Hugging Face's logo

mlx-community
/
Qwen3-14B-3bit

Text Generation
MLX
Safetensors
qwen3
conversational
3-bit
Model card Files
xet
Community
1
KCh3dRi4n commited on
Commit
005507d
·
verified ·
1 Parent(s): c242965

Upload folder using huggingface_hub

Browse files
Files changed (8) hide show
  1. README.md +38 -18
  2. chat_template.jinja +89 -0
  3. config.json +968 -2
  4. generation_config.json +13 -0
  5. model-00001-of-00002.safetensors +2 -2
  6. model-00002-of-00002.safetensors +2 -2
  7. model.safetensors.index.json +2 -1
  8. tokenizer_config.json +0 -1
README.md CHANGED
@@ -1,37 +1,57 @@
1
  ---
2
- library_name: mlx
3
  license: apache-2.0
4
- license_link: https://huggingface.co/Qwen/Qwen3-14B/blob/main/LICENSE
5
- pipeline_tag: text-generation
6
  base_model: Qwen/Qwen3-14B
7
  tags:
8
  - mlx
 
 
9
  ---
10
 
11
- # mlx-community/Qwen3-14B-3bit
12
 
13
- This model [mlx-community/Qwen3-14B-3bit](https://huggingface.co/mlx-community/Qwen3-14B-3bit) was
14
- converted to MLX format from [Qwen/Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B)
15
- using mlx-lm version **0.24.0**.
16
 
17
- ## Use with mlx
 
 
 
 
 
18
 
19
- ```bash
20
- pip install mlx-lm
21
- ```
22
 
23
  ```python
24
  from mlx_lm import load, generate
25
 
26
  model, tokenizer = load("mlx-community/Qwen3-14B-3bit")
27
 
28
- prompt = "hello"
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
29
 
30
- if tokenizer.chat_template is not None:
31
- messages = [{"role": "user", "content": prompt}]
32
- prompt = tokenizer.apply_chat_template(
33
- messages, add_generation_prompt=True
34
- )
35
 
36
- response = generate(model, tokenizer, prompt=prompt, verbose=True)
 
 
 
 
 
 
 
37
  ```
 
1
  ---
 
2
  license: apache-2.0
 
 
3
  base_model: Qwen/Qwen3-14B
4
  tags:
5
  - mlx
6
+ - 3bit
7
+ - quantized
8
  ---
9
 
10
+ # Qwen3-14B 3bit MLX
11
 
12
+ This model is a 3-bit quantized version of [Qwen/Qwen3-14B](https://huggingface.co/Qwen/Qwen3-14B) using MLX.
 
 
13
 
14
+ ## Model Details
15
+ - **Quantization**: 3-bit
16
+ - **Framework**: MLX
17
+ - **Base Model**: Qwen/Qwen3-14B
18
+ - **Model Size**: ~5.25GB (3-bit quantized)
19
+ - **Original Size**: 14B parameters
20
 
21
+ ## Usage
 
 
22
 
23
  ```python
24
  from mlx_lm import load, generate
25
 
26
  model, tokenizer = load("mlx-community/Qwen3-14B-3bit")
27
 
28
+ prompt = "Hello, how are you?"
29
+ messages = [{"role": "user", "content": prompt}]
30
+ formatted_prompt = tokenizer.apply_chat_template(messages, add_generation_prompt=True)
31
+
32
+ response = generate(model, tokenizer, prompt=formatted_prompt, max_tokens=100)
33
+ print(response)
34
+ ```
35
+
36
+ ## Performance on M1/M2/M3
37
+ - **Memory Usage**: ~7-8GB
38
+ - **Inference Speed**: ~15-25 tokens/sec (M1 Max)
39
+ - **First Token Latency**: ~1-3 seconds
40
+
41
+ ## Requirements
42
+ - Apple Silicon Mac (M1/M2/M3)
43
+ - macOS 13.0+
44
+ - Python 3.8+
45
+ - MLX and mlx-lm packages
46
 
47
+ ## Installation
 
 
 
 
48
 
49
+ ```bash
50
+ pip install mlx mlx-lm
51
+ ```
52
+
53
+ ## Chat Mode
54
+
55
+ ```bash
56
+ mlx_lm.chat --model mlx-community/Qwen3-14B-3bit
57
  ```
chat_template.jinja ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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 message.content is string 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.content is string %}
27
+ {%- set content = message.content %}
28
+ {%- else %}
29
+ {%- set content = '' %}
30
+ {%- endif %}
31
+ {%- if (message.role == "user") or (message.role == "system" and not loop.first) %}
32
+ {{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
33
+ {%- elif message.role == "assistant" %}
34
+ {%- set reasoning_content = '' %}
35
+ {%- if message.reasoning_content is string %}
36
+ {%- set reasoning_content = message.reasoning_content %}
37
+ {%- else %}
38
+ {%- if '</think>' in content %}
39
+ {%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
40
+ {%- set content = content.split('</think>')[-1].lstrip('\n') %}
41
+ {%- endif %}
42
+ {%- endif %}
43
+ {%- if loop.index0 > ns.last_query_index %}
44
+ {%- if loop.last or (not loop.last and reasoning_content) %}
45
+ {{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content.strip('\n') + '\n</think>\n\n' + content.lstrip('\n') }}
46
+ {%- else %}
47
+ {{- '<|im_start|>' + message.role + '\n' + content }}
48
+ {%- endif %}
49
+ {%- else %}
50
+ {{- '<|im_start|>' + message.role + '\n' + content }}
51
+ {%- endif %}
52
+ {%- if message.tool_calls %}
53
+ {%- for tool_call in message.tool_calls %}
54
+ {%- if (loop.first and content) or (not loop.first) %}
55
+ {{- '\n' }}
56
+ {%- endif %}
57
+ {%- if tool_call.function %}
58
+ {%- set tool_call = tool_call.function %}
59
+ {%- endif %}
60
+ {{- '<tool_call>\n{"name": "' }}
61
+ {{- tool_call.name }}
62
+ {{- '", "arguments": ' }}
63
+ {%- if tool_call.arguments is string %}
64
+ {{- tool_call.arguments }}
65
+ {%- else %}
66
+ {{- tool_call.arguments | tojson }}
67
+ {%- endif %}
68
+ {{- '}\n</tool_call>' }}
69
+ {%- endfor %}
70
+ {%- endif %}
71
+ {{- '<|im_end|>\n' }}
72
+ {%- elif message.role == "tool" %}
73
+ {%- if loop.first or (messages[loop.index0 - 1].role != "tool") %}
74
+ {{- '<|im_start|>user' }}
75
+ {%- endif %}
76
+ {{- '\n<tool_response>\n' }}
77
+ {{- content }}
78
+ {{- '\n</tool_response>' }}
79
+ {%- if loop.last or (messages[loop.index0 + 1].role != "tool") %}
80
+ {{- '<|im_end|>\n' }}
81
+ {%- endif %}
82
+ {%- endif %}
83
+ {%- endfor %}
84
+ {%- if add_generation_prompt %}
85
+ {{- '<|im_start|>assistant\n' }}
86
+ {%- if enable_thinking is defined and enable_thinking is false %}
87
+ {{- '<think>\n\n</think>\n\n' }}
88
+ {%- endif %}
89
+ {%- endif %}
config.json CHANGED
@@ -19,11 +19,977 @@
19
  "num_key_value_heads": 8,
20
  "quantization": {
21
  "group_size": 64,
22
- "bits": 3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
23
  },
24
  "quantization_config": {
25
  "group_size": 64,
26
- "bits": 3
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
27
  },
28
  "rms_norm_eps": 1e-06,
29
  "rope_scaling": null,
 
19
  "num_key_value_heads": 8,
20
  "quantization": {
21
  "group_size": 64,
22
+ "bits": 3,
23
+ "model.embed_tokens": true,
24
+ "model.layers.0.self_attn.q_proj": true,
25
+ "model.layers.0.self_attn.k_proj": true,
26
+ "model.layers.0.self_attn.v_proj": true,
27
+ "model.layers.0.self_attn.o_proj": true,
28
+ "model.layers.0.self_attn.q_norm": false,
29
+ "model.layers.0.self_attn.k_norm": false,
30
+ "model.layers.0.self_attn.rope": false,
31
+ "model.layers.0.mlp.gate_proj": true,
32
+ "model.layers.0.mlp.down_proj": true,
33
+ "model.layers.0.mlp.up_proj": true,
34
+ "model.layers.0.input_layernorm": false,
35
+ "model.layers.0.post_attention_layernorm": false,
36
+ "model.layers.1.self_attn.q_proj": true,
37
+ "model.layers.1.self_attn.k_proj": true,
38
+ "model.layers.1.self_attn.v_proj": true,
39
+ "model.layers.1.self_attn.o_proj": true,
40
+ "model.layers.1.self_attn.q_norm": false,
41
+ "model.layers.1.self_attn.k_norm": false,
42
+ "model.layers.1.self_attn.rope": false,
43
+ "model.layers.1.mlp.gate_proj": true,
44
+ "model.layers.1.mlp.down_proj": true,
45
+ "model.layers.1.mlp.up_proj": true,
46
+ "model.layers.1.input_layernorm": false,
47
+ "model.layers.1.post_attention_layernorm": false,
48
+ "model.layers.2.self_attn.q_proj": true,
49
+ "model.layers.2.self_attn.k_proj": true,
50
+ "model.layers.2.self_attn.v_proj": true,
51
+ "model.layers.2.self_attn.o_proj": true,
52
+ "model.layers.2.self_attn.q_norm": false,
53
+ "model.layers.2.self_attn.k_norm": false,
54
+ "model.layers.2.self_attn.rope": false,
55
+ "model.layers.2.mlp.gate_proj": true,
56
+ "model.layers.2.mlp.down_proj": true,
57
+ "model.layers.2.mlp.up_proj": true,
58
+ "model.layers.2.input_layernorm": false,
59
+ "model.layers.2.post_attention_layernorm": false,
60
+ "model.layers.3.self_attn.q_proj": true,
61
+ "model.layers.3.self_attn.k_proj": true,
62
+ "model.layers.3.self_attn.v_proj": true,
63
+ "model.layers.3.self_attn.o_proj": true,
64
+ "model.layers.3.self_attn.q_norm": false,
65
+ "model.layers.3.self_attn.k_norm": false,
66
+ "model.layers.3.self_attn.rope": false,
67
+ "model.layers.3.mlp.gate_proj": true,
68
+ "model.layers.3.mlp.down_proj": true,
69
+ "model.layers.3.mlp.up_proj": true,
70
+ "model.layers.3.input_layernorm": false,
71
+ "model.layers.3.post_attention_layernorm": false,
72
+ "model.layers.4.self_attn.q_proj": true,
73
+ "model.layers.4.self_attn.k_proj": true,
74
+ "model.layers.4.self_attn.v_proj": true,
75
+ "model.layers.4.self_attn.o_proj": true,
76
+ "model.layers.4.self_attn.q_norm": false,
77
+ "model.layers.4.self_attn.k_norm": false,
78
+ "model.layers.4.self_attn.rope": false,
79
+ "model.layers.4.mlp.gate_proj": true,
80
+ "model.layers.4.mlp.down_proj": true,
81
+ "model.layers.4.mlp.up_proj": true,
82
+ "model.layers.4.input_layernorm": false,
83
+ "model.layers.4.post_attention_layernorm": false,
84
+ "model.layers.5.self_attn.q_proj": true,
85
+ "model.layers.5.self_attn.k_proj": true,
86
+ "model.layers.5.self_attn.v_proj": true,
87
+ "model.layers.5.self_attn.o_proj": true,
88
+ "model.layers.5.self_attn.q_norm": false,
89
+ "model.layers.5.self_attn.k_norm": false,
90
+ "model.layers.5.self_attn.rope": false,
91
+ "model.layers.5.mlp.gate_proj": true,
92
+ "model.layers.5.mlp.down_proj": true,
93
+ "model.layers.5.mlp.up_proj": true,
94
+ "model.layers.5.input_layernorm": false,
95
+ "model.layers.5.post_attention_layernorm": false,
96
+ "model.layers.6.self_attn.q_proj": true,
97
+ "model.layers.6.self_attn.k_proj": true,
98
+ "model.layers.6.self_attn.v_proj": true,
99
+ "model.layers.6.self_attn.o_proj": true,
100
+ "model.layers.6.self_attn.q_norm": false,
101
+ "model.layers.6.self_attn.k_norm": false,
102
+ "model.layers.6.self_attn.rope": false,
103
+ "model.layers.6.mlp.gate_proj": true,
104
+ "model.layers.6.mlp.down_proj": true,
105
+ "model.layers.6.mlp.up_proj": true,
106
+ "model.layers.6.input_layernorm": false,
107
+ "model.layers.6.post_attention_layernorm": false,
108
+ "model.layers.7.self_attn.q_proj": true,
109
+ "model.layers.7.self_attn.k_proj": true,
110
+ "model.layers.7.self_attn.v_proj": true,
111
+ "model.layers.7.self_attn.o_proj": true,
112
+ "model.layers.7.self_attn.q_norm": false,
113
+ "model.layers.7.self_attn.k_norm": false,
114
+ "model.layers.7.self_attn.rope": false,
115
+ "model.layers.7.mlp.gate_proj": true,
116
+ "model.layers.7.mlp.down_proj": true,
117
+ "model.layers.7.mlp.up_proj": true,
118
+ "model.layers.7.input_layernorm": false,
119
+ "model.layers.7.post_attention_layernorm": false,
120
+ "model.layers.8.self_attn.q_proj": true,
121
+ "model.layers.8.self_attn.k_proj": true,
122
+ "model.layers.8.self_attn.v_proj": true,
123
+ "model.layers.8.self_attn.o_proj": true,
124
+ "model.layers.8.self_attn.q_norm": false,
125
+ "model.layers.8.self_attn.k_norm": false,
126
+ "model.layers.8.self_attn.rope": false,
127
+ "model.layers.8.mlp.gate_proj": true,
128
+ "model.layers.8.mlp.down_proj": true,
129
+ "model.layers.8.mlp.up_proj": true,
130
+ "model.layers.8.input_layernorm": false,
131
+ "model.layers.8.post_attention_layernorm": false,
132
+ "model.layers.9.self_attn.q_proj": true,
133
+ "model.layers.9.self_attn.k_proj": true,
134
+ "model.layers.9.self_attn.v_proj": true,
135
+ "model.layers.9.self_attn.o_proj": true,
136
+ "model.layers.9.self_attn.q_norm": false,
137
+ "model.layers.9.self_attn.k_norm": false,
138
+ "model.layers.9.self_attn.rope": false,
139
+ "model.layers.9.mlp.gate_proj": true,
140
+ "model.layers.9.mlp.down_proj": true,
141
+ "model.layers.9.mlp.up_proj": true,
142
+ "model.layers.9.input_layernorm": false,
143
+ "model.layers.9.post_attention_layernorm": false,
144
+ "model.layers.10.self_attn.q_proj": true,
145
+ "model.layers.10.self_attn.k_proj": true,
146
+ "model.layers.10.self_attn.v_proj": true,
147
+ "model.layers.10.self_attn.o_proj": true,
148
+ "model.layers.10.self_attn.q_norm": false,
149
+ "model.layers.10.self_attn.k_norm": false,
150
+ "model.layers.10.self_attn.rope": false,
151
+ "model.layers.10.mlp.gate_proj": true,
152
+ "model.layers.10.mlp.down_proj": true,
153
+ "model.layers.10.mlp.up_proj": true,
154
+ "model.layers.10.input_layernorm": false,
155
+ "model.layers.10.post_attention_layernorm": false,
156
+ "model.layers.11.self_attn.q_proj": true,
157
+ "model.layers.11.self_attn.k_proj": true,
158
+ "model.layers.11.self_attn.v_proj": true,
159
+ "model.layers.11.self_attn.o_proj": true,
160
+ "model.layers.11.self_attn.q_norm": false,
161
+ "model.layers.11.self_attn.k_norm": false,
162
+ "model.layers.11.self_attn.rope": false,
163
+ "model.layers.11.mlp.gate_proj": true,
164
+ "model.layers.11.mlp.down_proj": true,
165
+ "model.layers.11.mlp.up_proj": true,
166
+ "model.layers.11.input_layernorm": false,
167
+ "model.layers.11.post_attention_layernorm": false,
168
+ "model.layers.12.self_attn.q_proj": true,
169
+ "model.layers.12.self_attn.k_proj": true,
170
+ "model.layers.12.self_attn.v_proj": true,
171
+ "model.layers.12.self_attn.o_proj": true,
172
+ "model.layers.12.self_attn.q_norm": false,
173
+ "model.layers.12.self_attn.k_norm": false,
174
+ "model.layers.12.self_attn.rope": false,
175
+ "model.layers.12.mlp.gate_proj": true,
176
+ "model.layers.12.mlp.down_proj": true,
177
+ "model.layers.12.mlp.up_proj": true,
178
+ "model.layers.12.input_layernorm": false,
179
+ "model.layers.12.post_attention_layernorm": false,
180
+ "model.layers.13.self_attn.q_proj": true,
181
+ "model.layers.13.self_attn.k_proj": true,
182
+ "model.layers.13.self_attn.v_proj": true,
183
+ "model.layers.13.self_attn.o_proj": true,
184
+ "model.layers.13.self_attn.q_norm": false,
185
+ "model.layers.13.self_attn.k_norm": false,
186
+ "model.layers.13.self_attn.rope": false,
187
+ "model.layers.13.mlp.gate_proj": true,
188
+ "model.layers.13.mlp.down_proj": true,
189
+ "model.layers.13.mlp.up_proj": true,
190
+ "model.layers.13.input_layernorm": false,
191
+ "model.layers.13.post_attention_layernorm": false,
192
+ "model.layers.14.self_attn.q_proj": true,
193
+ "model.layers.14.self_attn.k_proj": true,
194
+ "model.layers.14.self_attn.v_proj": true,
195
+ "model.layers.14.self_attn.o_proj": true,
196
+ "model.layers.14.self_attn.q_norm": false,
197
+ "model.layers.14.self_attn.k_norm": false,
198
+ "model.layers.14.self_attn.rope": false,
199
+ "model.layers.14.mlp.gate_proj": true,
200
+ "model.layers.14.mlp.down_proj": true,
201
+ "model.layers.14.mlp.up_proj": true,
202
+ "model.layers.14.input_layernorm": false,
203
+ "model.layers.14.post_attention_layernorm": false,
204
+ "model.layers.15.self_attn.q_proj": true,
205
+ "model.layers.15.self_attn.k_proj": true,
206
+ "model.layers.15.self_attn.v_proj": true,
207
+ "model.layers.15.self_attn.o_proj": true,
208
+ "model.layers.15.self_attn.q_norm": false,
209
+ "model.layers.15.self_attn.k_norm": false,
210
+ "model.layers.15.self_attn.rope": false,
211
+ "model.layers.15.mlp.gate_proj": true,
212
+ "model.layers.15.mlp.down_proj": true,
213
+ "model.layers.15.mlp.up_proj": true,
214
+ "model.layers.15.input_layernorm": false,
215
+ "model.layers.15.post_attention_layernorm": false,
216
+ "model.layers.16.self_attn.q_proj": true,
217
+ "model.layers.16.self_attn.k_proj": true,
218
+ "model.layers.16.self_attn.v_proj": true,
219
+ "model.layers.16.self_attn.o_proj": true,
220
+ "model.layers.16.self_attn.q_norm": false,
221
+ "model.layers.16.self_attn.k_norm": false,
222
+ "model.layers.16.self_attn.rope": false,
223
+ "model.layers.16.mlp.gate_proj": true,
224
+ "model.layers.16.mlp.down_proj": true,
225
+ "model.layers.16.mlp.up_proj": true,
226
+ "model.layers.16.input_layernorm": false,
227
+ "model.layers.16.post_attention_layernorm": false,
228
+ "model.layers.17.self_attn.q_proj": true,
229
+ "model.layers.17.self_attn.k_proj": true,
230
+ "model.layers.17.self_attn.v_proj": true,
231
+ "model.layers.17.self_attn.o_proj": true,
232
+ "model.layers.17.self_attn.q_norm": false,
233
+ "model.layers.17.self_attn.k_norm": false,
234
+ "model.layers.17.self_attn.rope": false,
235
+ "model.layers.17.mlp.gate_proj": true,
236
+ "model.layers.17.mlp.down_proj": true,
237
+ "model.layers.17.mlp.up_proj": true,
238
+ "model.layers.17.input_layernorm": false,
239
+ "model.layers.17.post_attention_layernorm": false,
240
+ "model.layers.18.self_attn.q_proj": true,
241
+ "model.layers.18.self_attn.k_proj": true,
242
+ "model.layers.18.self_attn.v_proj": true,
243
+ "model.layers.18.self_attn.o_proj": true,
244
+ "model.layers.18.self_attn.q_norm": false,
245
+ "model.layers.18.self_attn.k_norm": false,
246
+ "model.layers.18.self_attn.rope": false,
247
+ "model.layers.18.mlp.gate_proj": true,
248
+ "model.layers.18.mlp.down_proj": true,
249
+ "model.layers.18.mlp.up_proj": true,
250
+ "model.layers.18.input_layernorm": false,
251
+ "model.layers.18.post_attention_layernorm": false,
252
+ "model.layers.19.self_attn.q_proj": true,
253
+ "model.layers.19.self_attn.k_proj": true,
254
+ "model.layers.19.self_attn.v_proj": true,
255
+ "model.layers.19.self_attn.o_proj": true,
256
+ "model.layers.19.self_attn.q_norm": false,
257
+ "model.layers.19.self_attn.k_norm": false,
258
+ "model.layers.19.self_attn.rope": false,
259
+ "model.layers.19.mlp.gate_proj": true,
260
+ "model.layers.19.mlp.down_proj": true,
261
+ "model.layers.19.mlp.up_proj": true,
262
+ "model.layers.19.input_layernorm": false,
263
+ "model.layers.19.post_attention_layernorm": false,
264
+ "model.layers.20.self_attn.q_proj": true,
265
+ "model.layers.20.self_attn.k_proj": true,
266
+ "model.layers.20.self_attn.v_proj": true,
267
+ "model.layers.20.self_attn.o_proj": true,
268
+ "model.layers.20.self_attn.q_norm": false,
269
+ "model.layers.20.self_attn.k_norm": false,
270
+ "model.layers.20.self_attn.rope": false,
271
+ "model.layers.20.mlp.gate_proj": true,
272
+ "model.layers.20.mlp.down_proj": true,
273
+ "model.layers.20.mlp.up_proj": true,
274
+ "model.layers.20.input_layernorm": false,
275
+ "model.layers.20.post_attention_layernorm": false,
276
+ "model.layers.21.self_attn.q_proj": true,
277
+ "model.layers.21.self_attn.k_proj": true,
278
+ "model.layers.21.self_attn.v_proj": true,
279
+ "model.layers.21.self_attn.o_proj": true,
280
+ "model.layers.21.self_attn.q_norm": false,
281
+ "model.layers.21.self_attn.k_norm": false,
282
+ "model.layers.21.self_attn.rope": false,
283
+ "model.layers.21.mlp.gate_proj": true,
284
+ "model.layers.21.mlp.down_proj": true,
285
+ "model.layers.21.mlp.up_proj": true,
286
+ "model.layers.21.input_layernorm": false,
287
+ "model.layers.21.post_attention_layernorm": false,
288
+ "model.layers.22.self_attn.q_proj": true,
289
+ "model.layers.22.self_attn.k_proj": true,
290
+ "model.layers.22.self_attn.v_proj": true,
291
+ "model.layers.22.self_attn.o_proj": true,
292
+ "model.layers.22.self_attn.q_norm": false,
293
+ "model.layers.22.self_attn.k_norm": false,
294
+ "model.layers.22.self_attn.rope": false,
295
+ "model.layers.22.mlp.gate_proj": true,
296
+ "model.layers.22.mlp.down_proj": true,
297
+ "model.layers.22.mlp.up_proj": true,
298
+ "model.layers.22.input_layernorm": false,
299
+ "model.layers.22.post_attention_layernorm": false,
300
+ "model.layers.23.self_attn.q_proj": true,
301
+ "model.layers.23.self_attn.k_proj": true,
302
+ "model.layers.23.self_attn.v_proj": true,
303
+ "model.layers.23.self_attn.o_proj": true,
304
+ "model.layers.23.self_attn.q_norm": false,
305
+ "model.layers.23.self_attn.k_norm": false,
306
+ "model.layers.23.self_attn.rope": false,
307
+ "model.layers.23.mlp.gate_proj": true,
308
+ "model.layers.23.mlp.down_proj": true,
309
+ "model.layers.23.mlp.up_proj": true,
310
+ "model.layers.23.input_layernorm": false,
311
+ "model.layers.23.post_attention_layernorm": false,
312
+ "model.layers.24.self_attn.q_proj": true,
313
+ "model.layers.24.self_attn.k_proj": true,
314
+ "model.layers.24.self_attn.v_proj": true,
315
+ "model.layers.24.self_attn.o_proj": true,
316
+ "model.layers.24.self_attn.q_norm": false,
317
+ "model.layers.24.self_attn.k_norm": false,
318
+ "model.layers.24.self_attn.rope": false,
319
+ "model.layers.24.mlp.gate_proj": true,
320
+ "model.layers.24.mlp.down_proj": true,
321
+ "model.layers.24.mlp.up_proj": true,
322
+ "model.layers.24.input_layernorm": false,
323
+ "model.layers.24.post_attention_layernorm": false,
324
+ "model.layers.25.self_attn.q_proj": true,
325
+ "model.layers.25.self_attn.k_proj": true,
326
+ "model.layers.25.self_attn.v_proj": true,
327
+ "model.layers.25.self_attn.o_proj": true,
328
+ "model.layers.25.self_attn.q_norm": false,
329
+ "model.layers.25.self_attn.k_norm": false,
330
+ "model.layers.25.self_attn.rope": false,
331
+ "model.layers.25.mlp.gate_proj": true,
332
+ "model.layers.25.mlp.down_proj": true,
333
+ "model.layers.25.mlp.up_proj": true,
334
+ "model.layers.25.input_layernorm": false,
335
+ "model.layers.25.post_attention_layernorm": false,
336
+ "model.layers.26.self_attn.q_proj": true,
337
+ "model.layers.26.self_attn.k_proj": true,
338
+ "model.layers.26.self_attn.v_proj": true,
339
+ "model.layers.26.self_attn.o_proj": true,
340
+ "model.layers.26.self_attn.q_norm": false,
341
+ "model.layers.26.self_attn.k_norm": false,
342
+ "model.layers.26.self_attn.rope": false,
343
+ "model.layers.26.mlp.gate_proj": true,
344
+ "model.layers.26.mlp.down_proj": true,
345
+ "model.layers.26.mlp.up_proj": true,
346
+ "model.layers.26.input_layernorm": false,
347
+ "model.layers.26.post_attention_layernorm": false,
348
+ "model.layers.27.self_attn.q_proj": true,
349
+ "model.layers.27.self_attn.k_proj": true,
350
+ "model.layers.27.self_attn.v_proj": true,
351
+ "model.layers.27.self_attn.o_proj": true,
352
+ "model.layers.27.self_attn.q_norm": false,
353
+ "model.layers.27.self_attn.k_norm": false,
354
+ "model.layers.27.self_attn.rope": false,
355
+ "model.layers.27.mlp.gate_proj": true,
356
+ "model.layers.27.mlp.down_proj": true,
357
+ "model.layers.27.mlp.up_proj": true,
358
+ "model.layers.27.input_layernorm": false,
359
+ "model.layers.27.post_attention_layernorm": false,
360
+ "model.layers.28.self_attn.q_proj": true,
361
+ "model.layers.28.self_attn.k_proj": true,
362
+ "model.layers.28.self_attn.v_proj": true,
363
+ "model.layers.28.self_attn.o_proj": true,
364
+ "model.layers.28.self_attn.q_norm": false,
365
+ "model.layers.28.self_attn.k_norm": false,
366
+ "model.layers.28.self_attn.rope": false,
367
+ "model.layers.28.mlp.gate_proj": true,
368
+ "model.layers.28.mlp.down_proj": true,
369
+ "model.layers.28.mlp.up_proj": true,
370
+ "model.layers.28.input_layernorm": false,
371
+ "model.layers.28.post_attention_layernorm": false,
372
+ "model.layers.29.self_attn.q_proj": true,
373
+ "model.layers.29.self_attn.k_proj": true,
374
+ "model.layers.29.self_attn.v_proj": true,
375
+ "model.layers.29.self_attn.o_proj": true,
376
+ "model.layers.29.self_attn.q_norm": false,
377
+ "model.layers.29.self_attn.k_norm": false,
378
+ "model.layers.29.self_attn.rope": false,
379
+ "model.layers.29.mlp.gate_proj": true,
380
+ "model.layers.29.mlp.down_proj": true,
381
+ "model.layers.29.mlp.up_proj": true,
382
+ "model.layers.29.input_layernorm": false,
383
+ "model.layers.29.post_attention_layernorm": false,
384
+ "model.layers.30.self_attn.q_proj": true,
385
+ "model.layers.30.self_attn.k_proj": true,
386
+ "model.layers.30.self_attn.v_proj": true,
387
+ "model.layers.30.self_attn.o_proj": true,
388
+ "model.layers.30.self_attn.q_norm": false,
389
+ "model.layers.30.self_attn.k_norm": false,
390
+ "model.layers.30.self_attn.rope": false,
391
+ "model.layers.30.mlp.gate_proj": true,
392
+ "model.layers.30.mlp.down_proj": true,
393
+ "model.layers.30.mlp.up_proj": true,
394
+ "model.layers.30.input_layernorm": false,
395
+ "model.layers.30.post_attention_layernorm": false,
396
+ "model.layers.31.self_attn.q_proj": true,
397
+ "model.layers.31.self_attn.k_proj": true,
398
+ "model.layers.31.self_attn.v_proj": true,
399
+ "model.layers.31.self_attn.o_proj": true,
400
+ "model.layers.31.self_attn.q_norm": false,
401
+ "model.layers.31.self_attn.k_norm": false,
402
+ "model.layers.31.self_attn.rope": false,
403
+ "model.layers.31.mlp.gate_proj": true,
404
+ "model.layers.31.mlp.down_proj": true,
405
+ "model.layers.31.mlp.up_proj": true,
406
+ "model.layers.31.input_layernorm": false,
407
+ "model.layers.31.post_attention_layernorm": false,
408
+ "model.layers.32.self_attn.q_proj": true,
409
+ "model.layers.32.self_attn.k_proj": true,
410
+ "model.layers.32.self_attn.v_proj": true,
411
+ "model.layers.32.self_attn.o_proj": true,
412
+ "model.layers.32.self_attn.q_norm": false,
413
+ "model.layers.32.self_attn.k_norm": false,
414
+ "model.layers.32.self_attn.rope": false,
415
+ "model.layers.32.mlp.gate_proj": true,
416
+ "model.layers.32.mlp.down_proj": true,
417
+ "model.layers.32.mlp.up_proj": true,
418
+ "model.layers.32.input_layernorm": false,
419
+ "model.layers.32.post_attention_layernorm": false,
420
+ "model.layers.33.self_attn.q_proj": true,
421
+ "model.layers.33.self_attn.k_proj": true,
422
+ "model.layers.33.self_attn.v_proj": true,
423
+ "model.layers.33.self_attn.o_proj": true,
424
+ "model.layers.33.self_attn.q_norm": false,
425
+ "model.layers.33.self_attn.k_norm": false,
426
+ "model.layers.33.self_attn.rope": false,
427
+ "model.layers.33.mlp.gate_proj": true,
428
+ "model.layers.33.mlp.down_proj": true,
429
+ "model.layers.33.mlp.up_proj": true,
430
+ "model.layers.33.input_layernorm": false,
431
+ "model.layers.33.post_attention_layernorm": false,
432
+ "model.layers.34.self_attn.q_proj": true,
433
+ "model.layers.34.self_attn.k_proj": true,
434
+ "model.layers.34.self_attn.v_proj": true,
435
+ "model.layers.34.self_attn.o_proj": true,
436
+ "model.layers.34.self_attn.q_norm": false,
437
+ "model.layers.34.self_attn.k_norm": false,
438
+ "model.layers.34.self_attn.rope": false,
439
+ "model.layers.34.mlp.gate_proj": true,
440
+ "model.layers.34.mlp.down_proj": true,
441
+ "model.layers.34.mlp.up_proj": true,
442
+ "model.layers.34.input_layernorm": false,
443
+ "model.layers.34.post_attention_layernorm": false,
444
+ "model.layers.35.self_attn.q_proj": true,
445
+ "model.layers.35.self_attn.k_proj": true,
446
+ "model.layers.35.self_attn.v_proj": true,
447
+ "model.layers.35.self_attn.o_proj": true,
448
+ "model.layers.35.self_attn.q_norm": false,
449
+ "model.layers.35.self_attn.k_norm": false,
450
+ "model.layers.35.self_attn.rope": false,
451
+ "model.layers.35.mlp.gate_proj": true,
452
+ "model.layers.35.mlp.down_proj": true,
453
+ "model.layers.35.mlp.up_proj": true,
454
+ "model.layers.35.input_layernorm": false,
455
+ "model.layers.35.post_attention_layernorm": false,
456
+ "model.layers.36.self_attn.q_proj": true,
457
+ "model.layers.36.self_attn.k_proj": true,
458
+ "model.layers.36.self_attn.v_proj": true,
459
+ "model.layers.36.self_attn.o_proj": true,
460
+ "model.layers.36.self_attn.q_norm": false,
461
+ "model.layers.36.self_attn.k_norm": false,
462
+ "model.layers.36.self_attn.rope": false,
463
+ "model.layers.36.mlp.gate_proj": true,
464
+ "model.layers.36.mlp.down_proj": true,
465
+ "model.layers.36.mlp.up_proj": true,
466
+ "model.layers.36.input_layernorm": false,
467
+ "model.layers.36.post_attention_layernorm": false,
468
+ "model.layers.37.self_attn.q_proj": true,
469
+ "model.layers.37.self_attn.k_proj": true,
470
+ "model.layers.37.self_attn.v_proj": true,
471
+ "model.layers.37.self_attn.o_proj": true,
472
+ "model.layers.37.self_attn.q_norm": false,
473
+ "model.layers.37.self_attn.k_norm": false,
474
+ "model.layers.37.self_attn.rope": false,
475
+ "model.layers.37.mlp.gate_proj": true,
476
+ "model.layers.37.mlp.down_proj": true,
477
+ "model.layers.37.mlp.up_proj": true,
478
+ "model.layers.37.input_layernorm": false,
479
+ "model.layers.37.post_attention_layernorm": false,
480
+ "model.layers.38.self_attn.q_proj": true,
481
+ "model.layers.38.self_attn.k_proj": true,
482
+ "model.layers.38.self_attn.v_proj": true,
483
+ "model.layers.38.self_attn.o_proj": true,
484
+ "model.layers.38.self_attn.q_norm": false,
485
+ "model.layers.38.self_attn.k_norm": false,
486
+ "model.layers.38.self_attn.rope": false,
487
+ "model.layers.38.mlp.gate_proj": true,
488
+ "model.layers.38.mlp.down_proj": true,
489
+ "model.layers.38.mlp.up_proj": true,
490
+ "model.layers.38.input_layernorm": false,
491
+ "model.layers.38.post_attention_layernorm": false,
492
+ "model.layers.39.self_attn.q_proj": true,
493
+ "model.layers.39.self_attn.k_proj": true,
494
+ "model.layers.39.self_attn.v_proj": true,
495
+ "model.layers.39.self_attn.o_proj": true,
496
+ "model.layers.39.self_attn.q_norm": false,
497
+ "model.layers.39.self_attn.k_norm": false,
498
+ "model.layers.39.self_attn.rope": false,
499
+ "model.layers.39.mlp.gate_proj": true,
500
+ "model.layers.39.mlp.down_proj": true,
501
+ "model.layers.39.mlp.up_proj": true,
502
+ "model.layers.39.input_layernorm": false,
503
+ "model.layers.39.post_attention_layernorm": false,
504
+ "model.norm": false,
505
+ "lm_head": true
506
  },
507
  "quantization_config": {
508
  "group_size": 64,
509
+ "bits": 3,
510
+ "model.embed_tokens": true,
511
+ "model.layers.0.self_attn.q_proj": true,
512
+ "model.layers.0.self_attn.k_proj": true,
513
+ "model.layers.0.self_attn.v_proj": true,
514
+ "model.layers.0.self_attn.o_proj": true,
515
+ "model.layers.0.self_attn.q_norm": false,
516
+ "model.layers.0.self_attn.k_norm": false,
517
+ "model.layers.0.self_attn.rope": false,
518
+ "model.layers.0.mlp.gate_proj": true,
519
+ "model.layers.0.mlp.down_proj": true,
520
+ "model.layers.0.mlp.up_proj": true,
521
+ "model.layers.0.input_layernorm": false,
522
+ "model.layers.0.post_attention_layernorm": false,
523
+ "model.layers.1.self_attn.q_proj": true,
524
+ "model.layers.1.self_attn.k_proj": true,
525
+ "model.layers.1.self_attn.v_proj": true,
526
+ "model.layers.1.self_attn.o_proj": true,
527
+ "model.layers.1.self_attn.q_norm": false,
528
+ "model.layers.1.self_attn.k_norm": false,
529
+ "model.layers.1.self_attn.rope": false,
530
+ "model.layers.1.mlp.gate_proj": true,
531
+ "model.layers.1.mlp.down_proj": true,
532
+ "model.layers.1.mlp.up_proj": true,
533
+ "model.layers.1.input_layernorm": false,
534
+ "model.layers.1.post_attention_layernorm": false,
535
+ "model.layers.2.self_attn.q_proj": true,
536
+ "model.layers.2.self_attn.k_proj": true,
537
+ "model.layers.2.self_attn.v_proj": true,
538
+ "model.layers.2.self_attn.o_proj": true,
539
+ "model.layers.2.self_attn.q_norm": false,
540
+ "model.layers.2.self_attn.k_norm": false,
541
+ "model.layers.2.self_attn.rope": false,
542
+ "model.layers.2.mlp.gate_proj": true,
543
+ "model.layers.2.mlp.down_proj": true,
544
+ "model.layers.2.mlp.up_proj": true,
545
+ "model.layers.2.input_layernorm": false,
546
+ "model.layers.2.post_attention_layernorm": false,
547
+ "model.layers.3.self_attn.q_proj": true,
548
+ "model.layers.3.self_attn.k_proj": true,
549
+ "model.layers.3.self_attn.v_proj": true,
550
+ "model.layers.3.self_attn.o_proj": true,
551
+ "model.layers.3.self_attn.q_norm": false,
552
+ "model.layers.3.self_attn.k_norm": false,
553
+ "model.layers.3.self_attn.rope": false,
554
+ "model.layers.3.mlp.gate_proj": true,
555
+ "model.layers.3.mlp.down_proj": true,
556
+ "model.layers.3.mlp.up_proj": true,
557
+ "model.layers.3.input_layernorm": false,
558
+ "model.layers.3.post_attention_layernorm": false,
559
+ "model.layers.4.self_attn.q_proj": true,
560
+ "model.layers.4.self_attn.k_proj": true,
561
+ "model.layers.4.self_attn.v_proj": true,
562
+ "model.layers.4.self_attn.o_proj": true,
563
+ "model.layers.4.self_attn.q_norm": false,
564
+ "model.layers.4.self_attn.k_norm": false,
565
+ "model.layers.4.self_attn.rope": false,
566
+ "model.layers.4.mlp.gate_proj": true,
567
+ "model.layers.4.mlp.down_proj": true,
568
+ "model.layers.4.mlp.up_proj": true,
569
+ "model.layers.4.input_layernorm": false,
570
+ "model.layers.4.post_attention_layernorm": false,
571
+ "model.layers.5.self_attn.q_proj": true,
572
+ "model.layers.5.self_attn.k_proj": true,
573
+ "model.layers.5.self_attn.v_proj": true,
574
+ "model.layers.5.self_attn.o_proj": true,
575
+ "model.layers.5.self_attn.q_norm": false,
576
+ "model.layers.5.self_attn.k_norm": false,
577
+ "model.layers.5.self_attn.rope": false,
578
+ "model.layers.5.mlp.gate_proj": true,
579
+ "model.layers.5.mlp.down_proj": true,
580
+ "model.layers.5.mlp.up_proj": true,
581
+ "model.layers.5.input_layernorm": false,
582
+ "model.layers.5.post_attention_layernorm": false,
583
+ "model.layers.6.self_attn.q_proj": true,
584
+ "model.layers.6.self_attn.k_proj": true,
585
+ "model.layers.6.self_attn.v_proj": true,
586
+ "model.layers.6.self_attn.o_proj": true,
587
+ "model.layers.6.self_attn.q_norm": false,
588
+ "model.layers.6.self_attn.k_norm": false,
589
+ "model.layers.6.self_attn.rope": false,
590
+ "model.layers.6.mlp.gate_proj": true,
591
+ "model.layers.6.mlp.down_proj": true,
592
+ "model.layers.6.mlp.up_proj": true,
593
+ "model.layers.6.input_layernorm": false,
594
+ "model.layers.6.post_attention_layernorm": false,
595
+ "model.layers.7.self_attn.q_proj": true,
596
+ "model.layers.7.self_attn.k_proj": true,
597
+ "model.layers.7.self_attn.v_proj": true,
598
+ "model.layers.7.self_attn.o_proj": true,
599
+ "model.layers.7.self_attn.q_norm": false,
600
+ "model.layers.7.self_attn.k_norm": false,
601
+ "model.layers.7.self_attn.rope": false,
602
+ "model.layers.7.mlp.gate_proj": true,
603
+ "model.layers.7.mlp.down_proj": true,
604
+ "model.layers.7.mlp.up_proj": true,
605
+ "model.layers.7.input_layernorm": false,
606
+ "model.layers.7.post_attention_layernorm": false,
607
+ "model.layers.8.self_attn.q_proj": true,
608
+ "model.layers.8.self_attn.k_proj": true,
609
+ "model.layers.8.self_attn.v_proj": true,
610
+ "model.layers.8.self_attn.o_proj": true,
611
+ "model.layers.8.self_attn.q_norm": false,
612
+ "model.layers.8.self_attn.k_norm": false,
613
+ "model.layers.8.self_attn.rope": false,
614
+ "model.layers.8.mlp.gate_proj": true,
615
+ "model.layers.8.mlp.down_proj": true,
616
+ "model.layers.8.mlp.up_proj": true,
617
+ "model.layers.8.input_layernorm": false,
618
+ "model.layers.8.post_attention_layernorm": false,
619
+ "model.layers.9.self_attn.q_proj": true,
620
+ "model.layers.9.self_attn.k_proj": true,
621
+ "model.layers.9.self_attn.v_proj": true,
622
+ "model.layers.9.self_attn.o_proj": true,
623
+ "model.layers.9.self_attn.q_norm": false,
624
+ "model.layers.9.self_attn.k_norm": false,
625
+ "model.layers.9.self_attn.rope": false,
626
+ "model.layers.9.mlp.gate_proj": true,
627
+ "model.layers.9.mlp.down_proj": true,
628
+ "model.layers.9.mlp.up_proj": true,
629
+ "model.layers.9.input_layernorm": false,
630
+ "model.layers.9.post_attention_layernorm": false,
631
+ "model.layers.10.self_attn.q_proj": true,
632
+ "model.layers.10.self_attn.k_proj": true,
633
+ "model.layers.10.self_attn.v_proj": true,
634
+ "model.layers.10.self_attn.o_proj": true,
635
+ "model.layers.10.self_attn.q_norm": false,
636
+ "model.layers.10.self_attn.k_norm": false,
637
+ "model.layers.10.self_attn.rope": false,
638
+ "model.layers.10.mlp.gate_proj": true,
639
+ "model.layers.10.mlp.down_proj": true,
640
+ "model.layers.10.mlp.up_proj": true,
641
+ "model.layers.10.input_layernorm": false,
642
+ "model.layers.10.post_attention_layernorm": false,
643
+ "model.layers.11.self_attn.q_proj": true,
644
+ "model.layers.11.self_attn.k_proj": true,
645
+ "model.layers.11.self_attn.v_proj": true,
646
+ "model.layers.11.self_attn.o_proj": true,
647
+ "model.layers.11.self_attn.q_norm": false,
648
+ "model.layers.11.self_attn.k_norm": false,
649
+ "model.layers.11.self_attn.rope": false,
650
+ "model.layers.11.mlp.gate_proj": true,
651
+ "model.layers.11.mlp.down_proj": true,
652
+ "model.layers.11.mlp.up_proj": true,
653
+ "model.layers.11.input_layernorm": false,
654
+ "model.layers.11.post_attention_layernorm": false,
655
+ "model.layers.12.self_attn.q_proj": true,
656
+ "model.layers.12.self_attn.k_proj": true,
657
+ "model.layers.12.self_attn.v_proj": true,
658
+ "model.layers.12.self_attn.o_proj": true,
659
+ "model.layers.12.self_attn.q_norm": false,
660
+ "model.layers.12.self_attn.k_norm": false,
661
+ "model.layers.12.self_attn.rope": false,
662
+ "model.layers.12.mlp.gate_proj": true,
663
+ "model.layers.12.mlp.down_proj": true,
664
+ "model.layers.12.mlp.up_proj": true,
665
+ "model.layers.12.input_layernorm": false,
666
+ "model.layers.12.post_attention_layernorm": false,
667
+ "model.layers.13.self_attn.q_proj": true,
668
+ "model.layers.13.self_attn.k_proj": true,
669
+ "model.layers.13.self_attn.v_proj": true,
670
+ "model.layers.13.self_attn.o_proj": true,
671
+ "model.layers.13.self_attn.q_norm": false,
672
+ "model.layers.13.self_attn.k_norm": false,
673
+ "model.layers.13.self_attn.rope": false,
674
+ "model.layers.13.mlp.gate_proj": true,
675
+ "model.layers.13.mlp.down_proj": true,
676
+ "model.layers.13.mlp.up_proj": true,
677
+ "model.layers.13.input_layernorm": false,
678
+ "model.layers.13.post_attention_layernorm": false,
679
+ "model.layers.14.self_attn.q_proj": true,
680
+ "model.layers.14.self_attn.k_proj": true,
681
+ "model.layers.14.self_attn.v_proj": true,
682
+ "model.layers.14.self_attn.o_proj": true,
683
+ "model.layers.14.self_attn.q_norm": false,
684
+ "model.layers.14.self_attn.k_norm": false,
685
+ "model.layers.14.self_attn.rope": false,
686
+ "model.layers.14.mlp.gate_proj": true,
687
+ "model.layers.14.mlp.down_proj": true,
688
+ "model.layers.14.mlp.up_proj": true,
689
+ "model.layers.14.input_layernorm": false,
690
+ "model.layers.14.post_attention_layernorm": false,
691
+ "model.layers.15.self_attn.q_proj": true,
692
+ "model.layers.15.self_attn.k_proj": true,
693
+ "model.layers.15.self_attn.v_proj": true,
694
+ "model.layers.15.self_attn.o_proj": true,
695
+ "model.layers.15.self_attn.q_norm": false,
696
+ "model.layers.15.self_attn.k_norm": false,
697
+ "model.layers.15.self_attn.rope": false,
698
+ "model.layers.15.mlp.gate_proj": true,
699
+ "model.layers.15.mlp.down_proj": true,
700
+ "model.layers.15.mlp.up_proj": true,
701
+ "model.layers.15.input_layernorm": false,
702
+ "model.layers.15.post_attention_layernorm": false,
703
+ "model.layers.16.self_attn.q_proj": true,
704
+ "model.layers.16.self_attn.k_proj": true,
705
+ "model.layers.16.self_attn.v_proj": true,
706
+ "model.layers.16.self_attn.o_proj": true,
707
+ "model.layers.16.self_attn.q_norm": false,
708
+ "model.layers.16.self_attn.k_norm": false,
709
+ "model.layers.16.self_attn.rope": false,
710
+ "model.layers.16.mlp.gate_proj": true,
711
+ "model.layers.16.mlp.down_proj": true,
712
+ "model.layers.16.mlp.up_proj": true,
713
+ "model.layers.16.input_layernorm": false,
714
+ "model.layers.16.post_attention_layernorm": false,
715
+ "model.layers.17.self_attn.q_proj": true,
716
+ "model.layers.17.self_attn.k_proj": true,
717
+ "model.layers.17.self_attn.v_proj": true,
718
+ "model.layers.17.self_attn.o_proj": true,
719
+ "model.layers.17.self_attn.q_norm": false,
720
+ "model.layers.17.self_attn.k_norm": false,
721
+ "model.layers.17.self_attn.rope": false,
722
+ "model.layers.17.mlp.gate_proj": true,
723
+ "model.layers.17.mlp.down_proj": true,
724
+ "model.layers.17.mlp.up_proj": true,
725
+ "model.layers.17.input_layernorm": false,
726
+ "model.layers.17.post_attention_layernorm": false,
727
+ "model.layers.18.self_attn.q_proj": true,
728
+ "model.layers.18.self_attn.k_proj": true,
729
+ "model.layers.18.self_attn.v_proj": true,
730
+ "model.layers.18.self_attn.o_proj": true,
731
+ "model.layers.18.self_attn.q_norm": false,
732
+ "model.layers.18.self_attn.k_norm": false,
733
+ "model.layers.18.self_attn.rope": false,
734
+ "model.layers.18.mlp.gate_proj": true,
735
+ "model.layers.18.mlp.down_proj": true,
736
+ "model.layers.18.mlp.up_proj": true,
737
+ "model.layers.18.input_layernorm": false,
738
+ "model.layers.18.post_attention_layernorm": false,
739
+ "model.layers.19.self_attn.q_proj": true,
740
+ "model.layers.19.self_attn.k_proj": true,
741
+ "model.layers.19.self_attn.v_proj": true,
742
+ "model.layers.19.self_attn.o_proj": true,
743
+ "model.layers.19.self_attn.q_norm": false,
744
+ "model.layers.19.self_attn.k_norm": false,
745
+ "model.layers.19.self_attn.rope": false,
746
+ "model.layers.19.mlp.gate_proj": true,
747
+ "model.layers.19.mlp.down_proj": true,
748
+ "model.layers.19.mlp.up_proj": true,
749
+ "model.layers.19.input_layernorm": false,
750
+ "model.layers.19.post_attention_layernorm": false,
751
+ "model.layers.20.self_attn.q_proj": true,
752
+ "model.layers.20.self_attn.k_proj": true,
753
+ "model.layers.20.self_attn.v_proj": true,
754
+ "model.layers.20.self_attn.o_proj": true,
755
+ "model.layers.20.self_attn.q_norm": false,
756
+ "model.layers.20.self_attn.k_norm": false,
757
+ "model.layers.20.self_attn.rope": false,
758
+ "model.layers.20.mlp.gate_proj": true,
759
+ "model.layers.20.mlp.down_proj": true,
760
+ "model.layers.20.mlp.up_proj": true,
761
+ "model.layers.20.input_layernorm": false,
762
+ "model.layers.20.post_attention_layernorm": false,
763
+ "model.layers.21.self_attn.q_proj": true,
764
+ "model.layers.21.self_attn.k_proj": true,
765
+ "model.layers.21.self_attn.v_proj": true,
766
+ "model.layers.21.self_attn.o_proj": true,
767
+ "model.layers.21.self_attn.q_norm": false,
768
+ "model.layers.21.self_attn.k_norm": false,
769
+ "model.layers.21.self_attn.rope": false,
770
+ "model.layers.21.mlp.gate_proj": true,
771
+ "model.layers.21.mlp.down_proj": true,
772
+ "model.layers.21.mlp.up_proj": true,
773
+ "model.layers.21.input_layernorm": false,
774
+ "model.layers.21.post_attention_layernorm": false,
775
+ "model.layers.22.self_attn.q_proj": true,
776
+ "model.layers.22.self_attn.k_proj": true,
777
+ "model.layers.22.self_attn.v_proj": true,
778
+ "model.layers.22.self_attn.o_proj": true,
779
+ "model.layers.22.self_attn.q_norm": false,
780
+ "model.layers.22.self_attn.k_norm": false,
781
+ "model.layers.22.self_attn.rope": false,
782
+ "model.layers.22.mlp.gate_proj": true,
783
+ "model.layers.22.mlp.down_proj": true,
784
+ "model.layers.22.mlp.up_proj": true,
785
+ "model.layers.22.input_layernorm": false,
786
+ "model.layers.22.post_attention_layernorm": false,
787
+ "model.layers.23.self_attn.q_proj": true,
788
+ "model.layers.23.self_attn.k_proj": true,
789
+ "model.layers.23.self_attn.v_proj": true,
790
+ "model.layers.23.self_attn.o_proj": true,
791
+ "model.layers.23.self_attn.q_norm": false,
792
+ "model.layers.23.self_attn.k_norm": false,
793
+ "model.layers.23.self_attn.rope": false,
794
+ "model.layers.23.mlp.gate_proj": true,
795
+ "model.layers.23.mlp.down_proj": true,
796
+ "model.layers.23.mlp.up_proj": true,
797
+ "model.layers.23.input_layernorm": false,
798
+ "model.layers.23.post_attention_layernorm": false,
799
+ "model.layers.24.self_attn.q_proj": true,
800
+ "model.layers.24.self_attn.k_proj": true,
801
+ "model.layers.24.self_attn.v_proj": true,
802
+ "model.layers.24.self_attn.o_proj": true,
803
+ "model.layers.24.self_attn.q_norm": false,
804
+ "model.layers.24.self_attn.k_norm": false,
805
+ "model.layers.24.self_attn.rope": false,
806
+ "model.layers.24.mlp.gate_proj": true,
807
+ "model.layers.24.mlp.down_proj": true,
808
+ "model.layers.24.mlp.up_proj": true,
809
+ "model.layers.24.input_layernorm": false,
810
+ "model.layers.24.post_attention_layernorm": false,
811
+ "model.layers.25.self_attn.q_proj": true,
812
+ "model.layers.25.self_attn.k_proj": true,
813
+ "model.layers.25.self_attn.v_proj": true,
814
+ "model.layers.25.self_attn.o_proj": true,
815
+ "model.layers.25.self_attn.q_norm": false,
816
+ "model.layers.25.self_attn.k_norm": false,
817
+ "model.layers.25.self_attn.rope": false,
818
+ "model.layers.25.mlp.gate_proj": true,
819
+ "model.layers.25.mlp.down_proj": true,
820
+ "model.layers.25.mlp.up_proj": true,
821
+ "model.layers.25.input_layernorm": false,
822
+ "model.layers.25.post_attention_layernorm": false,
823
+ "model.layers.26.self_attn.q_proj": true,
824
+ "model.layers.26.self_attn.k_proj": true,
825
+ "model.layers.26.self_attn.v_proj": true,
826
+ "model.layers.26.self_attn.o_proj": true,
827
+ "model.layers.26.self_attn.q_norm": false,
828
+ "model.layers.26.self_attn.k_norm": false,
829
+ "model.layers.26.self_attn.rope": false,
830
+ "model.layers.26.mlp.gate_proj": true,
831
+ "model.layers.26.mlp.down_proj": true,
832
+ "model.layers.26.mlp.up_proj": true,
833
+ "model.layers.26.input_layernorm": false,
834
+ "model.layers.26.post_attention_layernorm": false,
835
+ "model.layers.27.self_attn.q_proj": true,
836
+ "model.layers.27.self_attn.k_proj": true,
837
+ "model.layers.27.self_attn.v_proj": true,
838
+ "model.layers.27.self_attn.o_proj": true,
839
+ "model.layers.27.self_attn.q_norm": false,
840
+ "model.layers.27.self_attn.k_norm": false,
841
+ "model.layers.27.self_attn.rope": false,
842
+ "model.layers.27.mlp.gate_proj": true,
843
+ "model.layers.27.mlp.down_proj": true,
844
+ "model.layers.27.mlp.up_proj": true,
845
+ "model.layers.27.input_layernorm": false,
846
+ "model.layers.27.post_attention_layernorm": false,
847
+ "model.layers.28.self_attn.q_proj": true,
848
+ "model.layers.28.self_attn.k_proj": true,
849
+ "model.layers.28.self_attn.v_proj": true,
850
+ "model.layers.28.self_attn.o_proj": true,
851
+ "model.layers.28.self_attn.q_norm": false,
852
+ "model.layers.28.self_attn.k_norm": false,
853
+ "model.layers.28.self_attn.rope": false,
854
+ "model.layers.28.mlp.gate_proj": true,
855
+ "model.layers.28.mlp.down_proj": true,
856
+ "model.layers.28.mlp.up_proj": true,
857
+ "model.layers.28.input_layernorm": false,
858
+ "model.layers.28.post_attention_layernorm": false,
859
+ "model.layers.29.self_attn.q_proj": true,
860
+ "model.layers.29.self_attn.k_proj": true,
861
+ "model.layers.29.self_attn.v_proj": true,
862
+ "model.layers.29.self_attn.o_proj": true,
863
+ "model.layers.29.self_attn.q_norm": false,
864
+ "model.layers.29.self_attn.k_norm": false,
865
+ "model.layers.29.self_attn.rope": false,
866
+ "model.layers.29.mlp.gate_proj": true,
867
+ "model.layers.29.mlp.down_proj": true,
868
+ "model.layers.29.mlp.up_proj": true,
869
+ "model.layers.29.input_layernorm": false,
870
+ "model.layers.29.post_attention_layernorm": false,
871
+ "model.layers.30.self_attn.q_proj": true,
872
+ "model.layers.30.self_attn.k_proj": true,
873
+ "model.layers.30.self_attn.v_proj": true,
874
+ "model.layers.30.self_attn.o_proj": true,
875
+ "model.layers.30.self_attn.q_norm": false,
876
+ "model.layers.30.self_attn.k_norm": false,
877
+ "model.layers.30.self_attn.rope": false,
878
+ "model.layers.30.mlp.gate_proj": true,
879
+ "model.layers.30.mlp.down_proj": true,
880
+ "model.layers.30.mlp.up_proj": true,
881
+ "model.layers.30.input_layernorm": false,
882
+ "model.layers.30.post_attention_layernorm": false,
883
+ "model.layers.31.self_attn.q_proj": true,
884
+ "model.layers.31.self_attn.k_proj": true,
885
+ "model.layers.31.self_attn.v_proj": true,
886
+ "model.layers.31.self_attn.o_proj": true,
887
+ "model.layers.31.self_attn.q_norm": false,
888
+ "model.layers.31.self_attn.k_norm": false,
889
+ "model.layers.31.self_attn.rope": false,
890
+ "model.layers.31.mlp.gate_proj": true,
891
+ "model.layers.31.mlp.down_proj": true,
892
+ "model.layers.31.mlp.up_proj": true,
893
+ "model.layers.31.input_layernorm": false,
894
+ "model.layers.31.post_attention_layernorm": false,
895
+ "model.layers.32.self_attn.q_proj": true,
896
+ "model.layers.32.self_attn.k_proj": true,
897
+ "model.layers.32.self_attn.v_proj": true,
898
+ "model.layers.32.self_attn.o_proj": true,
899
+ "model.layers.32.self_attn.q_norm": false,
900
+ "model.layers.32.self_attn.k_norm": false,
901
+ "model.layers.32.self_attn.rope": false,
902
+ "model.layers.32.mlp.gate_proj": true,
903
+ "model.layers.32.mlp.down_proj": true,
904
+ "model.layers.32.mlp.up_proj": true,
905
+ "model.layers.32.input_layernorm": false,
906
+ "model.layers.32.post_attention_layernorm": false,
907
+ "model.layers.33.self_attn.q_proj": true,
908
+ "model.layers.33.self_attn.k_proj": true,
909
+ "model.layers.33.self_attn.v_proj": true,
910
+ "model.layers.33.self_attn.o_proj": true,
911
+ "model.layers.33.self_attn.q_norm": false,
912
+ "model.layers.33.self_attn.k_norm": false,
913
+ "model.layers.33.self_attn.rope": false,
914
+ "model.layers.33.mlp.gate_proj": true,
915
+ "model.layers.33.mlp.down_proj": true,
916
+ "model.layers.33.mlp.up_proj": true,
917
+ "model.layers.33.input_layernorm": false,
918
+ "model.layers.33.post_attention_layernorm": false,
919
+ "model.layers.34.self_attn.q_proj": true,
920
+ "model.layers.34.self_attn.k_proj": true,
921
+ "model.layers.34.self_attn.v_proj": true,
922
+ "model.layers.34.self_attn.o_proj": true,
923
+ "model.layers.34.self_attn.q_norm": false,
924
+ "model.layers.34.self_attn.k_norm": false,
925
+ "model.layers.34.self_attn.rope": false,
926
+ "model.layers.34.mlp.gate_proj": true,
927
+ "model.layers.34.mlp.down_proj": true,
928
+ "model.layers.34.mlp.up_proj": true,
929
+ "model.layers.34.input_layernorm": false,
930
+ "model.layers.34.post_attention_layernorm": false,
931
+ "model.layers.35.self_attn.q_proj": true,
932
+ "model.layers.35.self_attn.k_proj": true,
933
+ "model.layers.35.self_attn.v_proj": true,
934
+ "model.layers.35.self_attn.o_proj": true,
935
+ "model.layers.35.self_attn.q_norm": false,
936
+ "model.layers.35.self_attn.k_norm": false,
937
+ "model.layers.35.self_attn.rope": false,
938
+ "model.layers.35.mlp.gate_proj": true,
939
+ "model.layers.35.mlp.down_proj": true,
940
+ "model.layers.35.mlp.up_proj": true,
941
+ "model.layers.35.input_layernorm": false,
942
+ "model.layers.35.post_attention_layernorm": false,
943
+ "model.layers.36.self_attn.q_proj": true,
944
+ "model.layers.36.self_attn.k_proj": true,
945
+ "model.layers.36.self_attn.v_proj": true,
946
+ "model.layers.36.self_attn.o_proj": true,
947
+ "model.layers.36.self_attn.q_norm": false,
948
+ "model.layers.36.self_attn.k_norm": false,
949
+ "model.layers.36.self_attn.rope": false,
950
+ "model.layers.36.mlp.gate_proj": true,
951
+ "model.layers.36.mlp.down_proj": true,
952
+ "model.layers.36.mlp.up_proj": true,
953
+ "model.layers.36.input_layernorm": false,
954
+ "model.layers.36.post_attention_layernorm": false,
955
+ "model.layers.37.self_attn.q_proj": true,
956
+ "model.layers.37.self_attn.k_proj": true,
957
+ "model.layers.37.self_attn.v_proj": true,
958
+ "model.layers.37.self_attn.o_proj": true,
959
+ "model.layers.37.self_attn.q_norm": false,
960
+ "model.layers.37.self_attn.k_norm": false,
961
+ "model.layers.37.self_attn.rope": false,
962
+ "model.layers.37.mlp.gate_proj": true,
963
+ "model.layers.37.mlp.down_proj": true,
964
+ "model.layers.37.mlp.up_proj": true,
965
+ "model.layers.37.input_layernorm": false,
966
+ "model.layers.37.post_attention_layernorm": false,
967
+ "model.layers.38.self_attn.q_proj": true,
968
+ "model.layers.38.self_attn.k_proj": true,
969
+ "model.layers.38.self_attn.v_proj": true,
970
+ "model.layers.38.self_attn.o_proj": true,
971
+ "model.layers.38.self_attn.q_norm": false,
972
+ "model.layers.38.self_attn.k_norm": false,
973
+ "model.layers.38.self_attn.rope": false,
974
+ "model.layers.38.mlp.gate_proj": true,
975
+ "model.layers.38.mlp.down_proj": true,
976
+ "model.layers.38.mlp.up_proj": true,
977
+ "model.layers.38.input_layernorm": false,
978
+ "model.layers.38.post_attention_layernorm": false,
979
+ "model.layers.39.self_attn.q_proj": true,
980
+ "model.layers.39.self_attn.k_proj": true,
981
+ "model.layers.39.self_attn.v_proj": true,
982
+ "model.layers.39.self_attn.o_proj": true,
983
+ "model.layers.39.self_attn.q_norm": false,
984
+ "model.layers.39.self_attn.k_norm": false,
985
+ "model.layers.39.self_attn.rope": false,
986
+ "model.layers.39.mlp.gate_proj": true,
987
+ "model.layers.39.mlp.down_proj": true,
988
+ "model.layers.39.mlp.up_proj": true,
989
+ "model.layers.39.input_layernorm": false,
990
+ "model.layers.39.post_attention_layernorm": false,
991
+ "model.norm": false,
992
+ "lm_head": true
993
  },
994
  "rms_norm_eps": 1e-06,
995
  "rope_scaling": null,
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": "4.51.0"
13
+ }
model-00001-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8de0be5f7c8e43083c94df53ec341ba44a622171129dc0af9bfbf211433c3b40
3
- size 5359897310
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6363a78368cd47e6e6c9b75a71e55ccbc37589725bfdbe5d2bfc47c630004a78
3
+ size 5359897354
model-00002-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2be14ff379c966f6ba4b50e19be7bdf4283e5c57b4d4619e2cd380bdb5b2a423
3
- size 1102015712
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:93993214f4e24d002f27d6b0393237e457a8511bb474cf7760f8d68f115338f1
3
+ size 1102015706
model.safetensors.index.json CHANGED
@@ -1,6 +1,7 @@
1
  {
2
  "metadata": {
3
- "total_size": 6461798400
 
4
  },
5
  "weight_map": {
6
  "lm_head.biases": "model-00002-of-00002.safetensors",
 
1
  {
2
  "metadata": {
3
+ "total_size": 6461798400,
4
+ "total_parameters": 14768307200
5
  },
6
  "weight_map": {
7
  "lm_head.biases": "model-00002-of-00002.safetensors",
tokenizer_config.json CHANGED
@@ -227,7 +227,6 @@
227
  "<|video_pad|>"
228
  ],
229
  "bos_token": null,
230
- "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 %}",
231
  "clean_up_tokenization_spaces": false,
232
  "eos_token": "<|im_end|>",
233
  "errors": "replace",
 
227
  "<|video_pad|>"
228
  ],
229
  "bos_token": null,
 
230
  "clean_up_tokenization_spaces": false,
231
  "eos_token": "<|im_end|>",
232
  "errors": "replace",