.END}")
385
+ print(f"{Colors.GRAY}{'─'*80}{Colors.END}")
386
+ for line in think_content.split('\n')[:10]: # Limit lines
387
+ print(f"{Colors.GRAY} {line}{Colors.END}")
388
+ print(f"{Colors.GRAY}{'─'*80}{Colors.END}\n")
389
+
390
+ # Display character response
391
+ print(f"{Colors.GREEN}{'═'*80}{Colors.END}")
392
+ print(f"{Colors.GREEN}🎭 【{character_name}'s Response】{Colors.END}")
393
+ print(f"{Colors.GREEN}{'═'*80}{Colors.END}")
394
+ display_response = format_for_display(clean_response, show_rolethink=show_rolethink)
395
+ print(f"{Colors.GREEN}{display_response}{Colors.END}")
396
+ print(f"{Colors.GREEN}{'═'*80}{Colors.END}\n")
397
+
398
+ messages.append({"role": "assistant", "content": clean_response})
399
+
400
+ except KeyboardInterrupt:
401
+ print(f"\n{Colors.YELLOW}👋 Goodbye!{Colors.END}")
402
+ break
403
+ except EOFError:
404
+ print(f"\n{Colors.YELLOW}👋 Goodbye!{Colors.END}")
405
+ break
406
+
407
+ return messages
408
+
409
+
410
+ def main():
411
+ parser = argparse.ArgumentParser(description="Interactive Chat Demo for HER Model")
412
+ parser.add_argument("--model-path", type=str, default=".",
413
+ help="Path to model directory (default: current directory)")
414
+ parser.add_argument("--show-think", action="store_true",
415
+ help="Show system_thinking")
416
+ parser.add_argument("--show-rolethink", action="store_true",
417
+ help="Show role_thinking (default: hidden)")
418
+ parser.add_argument("--scenario", type=int, default=None,
419
+ help="Scenario index (default: interactive selection)")
420
+ parser.add_argument("--character", type=int, default=None,
421
+ help="Character index (default: interactive selection)")
422
+ parser.add_argument("--simple", action="store_true",
423
+ help="Use simple built-in scenarios instead of CoSER dataset")
424
+
425
+ args = parser.parse_args()
426
+
427
+ # Load scenarios
428
+ scenarios = load_sample_scenarios(use_coser=not args.simple)
429
+ print(f"{Colors.GREEN}✅ Loaded {len(scenarios)} scenarios{Colors.END}")
430
+
431
+ # Select scenario
432
+ if args.scenario is not None:
433
+ if 0 <= args.scenario < len(scenarios):
434
+ scenario = scenarios[args.scenario]
435
+ else:
436
+ print(f"{Colors.RED}❌ Invalid scenario index{Colors.END}")
437
+ return
438
+ else:
439
+ print_scenarios(scenarios)
440
+ while True:
441
+ try:
442
+ idx = int(input(f"{Colors.CYAN}Select scenario (0-{len(scenarios)-1}): {Colors.END}"))
443
+ if 0 <= idx < len(scenarios):
444
+ scenario = scenarios[idx]
445
+ break
446
+ print(f"{Colors.RED}Invalid index{Colors.END}")
447
+ except (ValueError, KeyboardInterrupt, EOFError):
448
+ print(f"\n{Colors.YELLOW}👋 Goodbye!{Colors.END}")
449
+ return
450
+
451
+ print(f"\n{Colors.GREEN}✅ Selected: {scenario['book']}{Colors.END}")
452
+
453
+ # Select character
454
+ char_names = list(scenario['character_profiles'].keys())
455
+ print_characters(scenario)
456
+
457
+ if args.character is not None:
458
+ if 0 <= args.character < len(char_names):
459
+ character_name = char_names[args.character]
460
+ else:
461
+ print(f"{Colors.RED}❌ Invalid character index{Colors.END}")
462
+ return
463
+ else:
464
+ while True:
465
+ try:
466
+ idx = int(input(f"{Colors.CYAN}Select AI character (0-{len(char_names)-1}): {Colors.END}"))
467
+ if 0 <= idx < len(char_names):
468
+ character_name = char_names[idx]
469
+ break
470
+ print(f"{Colors.RED}Invalid index{Colors.END}")
471
+ except (ValueError, KeyboardInterrupt, EOFError):
472
+ print(f"\n{Colors.YELLOW}👋 Goodbye!{Colors.END}")
473
+ return
474
+
475
+ # Select user character
476
+ remaining_chars = [c for c in char_names if c != character_name]
477
+ if remaining_chars:
478
+ print(f"\n{Colors.CYAN}Who do you want to play?{Colors.END}")
479
+ for i, c in enumerate(remaining_chars):
480
+ print(f" [{i}] {c}")
481
+ print(f" [{len(remaining_chars)}] Custom name")
482
+
483
+ while True:
484
+ try:
485
+ idx = int(input(f"{Colors.CYAN}Select (0-{len(remaining_chars)}): {Colors.END}"))
486
+ if idx == len(remaining_chars):
487
+ user_character = input(f"{Colors.CYAN}Your name: {Colors.END}").strip() or "User"
488
+ break
489
+ elif 0 <= idx < len(remaining_chars):
490
+ user_character = remaining_chars[idx]
491
+ break
492
+ except (ValueError, KeyboardInterrupt, EOFError):
493
+ print(f"\n{Colors.YELLOW}👋 Goodbye!{Colors.END}")
494
+ return
495
+ else:
496
+ user_character = "User"
497
+
498
+ print(f"\n{Colors.GREEN}✅ AI plays: {character_name}{Colors.END}")
499
+ print(f"{Colors.BLUE}✅ You play: {user_character}{Colors.END}")
500
+
501
+ # Load model
502
+ print(f"\n{Colors.CYAN}🔧 Loading model...{Colors.END}")
503
+ try:
504
+ tokenizer = AutoTokenizer.from_pretrained(args.model_path)
505
+ model = AutoModelForCausalLM.from_pretrained(
506
+ args.model_path,
507
+ torch_dtype="auto",
508
+ device_map="auto"
509
+ )
510
+ print(f"{Colors.GREEN}✅ Model loaded successfully{Colors.END}")
511
+ except Exception as e:
512
+ print(f"{Colors.RED}❌ Model loading failed: {e}{Colors.END}")
513
+ return
514
+
515
+ # Start chat
516
+ messages = chat_loop(
517
+ model,
518
+ tokenizer,
519
+ scenario,
520
+ character_name,
521
+ user_character,
522
+ show_think=args.show_think,
523
+ show_rolethink=args.show_rolethink
524
+ )
525
+
526
+ # Save log
527
+ if messages and len(messages) > 3:
528
+ try:
529
+ log_path = save_chat_log(messages, scenario, character_name, user_character)
530
+ print(f"\n{Colors.CYAN}📝 Chat log saved: {log_path}{Colors.END}")
531
+ except Exception as e:
532
+ print(f"\n{Colors.RED}❌ Save failed: {e}{Colors.END}")
533
+
534
+
535
+ if __name__ == "__main__":
536
+ main()
chat_demo/coser_scenarios.json ADDED
The diff for this file is too large to render. See raw diff
 
chat_template.jinja ADDED
@@ -0,0 +1,85 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- if tools %}
2
+ {{- '<|im_start|>system\n' }}
3
+ {%- if messages[0].role == 'system' %}
4
+ {{- messages[0].content + '\n\n' }}
5
+ {%- endif %}
6
+ {{- "# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>" }}
7
+ {%- for tool in tools %}
8
+ {{- "\n" }}
9
+ {{- tool | tojson }}
10
+ {%- endfor %}
11
+ {{- "\n</tools>\n\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\n<tool_call>\n{\"name\": <function-name>, \"arguments\": <args-json-object>}\n</tool_call><|im_end|>\n" }}
12
+ {%- else %}
13
+ {%- if messages[0].role == 'system' %}
14
+ {{- '<|im_start|>system\n' + messages[0].content + '<|im_end|>\n' }}
15
+ {%- endif %}
16
+ {%- endif %}
17
+ {%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
18
+ {%- for message in messages[::-1] %}
19
+ {%- set index = (messages|length - 1) - loop.index0 %}
20
+ {%- if ns.multi_step_tool and message.role == "user" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}
21
+ {%- set ns.multi_step_tool = false %}
22
+ {%- set ns.last_query_index = index %}
23
+ {%- endif %}
24
+ {%- endfor %}
25
+ {%- for message in messages %}
26
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
27
+ {{- '<|im_start|>' + message.role + '\n' + message.content + '<|im_end|>' + '\n' }}
28
+ {%- elif message.role == "assistant" %}
29
+ {%- set content = message.content %}
30
+ {%- set reasoning_content = '' %}
31
+ {%- if message.reasoning_content is defined and message.reasoning_content is not none %}
32
+ {%- set reasoning_content = message.reasoning_content %}
33
+ {%- else %}
34
+ {%- if '</think>' in message.content %}
35
+ {%- set content = message.content.split('</think>')[-1].lstrip('\n') %}
36
+ {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
37
+ {%- endif %}
38
+ {%- endif %}
39
+ {%- if loop.index0 > ns.last_query_index %}
40
+ {%- if loop.last or (not loop.last and reasoning_content) %}
41
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
42
+ {%- else %}
43
+ {{- '<|im_start|>' + message.role + '\n' + content }}
44
+ {%- endif %}
45
+ {%- else %}
46
+ {{- '<|im_start|>' + message.role + '\n' + content }}
47
+ {%- endif %}
48
+ {%- if message.tool_calls %}
49
+ {%- for tool_call in message.tool_calls %}
50
+ {%- if (loop.first and content) or (not loop.first) %}
51
+ {{- '\n' }}
52
+ {%- endif %}
53
+ {%- if tool_call.function %}
54
+ {%- set tool_call = tool_call.function %}
55
+ {%- endif %}
56
+ {{- '<tool_call>\n{"name": "' }}
57
+ {{- tool_call.name }}
58
+ {{- '", "arguments": ' }}
59
+ {%- if tool_call.arguments is string %}
60
+ {{- tool_call.arguments }}
61
+ {%- else %}
62
+ {{- tool_call.arguments | tojson }}
63
+ {%- endif %}
64
+ {{- '}\n</tool_call>' }}
65
+ {%- endfor %}
66
+ {%- endif %}
67
+ {{- '<|im_end|>\n' }}
68
+ {%- elif message.role == "tool" %}
69
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
70
+ {{- '<|im_start|>user' }}
71
+ {%- endif %}
72
+ {{- '\n<tool_response>\n' }}
73
+ {{- message.content }}
74
+ {{- '\n</tool_response>' }}
75
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
76
+ {{- '<|im_end|>\n' }}
77
+ {%- endif %}
78
+ {%- endif %}
79
+ {%- endfor %}
80
+ {%- if add_generation_prompt %}
81
+ {{- '<|im_start|>assistant\n' }}
82
+ {%- if enable_thinking is defined and enable_thinking is false %}
83
+ {{- '<think>\n\n</think>\n\n' }}
84
+ {%- endif %}
85
+ {%- endif %}
config.json ADDED
@@ -0,0 +1,34 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "Qwen3ForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "eos_token_id": 151645,
8
+ "head_dim": 128,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 5120,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 25600,
13
+ "max_position_embeddings": 131072,
14
+ "max_window_layers": 64,
15
+ "model_type": "qwen3",
16
+ "num_attention_heads": 64,
17
+ "num_hidden_layers": 64,
18
+ "num_key_value_heads": 8,
19
+ "pad_token_id": 151643,
20
+ "rms_norm_eps": 1e-06,
21
+ "rope_scaling": {
22
+ "factor": 4.0,
23
+ "original_max_position_embeddings": 32768,
24
+ "rope_type": "yarn"
25
+ },
26
+ "rope_theta": 1000000,
27
+ "sliding_window": null,
28
+ "tie_word_embeddings": false,
29
+ "torch_dtype": "bfloat16",
30
+ "transformers_version": "4.52.4",
31
+ "use_cache": true,
32
+ "use_sliding_window": false,
33
+ "vocab_size": 151936
34
+ }
figure2github.png ADDED

Git LFS Details

  • SHA256: 657ea86a273a40a2ec1c930778fe18db9cf570d8bebcea754af5715b965df831
  • Pointer size: 132 Bytes
  • Size of remote file: 2.86 MB
generation_config.json ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151643,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 151645,
6
+ 151643
7
+ ],
8
+ "pad_token_id": 151643,
9
+ "temperature": 0.6,
10
+ "top_k": 20,
11
+ "top_p": 0.95,
12
+ "transformers_version": "4.52.4"
13
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model-00001-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:30e6f85e21a0e9d3483e8bb9e6010c80a8e797719552329ae795f76a26b52447
3
+ size 4928419424
model-00002-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:99db49822e9f64fd5502fd0d42e2e5a6fb43df4206fe161cb1c8120e32670637
3
+ size 4781605144
model-00003-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e07a88f5427346dfe5224e4b2c81368b18a293ae92b747c1e1a0cd18411032c5
3
+ size 4928450568
model-00004-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5888990fba12830a1c24bc6cc470697284f57fa3ff9bc68abb2f27c15ec9ae08
3
+ size 4980813680
model-00005-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:493e7423c2354cdc29e88cb6bbd439b9fe0a0c5dc44c9df7f912ef36f3de6997
3
+ size 4991315040
model-00006-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:540b426e81ddf3232e20258648997d52ece0532c0d3b88742c0dcd232beedda3
3
+ size 4949367504
model-00007-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:39b294addd72f12cb2c04c6ff91856a20f48ffc5e5300508ee3aa261a51b86c0
3
+ size 4854996904
model-00008-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:3a5971c2dc9c302a74844a6a8c0ac3c18776892563d5d07c6d473dbdd2e46d5b
3
+ size 3565289480
model-00009-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f948de65d6c0c4da9aa5c7b1b5634c10a92dd422f494c9554ffe06b9a68cd758
3
+ size 4890335312
model-00010-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b48a6226ab1fead18741840d0c6c2ff7a713a182210468b5861586b9a05656c7
3
+ size 4785473976
model-00011-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:307afed4ca7ac0aa48c4faf376bb3d654e143f2a6d4145b77416d8afc06cca2a
3
+ size 4834025448
model-00012-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0533a6fe98af4e719b356da4f5a996defcd6614f66fbd5c3787497ef0f9894e
3
+ size 4792103184
model-00013-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47c9088b8ab86b8c928cdfbb9a38e390e0a0ac716285e348e5ed4fcfd0854c70
3
+ size 4949465424
model-00014-of-00014.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31d997e89fb92893911b4214b4a51f7c6ed4ab25fccf47488ca5373b964e66e9
3
+ size 3292667304
model.safetensors.index.json ADDED
@@ -0,0 +1,714 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_size": 65524246528
4
+ },
5
+ "weight_map": {
6
+ "lm_head.weight": "model-00009-of-00014.safetensors",
7
+ "model.embed_tokens.weight": "model-00010-of-00014.safetensors",
8
+ "model.layers.0.input_layernorm.weight": "model-00008-of-00014.safetensors",
9
+ "model.layers.0.mlp.down_proj.weight": "model-00007-of-00014.safetensors",
10
+ "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00014.safetensors",
11
+ "model.layers.0.mlp.up_proj.weight": "model-00001-of-00014.safetensors",
12
+ "model.layers.0.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
13
+ "model.layers.0.self_attn.k_norm.weight": "model-00010-of-00014.safetensors",
14
+ "model.layers.0.self_attn.k_proj.weight": "model-00002-of-00014.safetensors",
15
+ "model.layers.0.self_attn.o_proj.weight": "model-00009-of-00014.safetensors",
16
+ "model.layers.0.self_attn.q_norm.weight": "model-00011-of-00014.safetensors",
17
+ "model.layers.0.self_attn.q_proj.weight": "model-00013-of-00014.safetensors",
18
+ "model.layers.0.self_attn.v_proj.weight": "model-00008-of-00014.safetensors",
19
+ "model.layers.1.input_layernorm.weight": "model-00010-of-00014.safetensors",
20
+ "model.layers.1.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
21
+ "model.layers.1.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
22
+ "model.layers.1.mlp.up_proj.weight": "model-00012-of-00014.safetensors",
23
+ "model.layers.1.post_attention_layernorm.weight": "model-00006-of-00014.safetensors",
24
+ "model.layers.1.self_attn.k_norm.weight": "model-00004-of-00014.safetensors",
25
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00014.safetensors",
26
+ "model.layers.1.self_attn.o_proj.weight": "model-00011-of-00014.safetensors",
27
+ "model.layers.1.self_attn.q_norm.weight": "model-00009-of-00014.safetensors",
28
+ "model.layers.1.self_attn.q_proj.weight": "model-00012-of-00014.safetensors",
29
+ "model.layers.1.self_attn.v_proj.weight": "model-00013-of-00014.safetensors",
30
+ "model.layers.10.input_layernorm.weight": "model-00005-of-00014.safetensors",
31
+ "model.layers.10.mlp.down_proj.weight": "model-00012-of-00014.safetensors",
32
+ "model.layers.10.mlp.gate_proj.weight": "model-00004-of-00014.safetensors",
33
+ "model.layers.10.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
34
+ "model.layers.10.post_attention_layernorm.weight": "model-00008-of-00014.safetensors",
35
+ "model.layers.10.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
36
+ "model.layers.10.self_attn.k_proj.weight": "model-00002-of-00014.safetensors",
37
+ "model.layers.10.self_attn.o_proj.weight": "model-00013-of-00014.safetensors",
38
+ "model.layers.10.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
39
+ "model.layers.10.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
40
+ "model.layers.10.self_attn.v_proj.weight": "model-00008-of-00014.safetensors",
41
+ "model.layers.11.input_layernorm.weight": "model-00006-of-00014.safetensors",
42
+ "model.layers.11.mlp.down_proj.weight": "model-00008-of-00014.safetensors",
43
+ "model.layers.11.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
44
+ "model.layers.11.mlp.up_proj.weight": "model-00010-of-00014.safetensors",
45
+ "model.layers.11.post_attention_layernorm.weight": "model-00013-of-00014.safetensors",
46
+ "model.layers.11.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
47
+ "model.layers.11.self_attn.k_proj.weight": "model-00006-of-00014.safetensors",
48
+ "model.layers.11.self_attn.o_proj.weight": "model-00014-of-00014.safetensors",
49
+ "model.layers.11.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
50
+ "model.layers.11.self_attn.q_proj.weight": "model-00010-of-00014.safetensors",
51
+ "model.layers.11.self_attn.v_proj.weight": "model-00013-of-00014.safetensors",
52
+ "model.layers.12.input_layernorm.weight": "model-00013-of-00014.safetensors",
53
+ "model.layers.12.mlp.down_proj.weight": "model-00010-of-00014.safetensors",
54
+ "model.layers.12.mlp.gate_proj.weight": "model-00011-of-00014.safetensors",
55
+ "model.layers.12.mlp.up_proj.weight": "model-00007-of-00014.safetensors",
56
+ "model.layers.12.post_attention_layernorm.weight": "model-00013-of-00014.safetensors",
57
+ "model.layers.12.self_attn.k_norm.weight": "model-00001-of-00014.safetensors",
58
+ "model.layers.12.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
59
+ "model.layers.12.self_attn.o_proj.weight": "model-00009-of-00014.safetensors",
60
+ "model.layers.12.self_attn.q_norm.weight": "model-00011-of-00014.safetensors",
61
+ "model.layers.12.self_attn.q_proj.weight": "model-00013-of-00014.safetensors",
62
+ "model.layers.12.self_attn.v_proj.weight": "model-00007-of-00014.safetensors",
63
+ "model.layers.13.input_layernorm.weight": "model-00004-of-00014.safetensors",
64
+ "model.layers.13.mlp.down_proj.weight": "model-00009-of-00014.safetensors",
65
+ "model.layers.13.mlp.gate_proj.weight": "model-00005-of-00014.safetensors",
66
+ "model.layers.13.mlp.up_proj.weight": "model-00002-of-00014.safetensors",
67
+ "model.layers.13.post_attention_layernorm.weight": "model-00005-of-00014.safetensors",
68
+ "model.layers.13.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
69
+ "model.layers.13.self_attn.k_proj.weight": "model-00014-of-00014.safetensors",
70
+ "model.layers.13.self_attn.o_proj.weight": "model-00007-of-00014.safetensors",
71
+ "model.layers.13.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
72
+ "model.layers.13.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
73
+ "model.layers.13.self_attn.v_proj.weight": "model-00006-of-00014.safetensors",
74
+ "model.layers.14.input_layernorm.weight": "model-00013-of-00014.safetensors",
75
+ "model.layers.14.mlp.down_proj.weight": "model-00013-of-00014.safetensors",
76
+ "model.layers.14.mlp.gate_proj.weight": "model-00012-of-00014.safetensors",
77
+ "model.layers.14.mlp.up_proj.weight": "model-00006-of-00014.safetensors",
78
+ "model.layers.14.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
79
+ "model.layers.14.self_attn.k_norm.weight": "model-00008-of-00014.safetensors",
80
+ "model.layers.14.self_attn.k_proj.weight": "model-00012-of-00014.safetensors",
81
+ "model.layers.14.self_attn.o_proj.weight": "model-00004-of-00014.safetensors",
82
+ "model.layers.14.self_attn.q_norm.weight": "model-00008-of-00014.safetensors",
83
+ "model.layers.14.self_attn.q_proj.weight": "model-00004-of-00014.safetensors",
84
+ "model.layers.14.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
85
+ "model.layers.15.input_layernorm.weight": "model-00012-of-00014.safetensors",
86
+ "model.layers.15.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
87
+ "model.layers.15.mlp.gate_proj.weight": "model-00012-of-00014.safetensors",
88
+ "model.layers.15.mlp.up_proj.weight": "model-00007-of-00014.safetensors",
89
+ "model.layers.15.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
90
+ "model.layers.15.self_attn.k_norm.weight": "model-00006-of-00014.safetensors",
91
+ "model.layers.15.self_attn.k_proj.weight": "model-00010-of-00014.safetensors",
92
+ "model.layers.15.self_attn.o_proj.weight": "model-00001-of-00014.safetensors",
93
+ "model.layers.15.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
94
+ "model.layers.15.self_attn.q_proj.weight": "model-00010-of-00014.safetensors",
95
+ "model.layers.15.self_attn.v_proj.weight": "model-00011-of-00014.safetensors",
96
+ "model.layers.16.input_layernorm.weight": "model-00013-of-00014.safetensors",
97
+ "model.layers.16.mlp.down_proj.weight": "model-00012-of-00014.safetensors",
98
+ "model.layers.16.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
99
+ "model.layers.16.mlp.up_proj.weight": "model-00001-of-00014.safetensors",
100
+ "model.layers.16.post_attention_layernorm.weight": "model-00012-of-00014.safetensors",
101
+ "model.layers.16.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
102
+ "model.layers.16.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
103
+ "model.layers.16.self_attn.o_proj.weight": "model-00003-of-00014.safetensors",
104
+ "model.layers.16.self_attn.q_norm.weight": "model-00010-of-00014.safetensors",
105
+ "model.layers.16.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
106
+ "model.layers.16.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
107
+ "model.layers.17.input_layernorm.weight": "model-00013-of-00014.safetensors",
108
+ "model.layers.17.mlp.down_proj.weight": "model-00002-of-00014.safetensors",
109
+ "model.layers.17.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
110
+ "model.layers.17.mlp.up_proj.weight": "model-00008-of-00014.safetensors",
111
+ "model.layers.17.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
112
+ "model.layers.17.self_attn.k_norm.weight": "model-00002-of-00014.safetensors",
113
+ "model.layers.17.self_attn.k_proj.weight": "model-00011-of-00014.safetensors",
114
+ "model.layers.17.self_attn.o_proj.weight": "model-00011-of-00014.safetensors",
115
+ "model.layers.17.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
116
+ "model.layers.17.self_attn.q_proj.weight": "model-00003-of-00014.safetensors",
117
+ "model.layers.17.self_attn.v_proj.weight": "model-00003-of-00014.safetensors",
118
+ "model.layers.18.input_layernorm.weight": "model-00001-of-00014.safetensors",
119
+ "model.layers.18.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
120
+ "model.layers.18.mlp.gate_proj.weight": "model-00001-of-00014.safetensors",
121
+ "model.layers.18.mlp.up_proj.weight": "model-00001-of-00014.safetensors",
122
+ "model.layers.18.post_attention_layernorm.weight": "model-00006-of-00014.safetensors",
123
+ "model.layers.18.self_attn.k_norm.weight": "model-00006-of-00014.safetensors",
124
+ "model.layers.18.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
125
+ "model.layers.18.self_attn.o_proj.weight": "model-00013-of-00014.safetensors",
126
+ "model.layers.18.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
127
+ "model.layers.18.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
128
+ "model.layers.18.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
129
+ "model.layers.19.input_layernorm.weight": "model-00014-of-00014.safetensors",
130
+ "model.layers.19.mlp.down_proj.weight": "model-00010-of-00014.safetensors",
131
+ "model.layers.19.mlp.gate_proj.weight": "model-00010-of-00014.safetensors",
132
+ "model.layers.19.mlp.up_proj.weight": "model-00012-of-00014.safetensors",
133
+ "model.layers.19.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
134
+ "model.layers.19.self_attn.k_norm.weight": "model-00002-of-00014.safetensors",
135
+ "model.layers.19.self_attn.k_proj.weight": "model-00013-of-00014.safetensors",
136
+ "model.layers.19.self_attn.o_proj.weight": "model-00002-of-00014.safetensors",
137
+ "model.layers.19.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
138
+ "model.layers.19.self_attn.q_proj.weight": "model-00009-of-00014.safetensors",
139
+ "model.layers.19.self_attn.v_proj.weight": "model-00012-of-00014.safetensors",
140
+ "model.layers.2.input_layernorm.weight": "model-00008-of-00014.safetensors",
141
+ "model.layers.2.mlp.down_proj.weight": "model-00004-of-00014.safetensors",
142
+ "model.layers.2.mlp.gate_proj.weight": "model-00003-of-00014.safetensors",
143
+ "model.layers.2.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
144
+ "model.layers.2.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
145
+ "model.layers.2.self_attn.k_norm.weight": "model-00011-of-00014.safetensors",
146
+ "model.layers.2.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
147
+ "model.layers.2.self_attn.o_proj.weight": "model-00013-of-00014.safetensors",
148
+ "model.layers.2.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
149
+ "model.layers.2.self_attn.q_proj.weight": "model-00004-of-00014.safetensors",
150
+ "model.layers.2.self_attn.v_proj.weight": "model-00006-of-00014.safetensors",
151
+ "model.layers.20.input_layernorm.weight": "model-00006-of-00014.safetensors",
152
+ "model.layers.20.mlp.down_proj.weight": "model-00002-of-00014.safetensors",
153
+ "model.layers.20.mlp.gate_proj.weight": "model-00002-of-00014.safetensors",
154
+ "model.layers.20.mlp.up_proj.weight": "model-00011-of-00014.safetensors",
155
+ "model.layers.20.post_attention_layernorm.weight": "model-00012-of-00014.safetensors",
156
+ "model.layers.20.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
157
+ "model.layers.20.self_attn.k_proj.weight": "model-00008-of-00014.safetensors",
158
+ "model.layers.20.self_attn.o_proj.weight": "model-00007-of-00014.safetensors",
159
+ "model.layers.20.self_attn.q_norm.weight": "model-00006-of-00014.safetensors",
160
+ "model.layers.20.self_attn.q_proj.weight": "model-00013-of-00014.safetensors",
161
+ "model.layers.20.self_attn.v_proj.weight": "model-00002-of-00014.safetensors",
162
+ "model.layers.21.input_layernorm.weight": "model-00004-of-00014.safetensors",
163
+ "model.layers.21.mlp.down_proj.weight": "model-00009-of-00014.safetensors",
164
+ "model.layers.21.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
165
+ "model.layers.21.mlp.up_proj.weight": "model-00006-of-00014.safetensors",
166
+ "model.layers.21.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
167
+ "model.layers.21.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
168
+ "model.layers.21.self_attn.k_proj.weight": "model-00013-of-00014.safetensors",
169
+ "model.layers.21.self_attn.o_proj.weight": "model-00005-of-00014.safetensors",
170
+ "model.layers.21.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
171
+ "model.layers.21.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
172
+ "model.layers.21.self_attn.v_proj.weight": "model-00009-of-00014.safetensors",
173
+ "model.layers.22.input_layernorm.weight": "model-00003-of-00014.safetensors",
174
+ "model.layers.22.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
175
+ "model.layers.22.mlp.gate_proj.weight": "model-00013-of-00014.safetensors",
176
+ "model.layers.22.mlp.up_proj.weight": "model-00002-of-00014.safetensors",
177
+ "model.layers.22.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
178
+ "model.layers.22.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
179
+ "model.layers.22.self_attn.k_proj.weight": "model-00011-of-00014.safetensors",
180
+ "model.layers.22.self_attn.o_proj.weight": "model-00012-of-00014.safetensors",
181
+ "model.layers.22.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
182
+ "model.layers.22.self_attn.q_proj.weight": "model-00006-of-00014.safetensors",
183
+ "model.layers.22.self_attn.v_proj.weight": "model-00006-of-00014.safetensors",
184
+ "model.layers.23.input_layernorm.weight": "model-00009-of-00014.safetensors",
185
+ "model.layers.23.mlp.down_proj.weight": "model-00007-of-00014.safetensors",
186
+ "model.layers.23.mlp.gate_proj.weight": "model-00004-of-00014.safetensors",
187
+ "model.layers.23.mlp.up_proj.weight": "model-00006-of-00014.safetensors",
188
+ "model.layers.23.post_attention_layernorm.weight": "model-00011-of-00014.safetensors",
189
+ "model.layers.23.self_attn.k_norm.weight": "model-00001-of-00014.safetensors",
190
+ "model.layers.23.self_attn.k_proj.weight": "model-00008-of-00014.safetensors",
191
+ "model.layers.23.self_attn.o_proj.weight": "model-00003-of-00014.safetensors",
192
+ "model.layers.23.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
193
+ "model.layers.23.self_attn.q_proj.weight": "model-00013-of-00014.safetensors",
194
+ "model.layers.23.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
195
+ "model.layers.24.input_layernorm.weight": "model-00013-of-00014.safetensors",
196
+ "model.layers.24.mlp.down_proj.weight": "model-00012-of-00014.safetensors",
197
+ "model.layers.24.mlp.gate_proj.weight": "model-00002-of-00014.safetensors",
198
+ "model.layers.24.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
199
+ "model.layers.24.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
200
+ "model.layers.24.self_attn.k_norm.weight": "model-00012-of-00014.safetensors",
201
+ "model.layers.24.self_attn.k_proj.weight": "model-00011-of-00014.safetensors",
202
+ "model.layers.24.self_attn.o_proj.weight": "model-00013-of-00014.safetensors",
203
+ "model.layers.24.self_attn.q_norm.weight": "model-00004-of-00014.safetensors",
204
+ "model.layers.24.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
205
+ "model.layers.24.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
206
+ "model.layers.25.input_layernorm.weight": "model-00008-of-00014.safetensors",
207
+ "model.layers.25.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
208
+ "model.layers.25.mlp.gate_proj.weight": "model-00001-of-00014.safetensors",
209
+ "model.layers.25.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
210
+ "model.layers.25.post_attention_layernorm.weight": "model-00011-of-00014.safetensors",
211
+ "model.layers.25.self_attn.k_norm.weight": "model-00002-of-00014.safetensors",
212
+ "model.layers.25.self_attn.k_proj.weight": "model-00002-of-00014.safetensors",
213
+ "model.layers.25.self_attn.o_proj.weight": "model-00006-of-00014.safetensors",
214
+ "model.layers.25.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
215
+ "model.layers.25.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
216
+ "model.layers.25.self_attn.v_proj.weight": "model-00007-of-00014.safetensors",
217
+ "model.layers.26.input_layernorm.weight": "model-00014-of-00014.safetensors",
218
+ "model.layers.26.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
219
+ "model.layers.26.mlp.gate_proj.weight": "model-00010-of-00014.safetensors",
220
+ "model.layers.26.mlp.up_proj.weight": "model-00009-of-00014.safetensors",
221
+ "model.layers.26.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
222
+ "model.layers.26.self_attn.k_norm.weight": "model-00012-of-00014.safetensors",
223
+ "model.layers.26.self_attn.k_proj.weight": "model-00004-of-00014.safetensors",
224
+ "model.layers.26.self_attn.o_proj.weight": "model-00005-of-00014.safetensors",
225
+ "model.layers.26.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
226
+ "model.layers.26.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
227
+ "model.layers.26.self_attn.v_proj.weight": "model-00012-of-00014.safetensors",
228
+ "model.layers.27.input_layernorm.weight": "model-00009-of-00014.safetensors",
229
+ "model.layers.27.mlp.down_proj.weight": "model-00011-of-00014.safetensors",
230
+ "model.layers.27.mlp.gate_proj.weight": "model-00005-of-00014.safetensors",
231
+ "model.layers.27.mlp.up_proj.weight": "model-00011-of-00014.safetensors",
232
+ "model.layers.27.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
233
+ "model.layers.27.self_attn.k_norm.weight": "model-00006-of-00014.safetensors",
234
+ "model.layers.27.self_attn.k_proj.weight": "model-00002-of-00014.safetensors",
235
+ "model.layers.27.self_attn.o_proj.weight": "model-00014-of-00014.safetensors",
236
+ "model.layers.27.self_attn.q_norm.weight": "model-00008-of-00014.safetensors",
237
+ "model.layers.27.self_attn.q_proj.weight": "model-00006-of-00014.safetensors",
238
+ "model.layers.27.self_attn.v_proj.weight": "model-00012-of-00014.safetensors",
239
+ "model.layers.28.input_layernorm.weight": "model-00004-of-00014.safetensors",
240
+ "model.layers.28.mlp.down_proj.weight": "model-00004-of-00014.safetensors",
241
+ "model.layers.28.mlp.gate_proj.weight": "model-00008-of-00014.safetensors",
242
+ "model.layers.28.mlp.up_proj.weight": "model-00014-of-00014.safetensors",
243
+ "model.layers.28.post_attention_layernorm.weight": "model-00008-of-00014.safetensors",
244
+ "model.layers.28.self_attn.k_norm.weight": "model-00011-of-00014.safetensors",
245
+ "model.layers.28.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
246
+ "model.layers.28.self_attn.o_proj.weight": "model-00007-of-00014.safetensors",
247
+ "model.layers.28.self_attn.q_norm.weight": "model-00009-of-00014.safetensors",
248
+ "model.layers.28.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
249
+ "model.layers.28.self_attn.v_proj.weight": "model-00009-of-00014.safetensors",
250
+ "model.layers.29.input_layernorm.weight": "model-00013-of-00014.safetensors",
251
+ "model.layers.29.mlp.down_proj.weight": "model-00009-of-00014.safetensors",
252
+ "model.layers.29.mlp.gate_proj.weight": "model-00012-of-00014.safetensors",
253
+ "model.layers.29.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
254
+ "model.layers.29.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
255
+ "model.layers.29.self_attn.k_norm.weight": "model-00012-of-00014.safetensors",
256
+ "model.layers.29.self_attn.k_proj.weight": "model-00008-of-00014.safetensors",
257
+ "model.layers.29.self_attn.o_proj.weight": "model-00005-of-00014.safetensors",
258
+ "model.layers.29.self_attn.q_norm.weight": "model-00010-of-00014.safetensors",
259
+ "model.layers.29.self_attn.q_proj.weight": "model-00013-of-00014.safetensors",
260
+ "model.layers.29.self_attn.v_proj.weight": "model-00011-of-00014.safetensors",
261
+ "model.layers.3.input_layernorm.weight": "model-00011-of-00014.safetensors",
262
+ "model.layers.3.mlp.down_proj.weight": "model-00004-of-00014.safetensors",
263
+ "model.layers.3.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
264
+ "model.layers.3.mlp.up_proj.weight": "model-00006-of-00014.safetensors",
265
+ "model.layers.3.post_attention_layernorm.weight": "model-00013-of-00014.safetensors",
266
+ "model.layers.3.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
267
+ "model.layers.3.self_attn.k_proj.weight": "model-00012-of-00014.safetensors",
268
+ "model.layers.3.self_attn.o_proj.weight": "model-00011-of-00014.safetensors",
269
+ "model.layers.3.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
270
+ "model.layers.3.self_attn.q_proj.weight": "model-00013-of-00014.safetensors",
271
+ "model.layers.3.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
272
+ "model.layers.30.input_layernorm.weight": "model-00013-of-00014.safetensors",
273
+ "model.layers.30.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
274
+ "model.layers.30.mlp.gate_proj.weight": "model-00012-of-00014.safetensors",
275
+ "model.layers.30.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
276
+ "model.layers.30.post_attention_layernorm.weight": "model-00008-of-00014.safetensors",
277
+ "model.layers.30.self_attn.k_norm.weight": "model-00003-of-00014.safetensors",
278
+ "model.layers.30.self_attn.k_proj.weight": "model-00001-of-00014.safetensors",
279
+ "model.layers.30.self_attn.o_proj.weight": "model-00012-of-00014.safetensors",
280
+ "model.layers.30.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
281
+ "model.layers.30.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
282
+ "model.layers.30.self_attn.v_proj.weight": "model-00001-of-00014.safetensors",
283
+ "model.layers.31.input_layernorm.weight": "model-00004-of-00014.safetensors",
284
+ "model.layers.31.mlp.down_proj.weight": "model-00001-of-00014.safetensors",
285
+ "model.layers.31.mlp.gate_proj.weight": "model-00004-of-00014.safetensors",
286
+ "model.layers.31.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
287
+ "model.layers.31.post_attention_layernorm.weight": "model-00008-of-00014.safetensors",
288
+ "model.layers.31.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
289
+ "model.layers.31.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
290
+ "model.layers.31.self_attn.o_proj.weight": "model-00005-of-00014.safetensors",
291
+ "model.layers.31.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
292
+ "model.layers.31.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
293
+ "model.layers.31.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
294
+ "model.layers.32.input_layernorm.weight": "model-00013-of-00014.safetensors",
295
+ "model.layers.32.mlp.down_proj.weight": "model-00013-of-00014.safetensors",
296
+ "model.layers.32.mlp.gate_proj.weight": "model-00001-of-00014.safetensors",
297
+ "model.layers.32.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
298
+ "model.layers.32.post_attention_layernorm.weight": "model-00010-of-00014.safetensors",
299
+ "model.layers.32.self_attn.k_norm.weight": "model-00001-of-00014.safetensors",
300
+ "model.layers.32.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
301
+ "model.layers.32.self_attn.o_proj.weight": "model-00007-of-00014.safetensors",
302
+ "model.layers.32.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
303
+ "model.layers.32.self_attn.q_proj.weight": "model-00012-of-00014.safetensors",
304
+ "model.layers.32.self_attn.v_proj.weight": "model-00001-of-00014.safetensors",
305
+ "model.layers.33.input_layernorm.weight": "model-00006-of-00014.safetensors",
306
+ "model.layers.33.mlp.down_proj.weight": "model-00014-of-00014.safetensors",
307
+ "model.layers.33.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
308
+ "model.layers.33.mlp.up_proj.weight": "model-00005-of-00014.safetensors",
309
+ "model.layers.33.post_attention_layernorm.weight": "model-00005-of-00014.safetensors",
310
+ "model.layers.33.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
311
+ "model.layers.33.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
312
+ "model.layers.33.self_attn.o_proj.weight": "model-00005-of-00014.safetensors",
313
+ "model.layers.33.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
314
+ "model.layers.33.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
315
+ "model.layers.33.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
316
+ "model.layers.34.input_layernorm.weight": "model-00005-of-00014.safetensors",
317
+ "model.layers.34.mlp.down_proj.weight": "model-00013-of-00014.safetensors",
318
+ "model.layers.34.mlp.gate_proj.weight": "model-00009-of-00014.safetensors",
319
+ "model.layers.34.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
320
+ "model.layers.34.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
321
+ "model.layers.34.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
322
+ "model.layers.34.self_attn.k_proj.weight": "model-00008-of-00014.safetensors",
323
+ "model.layers.34.self_attn.o_proj.weight": "model-00007-of-00014.safetensors",
324
+ "model.layers.34.self_attn.q_norm.weight": "model-00008-of-00014.safetensors",
325
+ "model.layers.34.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
326
+ "model.layers.34.self_attn.v_proj.weight": "model-00014-of-00014.safetensors",
327
+ "model.layers.35.input_layernorm.weight": "model-00003-of-00014.safetensors",
328
+ "model.layers.35.mlp.down_proj.weight": "model-00004-of-00014.safetensors",
329
+ "model.layers.35.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
330
+ "model.layers.35.mlp.up_proj.weight": "model-00012-of-00014.safetensors",
331
+ "model.layers.35.post_attention_layernorm.weight": "model-00012-of-00014.safetensors",
332
+ "model.layers.35.self_attn.k_norm.weight": "model-00004-of-00014.safetensors",
333
+ "model.layers.35.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
334
+ "model.layers.35.self_attn.o_proj.weight": "model-00008-of-00014.safetensors",
335
+ "model.layers.35.self_attn.q_norm.weight": "model-00008-of-00014.safetensors",
336
+ "model.layers.35.self_attn.q_proj.weight": "model-00010-of-00014.safetensors",
337
+ "model.layers.35.self_attn.v_proj.weight": "model-00008-of-00014.safetensors",
338
+ "model.layers.36.input_layernorm.weight": "model-00007-of-00014.safetensors",
339
+ "model.layers.36.mlp.down_proj.weight": "model-00002-of-00014.safetensors",
340
+ "model.layers.36.mlp.gate_proj.weight": "model-00013-of-00014.safetensors",
341
+ "model.layers.36.mlp.up_proj.weight": "model-00008-of-00014.safetensors",
342
+ "model.layers.36.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
343
+ "model.layers.36.self_attn.k_norm.weight": "model-00001-of-00014.safetensors",
344
+ "model.layers.36.self_attn.k_proj.weight": "model-00002-of-00014.safetensors",
345
+ "model.layers.36.self_attn.o_proj.weight": "model-00010-of-00014.safetensors",
346
+ "model.layers.36.self_attn.q_norm.weight": "model-00014-of-00014.safetensors",
347
+ "model.layers.36.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
348
+ "model.layers.36.self_attn.v_proj.weight": "model-00006-of-00014.safetensors",
349
+ "model.layers.37.input_layernorm.weight": "model-00013-of-00014.safetensors",
350
+ "model.layers.37.mlp.down_proj.weight": "model-00001-of-00014.safetensors",
351
+ "model.layers.37.mlp.gate_proj.weight": "model-00008-of-00014.safetensors",
352
+ "model.layers.37.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
353
+ "model.layers.37.post_attention_layernorm.weight": "model-00011-of-00014.safetensors",
354
+ "model.layers.37.self_attn.k_norm.weight": "model-00001-of-00014.safetensors",
355
+ "model.layers.37.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
356
+ "model.layers.37.self_attn.o_proj.weight": "model-00012-of-00014.safetensors",
357
+ "model.layers.37.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
358
+ "model.layers.37.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
359
+ "model.layers.37.self_attn.v_proj.weight": "model-00011-of-00014.safetensors",
360
+ "model.layers.38.input_layernorm.weight": "model-00001-of-00014.safetensors",
361
+ "model.layers.38.mlp.down_proj.weight": "model-00012-of-00014.safetensors",
362
+ "model.layers.38.mlp.gate_proj.weight": "model-00011-of-00014.safetensors",
363
+ "model.layers.38.mlp.up_proj.weight": "model-00002-of-00014.safetensors",
364
+ "model.layers.38.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
365
+ "model.layers.38.self_attn.k_norm.weight": "model-00002-of-00014.safetensors",
366
+ "model.layers.38.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
367
+ "model.layers.38.self_attn.o_proj.weight": "model-00008-of-00014.safetensors",
368
+ "model.layers.38.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
369
+ "model.layers.38.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
370
+ "model.layers.38.self_attn.v_proj.weight": "model-00007-of-00014.safetensors",
371
+ "model.layers.39.input_layernorm.weight": "model-00003-of-00014.safetensors",
372
+ "model.layers.39.mlp.down_proj.weight": "model-00012-of-00014.safetensors",
373
+ "model.layers.39.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
374
+ "model.layers.39.mlp.up_proj.weight": "model-00008-of-00014.safetensors",
375
+ "model.layers.39.post_attention_layernorm.weight": "model-00008-of-00014.safetensors",
376
+ "model.layers.39.self_attn.k_norm.weight": "model-00003-of-00014.safetensors",
377
+ "model.layers.39.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
378
+ "model.layers.39.self_attn.o_proj.weight": "model-00011-of-00014.safetensors",
379
+ "model.layers.39.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
380
+ "model.layers.39.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
381
+ "model.layers.39.self_attn.v_proj.weight": "model-00003-of-00014.safetensors",
382
+ "model.layers.4.input_layernorm.weight": "model-00014-of-00014.safetensors",
383
+ "model.layers.4.mlp.down_proj.weight": "model-00001-of-00014.safetensors",
384
+ "model.layers.4.mlp.gate_proj.weight": "model-00010-of-00014.safetensors",
385
+ "model.layers.4.mlp.up_proj.weight": "model-00008-of-00014.safetensors",
386
+ "model.layers.4.post_attention_layernorm.weight": "model-00001-of-00014.safetensors",
387
+ "model.layers.4.self_attn.k_norm.weight": "model-00012-of-00014.safetensors",
388
+ "model.layers.4.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
389
+ "model.layers.4.self_attn.o_proj.weight": "model-00006-of-00014.safetensors",
390
+ "model.layers.4.self_attn.q_norm.weight": "model-00014-of-00014.safetensors",
391
+ "model.layers.4.self_attn.q_proj.weight": "model-00003-of-00014.safetensors",
392
+ "model.layers.4.self_attn.v_proj.weight": "model-00014-of-00014.safetensors",
393
+ "model.layers.40.input_layernorm.weight": "model-00007-of-00014.safetensors",
394
+ "model.layers.40.mlp.down_proj.weight": "model-00006-of-00014.safetensors",
395
+ "model.layers.40.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
396
+ "model.layers.40.mlp.up_proj.weight": "model-00014-of-00014.safetensors",
397
+ "model.layers.40.post_attention_layernorm.weight": "model-00005-of-00014.safetensors",
398
+ "model.layers.40.self_attn.k_norm.weight": "model-00008-of-00014.safetensors",
399
+ "model.layers.40.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
400
+ "model.layers.40.self_attn.o_proj.weight": "model-00002-of-00014.safetensors",
401
+ "model.layers.40.self_attn.q_norm.weight": "model-00012-of-00014.safetensors",
402
+ "model.layers.40.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
403
+ "model.layers.40.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
404
+ "model.layers.41.input_layernorm.weight": "model-00004-of-00014.safetensors",
405
+ "model.layers.41.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
406
+ "model.layers.41.mlp.gate_proj.weight": "model-00014-of-00014.safetensors",
407
+ "model.layers.41.mlp.up_proj.weight": "model-00002-of-00014.safetensors",
408
+ "model.layers.41.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
409
+ "model.layers.41.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
410
+ "model.layers.41.self_attn.k_proj.weight": "model-00001-of-00014.safetensors",
411
+ "model.layers.41.self_attn.o_proj.weight": "model-00006-of-00014.safetensors",
412
+ "model.layers.41.self_attn.q_norm.weight": "model-00011-of-00014.safetensors",
413
+ "model.layers.41.self_attn.q_proj.weight": "model-00007-of-00014.safetensors",
414
+ "model.layers.41.self_attn.v_proj.weight": "model-00007-of-00014.safetensors",
415
+ "model.layers.42.input_layernorm.weight": "model-00008-of-00014.safetensors",
416
+ "model.layers.42.mlp.down_proj.weight": "model-00006-of-00014.safetensors",
417
+ "model.layers.42.mlp.gate_proj.weight": "model-00005-of-00014.safetensors",
418
+ "model.layers.42.mlp.up_proj.weight": "model-00011-of-00014.safetensors",
419
+ "model.layers.42.post_attention_layernorm.weight": "model-00006-of-00014.safetensors",
420
+ "model.layers.42.self_attn.k_norm.weight": "model-00006-of-00014.safetensors",
421
+ "model.layers.42.self_attn.k_proj.weight": "model-00008-of-00014.safetensors",
422
+ "model.layers.42.self_attn.o_proj.weight": "model-00006-of-00014.safetensors",
423
+ "model.layers.42.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
424
+ "model.layers.42.self_attn.q_proj.weight": "model-00012-of-00014.safetensors",
425
+ "model.layers.42.self_attn.v_proj.weight": "model-00007-of-00014.safetensors",
426
+ "model.layers.43.input_layernorm.weight": "model-00011-of-00014.safetensors",
427
+ "model.layers.43.mlp.down_proj.weight": "model-00009-of-00014.safetensors",
428
+ "model.layers.43.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
429
+ "model.layers.43.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
430
+ "model.layers.43.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
431
+ "model.layers.43.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
432
+ "model.layers.43.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
433
+ "model.layers.43.self_attn.o_proj.weight": "model-00009-of-00014.safetensors",
434
+ "model.layers.43.self_attn.q_norm.weight": "model-00009-of-00014.safetensors",
435
+ "model.layers.43.self_attn.q_proj.weight": "model-00006-of-00014.safetensors",
436
+ "model.layers.43.self_attn.v_proj.weight": "model-00014-of-00014.safetensors",
437
+ "model.layers.44.input_layernorm.weight": "model-00010-of-00014.safetensors",
438
+ "model.layers.44.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
439
+ "model.layers.44.mlp.gate_proj.weight": "model-00014-of-00014.safetensors",
440
+ "model.layers.44.mlp.up_proj.weight": "model-00011-of-00014.safetensors",
441
+ "model.layers.44.post_attention_layernorm.weight": "model-00011-of-00014.safetensors",
442
+ "model.layers.44.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
443
+ "model.layers.44.self_attn.k_proj.weight": "model-00007-of-00014.safetensors",
444
+ "model.layers.44.self_attn.o_proj.weight": "model-00012-of-00014.safetensors",
445
+ "model.layers.44.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
446
+ "model.layers.44.self_attn.q_proj.weight": "model-00003-of-00014.safetensors",
447
+ "model.layers.44.self_attn.v_proj.weight": "model-00011-of-00014.safetensors",
448
+ "model.layers.45.input_layernorm.weight": "model-00008-of-00014.safetensors",
449
+ "model.layers.45.mlp.down_proj.weight": "model-00014-of-00014.safetensors",
450
+ "model.layers.45.mlp.gate_proj.weight": "model-00013-of-00014.safetensors",
451
+ "model.layers.45.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
452
+ "model.layers.45.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
453
+ "model.layers.45.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
454
+ "model.layers.45.self_attn.k_proj.weight": "model-00006-of-00014.safetensors",
455
+ "model.layers.45.self_attn.o_proj.weight": "model-00011-of-00014.safetensors",
456
+ "model.layers.45.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
457
+ "model.layers.45.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
458
+ "model.layers.45.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
459
+ "model.layers.46.input_layernorm.weight": "model-00013-of-00014.safetensors",
460
+ "model.layers.46.mlp.down_proj.weight": "model-00011-of-00014.safetensors",
461
+ "model.layers.46.mlp.gate_proj.weight": "model-00004-of-00014.safetensors",
462
+ "model.layers.46.mlp.up_proj.weight": "model-00002-of-00014.safetensors",
463
+ "model.layers.46.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
464
+ "model.layers.46.self_attn.k_norm.weight": "model-00003-of-00014.safetensors",
465
+ "model.layers.46.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
466
+ "model.layers.46.self_attn.o_proj.weight": "model-00003-of-00014.safetensors",
467
+ "model.layers.46.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
468
+ "model.layers.46.self_attn.q_proj.weight": "model-00009-of-00014.safetensors",
469
+ "model.layers.46.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
470
+ "model.layers.47.input_layernorm.weight": "model-00007-of-00014.safetensors",
471
+ "model.layers.47.mlp.down_proj.weight": "model-00009-of-00014.safetensors",
472
+ "model.layers.47.mlp.gate_proj.weight": "model-00014-of-00014.safetensors",
473
+ "model.layers.47.mlp.up_proj.weight": "model-00010-of-00014.safetensors",
474
+ "model.layers.47.post_attention_layernorm.weight": "model-00013-of-00014.safetensors",
475
+ "model.layers.47.self_attn.k_norm.weight": "model-00008-of-00014.safetensors",
476
+ "model.layers.47.self_attn.k_proj.weight": "model-00012-of-00014.safetensors",
477
+ "model.layers.47.self_attn.o_proj.weight": "model-00003-of-00014.safetensors",
478
+ "model.layers.47.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
479
+ "model.layers.47.self_attn.q_proj.weight": "model-00002-of-00014.safetensors",
480
+ "model.layers.47.self_attn.v_proj.weight": "model-00013-of-00014.safetensors",
481
+ "model.layers.48.input_layernorm.weight": "model-00003-of-00014.safetensors",
482
+ "model.layers.48.mlp.down_proj.weight": "model-00012-of-00014.safetensors",
483
+ "model.layers.48.mlp.gate_proj.weight": "model-00005-of-00014.safetensors",
484
+ "model.layers.48.mlp.up_proj.weight": "model-00005-of-00014.safetensors",
485
+ "model.layers.48.post_attention_layernorm.weight": "model-00006-of-00014.safetensors",
486
+ "model.layers.48.self_attn.k_norm.weight": "model-00012-of-00014.safetensors",
487
+ "model.layers.48.self_attn.k_proj.weight": "model-00014-of-00014.safetensors",
488
+ "model.layers.48.self_attn.o_proj.weight": "model-00001-of-00014.safetensors",
489
+ "model.layers.48.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
490
+ "model.layers.48.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
491
+ "model.layers.48.self_attn.v_proj.weight": "model-00013-of-00014.safetensors",
492
+ "model.layers.49.input_layernorm.weight": "model-00012-of-00014.safetensors",
493
+ "model.layers.49.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
494
+ "model.layers.49.mlp.gate_proj.weight": "model-00014-of-00014.safetensors",
495
+ "model.layers.49.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
496
+ "model.layers.49.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
497
+ "model.layers.49.self_attn.k_norm.weight": "model-00008-of-00014.safetensors",
498
+ "model.layers.49.self_attn.k_proj.weight": "model-00013-of-00014.safetensors",
499
+ "model.layers.49.self_attn.o_proj.weight": "model-00005-of-00014.safetensors",
500
+ "model.layers.49.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
501
+ "model.layers.49.self_attn.q_proj.weight": "model-00008-of-00014.safetensors",
502
+ "model.layers.49.self_attn.v_proj.weight": "model-00013-of-00014.safetensors",
503
+ "model.layers.5.input_layernorm.weight": "model-00001-of-00014.safetensors",
504
+ "model.layers.5.mlp.down_proj.weight": "model-00010-of-00014.safetensors",
505
+ "model.layers.5.mlp.gate_proj.weight": "model-00008-of-00014.safetensors",
506
+ "model.layers.5.mlp.up_proj.weight": "model-00007-of-00014.safetensors",
507
+ "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
508
+ "model.layers.5.self_attn.k_norm.weight": "model-00008-of-00014.safetensors",
509
+ "model.layers.5.self_attn.k_proj.weight": "model-00014-of-00014.safetensors",
510
+ "model.layers.5.self_attn.o_proj.weight": "model-00008-of-00014.safetensors",
511
+ "model.layers.5.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
512
+ "model.layers.5.self_attn.q_proj.weight": "model-00004-of-00014.safetensors",
513
+ "model.layers.5.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
514
+ "model.layers.50.input_layernorm.weight": "model-00012-of-00014.safetensors",
515
+ "model.layers.50.mlp.down_proj.weight": "model-00008-of-00014.safetensors",
516
+ "model.layers.50.mlp.gate_proj.weight": "model-00010-of-00014.safetensors",
517
+ "model.layers.50.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
518
+ "model.layers.50.post_attention_layernorm.weight": "model-00001-of-00014.safetensors",
519
+ "model.layers.50.self_attn.k_norm.weight": "model-00010-of-00014.safetensors",
520
+ "model.layers.50.self_attn.k_proj.weight": "model-00001-of-00014.safetensors",
521
+ "model.layers.50.self_attn.o_proj.weight": "model-00012-of-00014.safetensors",
522
+ "model.layers.50.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
523
+ "model.layers.50.self_attn.q_proj.weight": "model-00014-of-00014.safetensors",
524
+ "model.layers.50.self_attn.v_proj.weight": "model-00005-of-00014.safetensors",
525
+ "model.layers.51.input_layernorm.weight": "model-00013-of-00014.safetensors",
526
+ "model.layers.51.mlp.down_proj.weight": "model-00007-of-00014.safetensors",
527
+ "model.layers.51.mlp.gate_proj.weight": "model-00001-of-00014.safetensors",
528
+ "model.layers.51.mlp.up_proj.weight": "model-00010-of-00014.safetensors",
529
+ "model.layers.51.post_attention_layernorm.weight": "model-00012-of-00014.safetensors",
530
+ "model.layers.51.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
531
+ "model.layers.51.self_attn.k_proj.weight": "model-00012-of-00014.safetensors",
532
+ "model.layers.51.self_attn.o_proj.weight": "model-00002-of-00014.safetensors",
533
+ "model.layers.51.self_attn.q_norm.weight": "model-00012-of-00014.safetensors",
534
+ "model.layers.51.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
535
+ "model.layers.51.self_attn.v_proj.weight": "model-00010-of-00014.safetensors",
536
+ "model.layers.52.input_layernorm.weight": "model-00013-of-00014.safetensors",
537
+ "model.layers.52.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
538
+ "model.layers.52.mlp.gate_proj.weight": "model-00011-of-00014.safetensors",
539
+ "model.layers.52.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
540
+ "model.layers.52.post_attention_layernorm.weight": "model-00001-of-00014.safetensors",
541
+ "model.layers.52.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
542
+ "model.layers.52.self_attn.k_proj.weight": "model-00004-of-00014.safetensors",
543
+ "model.layers.52.self_attn.o_proj.weight": "model-00010-of-00014.safetensors",
544
+ "model.layers.52.self_attn.q_norm.weight": "model-00005-of-00014.safetensors",
545
+ "model.layers.52.self_attn.q_proj.weight": "model-00009-of-00014.safetensors",
546
+ "model.layers.52.self_attn.v_proj.weight": "model-00003-of-00014.safetensors",
547
+ "model.layers.53.input_layernorm.weight": "model-00007-of-00014.safetensors",
548
+ "model.layers.53.mlp.down_proj.weight": "model-00001-of-00014.safetensors",
549
+ "model.layers.53.mlp.gate_proj.weight": "model-00002-of-00014.safetensors",
550
+ "model.layers.53.mlp.up_proj.weight": "model-00014-of-00014.safetensors",
551
+ "model.layers.53.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
552
+ "model.layers.53.self_attn.k_norm.weight": "model-00007-of-00014.safetensors",
553
+ "model.layers.53.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
554
+ "model.layers.53.self_attn.o_proj.weight": "model-00009-of-00014.safetensors",
555
+ "model.layers.53.self_attn.q_norm.weight": "model-00014-of-00014.safetensors",
556
+ "model.layers.53.self_attn.q_proj.weight": "model-00003-of-00014.safetensors",
557
+ "model.layers.53.self_attn.v_proj.weight": "model-00003-of-00014.safetensors",
558
+ "model.layers.54.input_layernorm.weight": "model-00003-of-00014.safetensors",
559
+ "model.layers.54.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
560
+ "model.layers.54.mlp.gate_proj.weight": "model-00002-of-00014.safetensors",
561
+ "model.layers.54.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
562
+ "model.layers.54.post_attention_layernorm.weight": "model-00001-of-00014.safetensors",
563
+ "model.layers.54.self_attn.k_norm.weight": "model-00011-of-00014.safetensors",
564
+ "model.layers.54.self_attn.k_proj.weight": "model-00002-of-00014.safetensors",
565
+ "model.layers.54.self_attn.o_proj.weight": "model-00001-of-00014.safetensors",
566
+ "model.layers.54.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
567
+ "model.layers.54.self_attn.q_proj.weight": "model-00004-of-00014.safetensors",
568
+ "model.layers.54.self_attn.v_proj.weight": "model-00009-of-00014.safetensors",
569
+ "model.layers.55.input_layernorm.weight": "model-00005-of-00014.safetensors",
570
+ "model.layers.55.mlp.down_proj.weight": "model-00002-of-00014.safetensors",
571
+ "model.layers.55.mlp.gate_proj.weight": "model-00007-of-00014.safetensors",
572
+ "model.layers.55.mlp.up_proj.weight": "model-00012-of-00014.safetensors",
573
+ "model.layers.55.post_attention_layernorm.weight": "model-00008-of-00014.safetensors",
574
+ "model.layers.55.self_attn.k_norm.weight": "model-00007-of-00014.safetensors",
575
+ "model.layers.55.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
576
+ "model.layers.55.self_attn.o_proj.weight": "model-00008-of-00014.safetensors",
577
+ "model.layers.55.self_attn.q_norm.weight": "model-00012-of-00014.safetensors",
578
+ "model.layers.55.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
579
+ "model.layers.55.self_attn.v_proj.weight": "model-00003-of-00014.safetensors",
580
+ "model.layers.56.input_layernorm.weight": "model-00003-of-00014.safetensors",
581
+ "model.layers.56.mlp.down_proj.weight": "model-00004-of-00014.safetensors",
582
+ "model.layers.56.mlp.gate_proj.weight": "model-00008-of-00014.safetensors",
583
+ "model.layers.56.mlp.up_proj.weight": "model-00006-of-00014.safetensors",
584
+ "model.layers.56.post_attention_layernorm.weight": "model-00005-of-00014.safetensors",
585
+ "model.layers.56.self_attn.k_norm.weight": "model-00011-of-00014.safetensors",
586
+ "model.layers.56.self_attn.k_proj.weight": "model-00013-of-00014.safetensors",
587
+ "model.layers.56.self_attn.o_proj.weight": "model-00011-of-00014.safetensors",
588
+ "model.layers.56.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
589
+ "model.layers.56.self_attn.q_proj.weight": "model-00006-of-00014.safetensors",
590
+ "model.layers.56.self_attn.v_proj.weight": "model-00007-of-00014.safetensors",
591
+ "model.layers.57.input_layernorm.weight": "model-00005-of-00014.safetensors",
592
+ "model.layers.57.mlp.down_proj.weight": "model-00014-of-00014.safetensors",
593
+ "model.layers.57.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
594
+ "model.layers.57.mlp.up_proj.weight": "model-00011-of-00014.safetensors",
595
+ "model.layers.57.post_attention_layernorm.weight": "model-00001-of-00014.safetensors",
596
+ "model.layers.57.self_attn.k_norm.weight": "model-00001-of-00014.safetensors",
597
+ "model.layers.57.self_attn.k_proj.weight": "model-00012-of-00014.safetensors",
598
+ "model.layers.57.self_attn.o_proj.weight": "model-00001-of-00014.safetensors",
599
+ "model.layers.57.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
600
+ "model.layers.57.self_attn.q_proj.weight": "model-00007-of-00014.safetensors",
601
+ "model.layers.57.self_attn.v_proj.weight": "model-00012-of-00014.safetensors",
602
+ "model.layers.58.input_layernorm.weight": "model-00011-of-00014.safetensors",
603
+ "model.layers.58.mlp.down_proj.weight": "model-00006-of-00014.safetensors",
604
+ "model.layers.58.mlp.gate_proj.weight": "model-00004-of-00014.safetensors",
605
+ "model.layers.58.mlp.up_proj.weight": "model-00007-of-00014.safetensors",
606
+ "model.layers.58.post_attention_layernorm.weight": "model-00009-of-00014.safetensors",
607
+ "model.layers.58.self_attn.k_norm.weight": "model-00004-of-00014.safetensors",
608
+ "model.layers.58.self_attn.k_proj.weight": "model-00009-of-00014.safetensors",
609
+ "model.layers.58.self_attn.o_proj.weight": "model-00002-of-00014.safetensors",
610
+ "model.layers.58.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
611
+ "model.layers.58.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
612
+ "model.layers.58.self_attn.v_proj.weight": "model-00009-of-00014.safetensors",
613
+ "model.layers.59.input_layernorm.weight": "model-00013-of-00014.safetensors",
614
+ "model.layers.59.mlp.down_proj.weight": "model-00003-of-00014.safetensors",
615
+ "model.layers.59.mlp.gate_proj.weight": "model-00008-of-00014.safetensors",
616
+ "model.layers.59.mlp.up_proj.weight": "model-00007-of-00014.safetensors",
617
+ "model.layers.59.post_attention_layernorm.weight": "model-00004-of-00014.safetensors",
618
+ "model.layers.59.self_attn.k_norm.weight": "model-00010-of-00014.safetensors",
619
+ "model.layers.59.self_attn.k_proj.weight": "model-00011-of-00014.safetensors",
620
+ "model.layers.59.self_attn.o_proj.weight": "model-00009-of-00014.safetensors",
621
+ "model.layers.59.self_attn.q_norm.weight": "model-00007-of-00014.safetensors",
622
+ "model.layers.59.self_attn.q_proj.weight": "model-00012-of-00014.safetensors",
623
+ "model.layers.59.self_attn.v_proj.weight": "model-00009-of-00014.safetensors",
624
+ "model.layers.6.input_layernorm.weight": "model-00001-of-00014.safetensors",
625
+ "model.layers.6.mlp.down_proj.weight": "model-00011-of-00014.safetensors",
626
+ "model.layers.6.mlp.gate_proj.weight": "model-00009-of-00014.safetensors",
627
+ "model.layers.6.mlp.up_proj.weight": "model-00012-of-00014.safetensors",
628
+ "model.layers.6.post_attention_layernorm.weight": "model-00007-of-00014.safetensors",
629
+ "model.layers.6.self_attn.k_norm.weight": "model-00002-of-00014.safetensors",
630
+ "model.layers.6.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
631
+ "model.layers.6.self_attn.o_proj.weight": "model-00003-of-00014.safetensors",
632
+ "model.layers.6.self_attn.q_norm.weight": "model-00008-of-00014.safetensors",
633
+ "model.layers.6.self_attn.q_proj.weight": "model-00001-of-00014.safetensors",
634
+ "model.layers.6.self_attn.v_proj.weight": "model-00014-of-00014.safetensors",
635
+ "model.layers.60.input_layernorm.weight": "model-00007-of-00014.safetensors",
636
+ "model.layers.60.mlp.down_proj.weight": "model-00002-of-00014.safetensors",
637
+ "model.layers.60.mlp.gate_proj.weight": "model-00013-of-00014.safetensors",
638
+ "model.layers.60.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
639
+ "model.layers.60.post_attention_layernorm.weight": "model-00003-of-00014.safetensors",
640
+ "model.layers.60.self_attn.k_norm.weight": "model-00007-of-00014.safetensors",
641
+ "model.layers.60.self_attn.k_proj.weight": "model-00012-of-00014.safetensors",
642
+ "model.layers.60.self_attn.o_proj.weight": "model-00009-of-00014.safetensors",
643
+ "model.layers.60.self_attn.q_norm.weight": "model-00003-of-00014.safetensors",
644
+ "model.layers.60.self_attn.q_proj.weight": "model-00009-of-00014.safetensors",
645
+ "model.layers.60.self_attn.v_proj.weight": "model-00002-of-00014.safetensors",
646
+ "model.layers.61.input_layernorm.weight": "model-00002-of-00014.safetensors",
647
+ "model.layers.61.mlp.down_proj.weight": "model-00006-of-00014.safetensors",
648
+ "model.layers.61.mlp.gate_proj.weight": "model-00001-of-00014.safetensors",
649
+ "model.layers.61.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
650
+ "model.layers.61.post_attention_layernorm.weight": "model-00007-of-00014.safetensors",
651
+ "model.layers.61.self_attn.k_norm.weight": "model-00011-of-00014.safetensors",
652
+ "model.layers.61.self_attn.k_proj.weight": "model-00003-of-00014.safetensors",
653
+ "model.layers.61.self_attn.o_proj.weight": "model-00008-of-00014.safetensors",
654
+ "model.layers.61.self_attn.q_norm.weight": "model-00006-of-00014.safetensors",
655
+ "model.layers.61.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
656
+ "model.layers.61.self_attn.v_proj.weight": "model-00001-of-00014.safetensors",
657
+ "model.layers.62.input_layernorm.weight": "model-00006-of-00014.safetensors",
658
+ "model.layers.62.mlp.down_proj.weight": "model-00001-of-00014.safetensors",
659
+ "model.layers.62.mlp.gate_proj.weight": "model-00006-of-00014.safetensors",
660
+ "model.layers.62.mlp.up_proj.weight": "model-00003-of-00014.safetensors",
661
+ "model.layers.62.post_attention_layernorm.weight": "model-00002-of-00014.safetensors",
662
+ "model.layers.62.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
663
+ "model.layers.62.self_attn.k_proj.weight": "model-00008-of-00014.safetensors",
664
+ "model.layers.62.self_attn.o_proj.weight": "model-00013-of-00014.safetensors",
665
+ "model.layers.62.self_attn.q_norm.weight": "model-00009-of-00014.safetensors",
666
+ "model.layers.62.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
667
+ "model.layers.62.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
668
+ "model.layers.63.input_layernorm.weight": "model-00012-of-00014.safetensors",
669
+ "model.layers.63.mlp.down_proj.weight": "model-00002-of-00014.safetensors",
670
+ "model.layers.63.mlp.gate_proj.weight": "model-00002-of-00014.safetensors",
671
+ "model.layers.63.mlp.up_proj.weight": "model-00001-of-00014.safetensors",
672
+ "model.layers.63.post_attention_layernorm.weight": "model-00004-of-00014.safetensors",
673
+ "model.layers.63.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
674
+ "model.layers.63.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
675
+ "model.layers.63.self_attn.o_proj.weight": "model-00014-of-00014.safetensors",
676
+ "model.layers.63.self_attn.q_norm.weight": "model-00012-of-00014.safetensors",
677
+ "model.layers.63.self_attn.q_proj.weight": "model-00008-of-00014.safetensors",
678
+ "model.layers.63.self_attn.v_proj.weight": "model-00004-of-00014.safetensors",
679
+ "model.layers.7.input_layernorm.weight": "model-00008-of-00014.safetensors",
680
+ "model.layers.7.mlp.down_proj.weight": "model-00011-of-00014.safetensors",
681
+ "model.layers.7.mlp.gate_proj.weight": "model-00011-of-00014.safetensors",
682
+ "model.layers.7.mlp.up_proj.weight": "model-00014-of-00014.safetensors",
683
+ "model.layers.7.post_attention_layernorm.weight": "model-00001-of-00014.safetensors",
684
+ "model.layers.7.self_attn.k_norm.weight": "model-00005-of-00014.safetensors",
685
+ "model.layers.7.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
686
+ "model.layers.7.self_attn.o_proj.weight": "model-00010-of-00014.safetensors",
687
+ "model.layers.7.self_attn.q_norm.weight": "model-00013-of-00014.safetensors",
688
+ "model.layers.7.self_attn.q_proj.weight": "model-00011-of-00014.safetensors",
689
+ "model.layers.7.self_attn.v_proj.weight": "model-00009-of-00014.safetensors",
690
+ "model.layers.8.input_layernorm.weight": "model-00012-of-00014.safetensors",
691
+ "model.layers.8.mlp.down_proj.weight": "model-00007-of-00014.safetensors",
692
+ "model.layers.8.mlp.gate_proj.weight": "model-00009-of-00014.safetensors",
693
+ "model.layers.8.mlp.up_proj.weight": "model-00004-of-00014.safetensors",
694
+ "model.layers.8.post_attention_layernorm.weight": "model-00014-of-00014.safetensors",
695
+ "model.layers.8.self_attn.k_norm.weight": "model-00009-of-00014.safetensors",
696
+ "model.layers.8.self_attn.k_proj.weight": "model-00004-of-00014.safetensors",
697
+ "model.layers.8.self_attn.o_proj.weight": "model-00010-of-00014.safetensors",
698
+ "model.layers.8.self_attn.q_norm.weight": "model-00014-of-00014.safetensors",
699
+ "model.layers.8.self_attn.q_proj.weight": "model-00005-of-00014.safetensors",
700
+ "model.layers.8.self_attn.v_proj.weight": "model-00006-of-00014.safetensors",
701
+ "model.layers.9.input_layernorm.weight": "model-00007-of-00014.safetensors",
702
+ "model.layers.9.mlp.down_proj.weight": "model-00005-of-00014.safetensors",
703
+ "model.layers.9.mlp.gate_proj.weight": "model-00013-of-00014.safetensors",
704
+ "model.layers.9.mlp.up_proj.weight": "model-00013-of-00014.safetensors",
705
+ "model.layers.9.post_attention_layernorm.weight": "model-00012-of-00014.safetensors",
706
+ "model.layers.9.self_attn.k_norm.weight": "model-00013-of-00014.safetensors",
707
+ "model.layers.9.self_attn.k_proj.weight": "model-00005-of-00014.safetensors",
708
+ "model.layers.9.self_attn.o_proj.weight": "model-00002-of-00014.safetensors",
709
+ "model.layers.9.self_attn.q_norm.weight": "model-00001-of-00014.safetensors",
710
+ "model.layers.9.self_attn.q_proj.weight": "model-00003-of-00014.safetensors",
711
+ "model.layers.9.self_attn.v_proj.weight": "model-00006-of-00014.safetensors",
712
+ "model.norm.weight": "model-00011-of-00014.safetensors"
713
+ }
714
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<|im_start|>",
4
+ "<|im_end|>",
5
+ "<|object_ref_start|>",
6
+ "<|object_ref_end|>",
7
+ "<|box_start|>",
8
+ "<|box_end|>",
9
+ "<|quad_start|>",
10
+ "<|quad_end|>",
11
+ "<|vision_start|>",
12
+ "<|vision_end|>",
13
+ "<|vision_pad|>",
14
+ "<|image_pad|>",
15
+ "<|video_pad|>"
16
+ ],
17
+ "eos_token": {
18
+ "content": "<|im_end|>",
19
+ "lstrip": false,
20
+ "normalized": false,
21
+ "rstrip": false,
22
+ "single_word": false
23
+ },
24
+ "pad_token": {
25
+ "content": "<|endoftext|>",
26
+ "lstrip": false,
27
+ "normalized": false,
28
+ "rstrip": false,
29
+ "single_word": false
30
+ }
31
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:aeb13307a71acd8fe81861d94ad54ab689df773318809eed3cbe794b4492dae4
3
+ size 11422654
tokenizer_config.json ADDED
@@ -0,0 +1,240 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ },
181
+ "151665": {
182
+ "content": "<tool_response>",
183
+ "lstrip": false,
184
+ "normalized": false,
185
+ "rstrip": false,
186
+ "single_word": false,
187
+ "special": false
188
+ },
189
+ "151666": {
190
+ "content": "</tool_response>",
191
+ "lstrip": false,
192
+ "normalized": false,
193
+ "rstrip": false,
194
+ "single_word": false,
195
+ "special": false
196
+ },
197
+ "151667": {
198
+ "content": "<think>",
199
+ "lstrip": false,
200
+ "normalized": false,
201
+ "rstrip": false,
202
+ "single_word": false,
203
+ "special": false
204
+ },
205
+ "151668": {
206
+ "content": "</think>",
207
+ "lstrip": false,
208
+ "normalized": false,
209
+ "rstrip": false,
210
+ "single_word": false,
211
+ "special": false
212
+ }
213
+ },
214
+ "additional_special_tokens": [
215
+ "<|im_start|>",
216
+ "<|im_end|>",
217
+ "<|object_ref_start|>",
218
+ "<|object_ref_end|>",
219
+ "<|box_start|>",
220
+ "<|box_end|>",
221
+ "<|quad_start|>",
222
+ "<|quad_end|>",
223
+ "<|vision_start|>",
224
+ "<|vision_end|>",
225
+ "<|vision_pad|>",
226
+ "<|image_pad|>",
227
+ "<|video_pad|>"
228
+ ],
229
+ "bos_token": null,
230
+ "clean_up_tokenization_spaces": false,
231
+ "eos_token": "<|im_end|>",
232
+ "errors": "replace",
233
+ "extra_special_tokens": {},
234
+ "model_max_length": 131072,
235
+ "pad_token": "<|endoftext|>",
236
+ "split_special_tokens": false,
237
+ "tokenizer_class": "Qwen2Tokenizer",
238
+ "unk_token": null,
239
+ "chat_template": "{%- if tools %}\n {{- '<|im_start|>system\\n' }}\n {%- if messages[0].role == 'system' %}\n {{- messages[0].content + '\\n\\n' }}\n {%- endif %}\n {{- \"# Tools\\n\\nYou may call one or more functions to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>\" }}\n {%- for tool in tools %}\n {{- \"\\n\" }}\n {{- tool | tojson }}\n {%- endfor %}\n {{- \"\\n</tools>\\n\\nFor each function call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:\\n<tool_call>\\n{\\\"name\\\": <function-name>, \\\"arguments\\\": <args-json-object>}\\n</tool_call><|im_end|>\\n\" }}\n{%- else %}\n {%- if messages[0].role == 'system' %}\n {{- '<|im_start|>system\\n' + messages[0].content + '<|im_end|>\\n' }}\n {%- endif %}\n{%- endif %}\n{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}\n{%- for message in messages[::-1] %}\n {%- set index = (messages|length - 1) - loop.index0 %}\n {%- if ns.multi_step_tool and message.role == \"user\" and not(message.content.startswith('<tool_response>') and message.content.endswith('</tool_response>')) %}\n {%- set ns.multi_step_tool = false %}\n {%- set ns.last_query_index = index %}\n {%- endif %}\n{%- endfor %}\n{%- for message in messages %}\n {%- if (message.role == \"user\") or (message.role == \"system\" and not loop.first) %}\n {{- '<|im_start|>' + message.role + '\\n' + message.content + '<|im_end|>' + '\\n' }}\n {%- elif message.role == \"assistant\" %}\n {%- set content = message.content %}\n {%- set reasoning_content = '' %}\n {%- if message.reasoning_content is defined and message.reasoning_content is not none %}\n {%- set reasoning_content = message.reasoning_content %}\n {%- else %}\n {%- if '</think>' in message.content %}\n {%- set content = message.content.split('</think>')[-1].lstrip('\\n') %}\n {%- set reasoning_content = message.content.split('</think>')[0].rstrip('\\n').split('<think>')[-1].lstrip('\\n') %}\n {%- endif %}\n {%- endif %}\n {%- if loop.index0 > ns.last_query_index %}\n {%- if loop.last or (not loop.last and reasoning_content) %}\n {{- '<|im_start|>' + message.role + '\\n<think>\\n' + reasoning_content.strip('\\n') + '\\n</think>\\n\\n' + content.lstrip('\\n') }}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- else %}\n {{- '<|im_start|>' + message.role + '\\n' + content }}\n {%- endif %}\n {%- if message.tool_calls %}\n {%- for tool_call in message.tool_calls %}\n {%- if (loop.first and content) or (not loop.first) %}\n {{- '\\n' }}\n {%- endif %}\n {%- if tool_call.function %}\n {%- set tool_call = tool_call.function %}\n {%- endif %}\n {{- '<tool_call>\\n{\"name\": \"' }}\n {{- tool_call.name }}\n {{- '\", \"arguments\": ' }}\n {%- if tool_call.arguments is string %}\n {{- tool_call.arguments }}\n {%- else %}\n {{- tool_call.arguments | tojson }}\n {%- endif %}\n {{- '}\\n</tool_call>' }}\n {%- endfor %}\n {%- endif %}\n {{- '<|im_end|>\\n' }}\n {%- elif message.role == \"tool\" %}\n {%- if loop.first or (messages[loop.index0 - 1].role != \"tool\") %}\n {{- '<|im_start|>user' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- message.content }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != \"tool\") %}\n {{- '<|im_end|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|im_start|>assistant\\n' }}\n {%- if enable_thinking is defined and enable_thinking is false %}\n {{- '<think>\\n\\n</think>\\n\\n' }}\n {%- endif %}\n{%- endif %}"
240
+ }
tokenizer_config.json.bak ADDED
@@ -0,0 +1,239 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "151643": {
6
+ "content": "<|endoftext|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "151644": {
14
+ "content": "<|im_start|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "151645": {
22
+ "content": "<|im_end|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ },
29
+ "151646": {
30
+ "content": "<|object_ref_start|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": true
36
+ },
37
+ "151647": {
38
+ "content": "<|object_ref_end|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": true
44
+ },
45
+ "151648": {
46
+ "content": "<|box_start|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": true
52
+ },
53
+ "151649": {
54
+ "content": "<|box_end|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": true
60
+ },
61
+ "151650": {
62
+ "content": "<|quad_start|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": true
68
+ },
69
+ "151651": {
70
+ "content": "<|quad_end|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "151652": {
78
+ "content": "<|vision_start|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "151653": {
86
+ "content": "<|vision_end|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "151654": {
94
+ "content": "<|vision_pad|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "151655": {
102
+ "content": "<|image_pad|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "151656": {
110
+ "content": "<|video_pad|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "151657": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "151658": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "151659": {
134
+ "content": "<|fim_prefix|>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "151660": {
142
+ "content": "<|fim_middle|>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "151661": {
150
+ "content": "<|fim_suffix|>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "151662": {
158
+ "content": "<|fim_pad|>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "151663": {
166
+ "content": "<|repo_name|>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": false
172
+ },
173
+ "151664": {
174
+ "content": "<|file_sep|>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": false
180
+ },
181
+ "151665": {
182
+ "content": "<tool_response>",
183
+ "lstrip": false,
184
+ "normalized": false,
185
+ "rstrip": false,
186
+ "single_word": false,
187
+ "special": false
188
+ },
189
+ "151666": {
190
+ "content": "</tool_response>",
191
+ "lstrip": false,
192
+ "normalized": false,
193
+ "rstrip": false,
194
+ "single_word": false,
195
+ "special": false
196
+ },
197
+ "151667": {
198
+ "content": "<think>",
199
+ "lstrip": false,
200
+ "normalized": false,
201
+ "rstrip": false,
202
+ "single_word": false,
203
+ "special": false
204
+ },
205
+ "151668": {
206
+ "content": "</think>",
207
+ "lstrip": false,
208
+ "normalized": false,
209
+ "rstrip": false,
210
+ "single_word": false,
211
+ "special": false
212
+ }
213
+ },
214
+ "additional_special_tokens": [
215
+ "<|im_start|>",
216
+ "<|im_end|>",
217
+ "<|object_ref_start|>",
218
+ "<|object_ref_end|>",
219
+ "<|box_start|>",
220
+ "<|box_end|>",
221
+ "<|quad_start|>",
222
+ "<|quad_end|>",
223
+ "<|vision_start|>",
224
+ "<|vision_end|>",
225
+ "<|vision_pad|>",
226
+ "<|image_pad|>",
227
+ "<|video_pad|>"
228
+ ],
229
+ "bos_token": null,
230
+ "clean_up_tokenization_spaces": false,
231
+ "eos_token": "<|im_end|>",
232
+ "errors": "replace",
233
+ "extra_special_tokens": {},
234
+ "model_max_length": 131072,
235
+ "pad_token": "<|endoftext|>",
236
+ "split_special_tokens": false,
237
+ "tokenizer_class": "Qwen2Tokenizer",
238
+ "unk_token": null
239
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff