Safetensors
gpt_oss
austinxu87 commited on
Commit
bc24a32
·
verified ·
1 Parent(s): f05a5e3

Upload folder using huggingface_hub

Browse files
.gitattributes CHANGED
@@ -33,3 +33,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
 
 
33
  *.zip filter=lfs diff=lfs merge=lfs -text
34
  *.zst filter=lfs diff=lfs merge=lfs -text
35
  *tfevents* filter=lfs diff=lfs merge=lfs -text
36
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md CHANGED
@@ -1,3 +1,139 @@
1
  ---
2
  license: cc-by-nc-4.0
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-nc-4.0
3
  ---
4
+ # Foundational Automatic Evaluators: Scaling Multi-Task Generative Evaluator Training for Reasoning-Centric Domains
5
+ Paper: [arXiv link](https://dummy)
6
+
7
+ Authors: Austin Xu, Xuan-Phi Nguyen, Yilun Zhou, Chien-Sheng Wu, Caiming Xiong, Shafiq Joty
8
+
9
+ FARE-20B is a multi-task evaluator model finetuned from [gpt-oss-20B](https://huggingface.co/openai/gpt-oss-20b). It is trained on a large-scale multi-task, multi-domain data mixture using rejection-sampling SFT to perform the following evaluation tasks: Pairwise comparisons, step-level evaluation, reference-based verification, reference-free verification, and single-rating assessment.
10
+
11
+ # Usage
12
+ > [!IMPORTANT]
13
+ > The FARE family of evaluators has been trained with specific system and user prompt templates.
14
+
15
+ We provide examples below for two evaluation tasks: Pairwise comparisons and step-level error identification evaluation. For other tasks, we provide prompt templates in our paper (Appendix E).
16
+
17
+ ## Pairwise comparisons
18
+ ```
19
+ PROMPT_PAIRWISE_SYSTEM = """
20
+ Please act as an impartial judge and evaluate the quality of the responses provided by two AI assistants to the user prompt displayed below. You will be given assistant A's answer and assistant B's answer. Your job is to determine which assistant's answer is better.
21
+ If assistant A is better, output [A]. If assistant B is better, output [B].
22
+
23
+ Here are some rules for evaluation
24
+ (1) When evaluating the assistants' answers, identify any mistakes or inaccurate information. Focus on the content each response and select the response that is logically sound and error free.
25
+ (2) If both responses contain inaccurate information, select the response that arrives at the correct response
26
+ (3) Avoid any biases, such as order of responses, length, or stylistic elements like formatting
27
+
28
+ Before outputting your final judgment, provide an explanation of your judgment. Your explanation should discuss why your chosen response is better based on the evaluation criteria. The explanation should concretely discuss strengths and weaknesses of both answers.
29
+ After outputting your explanation, provide your final judgment. Use the following format:
30
+ Explanation: Your explanation here
31
+ Verdict: Your final verdict
32
+ """.strip()
33
+
34
+ PROMPT_PAIRWISE="""
35
+ [User Question]
36
+ {instruction}
37
+
38
+ [The Start of Assistant A's Answer]
39
+ {response_a}
40
+ [The End of Assistant A's Answer]
41
+
42
+ [The Start of Assistant B's Answer]
43
+ {response_b}
44
+ [The End of Assistant B's Answer]
45
+ """.strip()
46
+ ```
47
+
48
+ ## Step-level evaluation
49
+ ```
50
+ PROMPT_PROCESS_SYSTEM_ERROR_ID = """
51
+ Please act as an impartial judge and evaluate the quality of the response provided by an AI assistant to the user prompt displayed below. You will be given the assistant's solution to a math problem, which is split into steps, starting with a <step [step number]> tag, where [step number] is indexed from 0. Your job is to identify which step an error occurs, if an error is present.
52
+ When evaluating the solution, consider each step separately. Evaluate the content of each step for correctness. If you encounter a mistake at <step [step number]>, output [step number] as your Verdict. If the full response is error free, then select step number -1. Avoid any biases, such as length of step, or stylistic elements like formatting.
53
+
54
+ Here are some rules for evaluation.
55
+ (1) The assistant's answer does not need to be complete or arrive at a final solution. You may receive a partially complete response. Your job is to assess the quality of each step.
56
+ (2) When evaluating the assistant's answer, identify any mistakes or inaccurate information. Focus on the content each step and determine if the step is logically valid.
57
+ (3) For each step, you should provide an explanation of your assessment. If you find an error, describe the nature and cause of the error.
58
+ (4) Avoid any biases, such as answer length, or stylistic elements like formatting.
59
+
60
+ Before providing an your final verdict, think through the judging process and output your thoughts as an explanation
61
+ After providing your explanation, you must output the corresponding step number with an error. Use the following format:
62
+ Explanation: Your explanation here
63
+ Verdict: The step number with the error or -1 if no error occurs
64
+ """.strip()
65
+
66
+ PROMPT_SINGLE="""
67
+ [User Question]
68
+ {instruction}
69
+
70
+ [The Start of Assistant's Answer]
71
+ {response}
72
+ [The End of Assistant's Answer]
73
+ """.strip()
74
+ ```
75
+
76
+ ## Example inference with SGLang
77
+ For FARE-20B (gpt-oss variant), our evaluations were conducted with SGLang. We provide a minimal working example below with pairwise evaluation. For example usage with vLLM, see [FARE-8B](https://huggingface.co/Salesforce/FARE-8B)
78
+ ```
79
+ # instantiate model
80
+ from sglang.srt.entrypoints.engine import Engine
81
+ from transformers import AutoTokenizer
82
+ from prompts import PROMPT_PAIRWISE_SYSTEM, PROMPT_PAIRWISE # Prompt templates saved in a prompts.py file
83
+
84
+ llm = Engine(model_path="Salesforce/FARE-20B", tp_size=8, context_length=65536, trust_remote_code=True, mem_fraction_static=0.85)
85
+ tokenizer = AutoTokenizer.from_pretrained("Salesforce/FARE-20B", trust_remote_code=True)
86
+
87
+ # format data
88
+ data = [
89
+ {"question": "What is 5 + 10?", "response_a": "The answer is 15!", "response_b": "The answer is 16!"}
90
+ ]
91
+
92
+ formatted = [
93
+ PROMPT_PAIRWISE.format(
94
+ instruction = d["question"],
95
+ response_a = d["response_a"],
96
+ response_b = d["response_b"],
97
+ )
98
+ for d in data
99
+ ]
100
+
101
+ messages_lst = [
102
+ [{"role": "system", "content": PROMPT_PAIRWISE_SYSTEM}, {"role": "user", "content": user_formatted}]
103
+ for user_formatted in formatted
104
+ ]
105
+
106
+ prompts = [tokenizer.apply_chat_template(messages, add_generation_prompt=True, tokenize=False) for messages in messages_lst]
107
+
108
+ # inference!
109
+ sampling_params = {
110
+ "temperature": 0.0,
111
+ "max_new_tokens": 32768,
112
+ "top_p": 1.0,
113
+ "top_k": -1,
114
+ "skip_special_tokens": False, #Needed for GPT-OSS to output channel tags correctly; Otherwise gets rendered as `assistantfinal blah blah`
115
+ }
116
+
117
+ outputs = llm.generate(prompts, sampling_params)
118
+
119
+ # extract CoT tokens and assistant turn
120
+
121
+ for output in outputs:
122
+ text = output["text"]
123
+ cot_text, evaluator_text = text.split("<|end|><|start|>assistant<|channel|>final<|message|>")
124
+ cot_text = cot_text.split("<|channel|>analysis<|message|>")[-1]
125
+ ```
126
+
127
+
128
+ # Ethics disclaimer for Salesforce AI models, data, code
129
+ This release is for research purposes only in support of an academic paper. Our models, datasets, and code are not specifically designed or evaluated for all downstream purposes. We strongly recommend users evaluate and address potential concerns related to accuracy, safety, and fairness before deploying this model. We encourage users to consider the common limitations of AI, comply with applicable laws, and leverage best practices when selecting use cases, particularly for high-risk scenarios where errors or misuse could significantly impact people’s lives, rights, or safety. For further guidance on use cases, refer to our standard [AUP](https://www.salesforce.com/content/dam/web/en_us/www/documents/legal/Agreements/policies/ExternalFacing_Services_Policy.pdf) and [AI AUP](https://www.salesforce.com/content/dam/web/en_us/www/documents/legal/Agreements/policies/ai-acceptable-use-policy.pdf).
130
+
131
+ # Citation
132
+ ```
133
+ @misc{xu2025foundational,
134
+ title={Foundational Automatic Evaluators: Scaling Multi-Task Generative Evaluator Training for Reasoning-Centric Domains},
135
+ author={Xu, Austin and Nguyen, Xuan-Phi and Zhou, Yilun and Wu, Chien-Sheng and Xiong, Caiming and Joty, Shafiq},
136
+ year={2025},
137
+ journal={arXiv preprint arXiv:2510.},
138
+ }
139
+ ```
chat_template.jinja ADDED
@@ -0,0 +1,331 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {#-
2
+ In addition to the normal inputs of `messages` and `tools`, this template also accepts the
3
+ following kwargs:
4
+ - "builtin_tools": A list, can contain "browser" and/or "python".
5
+ - "model_identity": A string that optionally describes the model identity.
6
+ - "reasoning_effort": A string that describes the reasoning effort, defaults to "medium".
7
+ #}
8
+
9
+ {#- Tool Definition Rendering ============================================== #}
10
+ {%- macro render_typescript_type(param_spec, required_params, is_nullable=false) -%}
11
+ {%- if param_spec.type == "array" -%}
12
+ {%- if param_spec['items'] -%}
13
+ {%- if param_spec['items']['type'] == "string" -%}
14
+ {{- "string[]" }}
15
+ {%- elif param_spec['items']['type'] == "number" -%}
16
+ {{- "number[]" }}
17
+ {%- elif param_spec['items']['type'] == "integer" -%}
18
+ {{- "number[]" }}
19
+ {%- elif param_spec['items']['type'] == "boolean" -%}
20
+ {{- "boolean[]" }}
21
+ {%- else -%}
22
+ {%- set inner_type = render_typescript_type(param_spec['items'], required_params) -%}
23
+ {%- if inner_type == "object | object" or inner_type|length > 50 -%}
24
+ {{- "any[]" }}
25
+ {%- else -%}
26
+ {{- inner_type + "[]" }}
27
+ {%- endif -%}
28
+ {%- endif -%}
29
+ {%- if param_spec.nullable -%}
30
+ {{- " | null" }}
31
+ {%- endif -%}
32
+ {%- else -%}
33
+ {{- "any[]" }}
34
+ {%- if param_spec.nullable -%}
35
+ {{- " | null" }}
36
+ {%- endif -%}
37
+ {%- endif -%}
38
+ {%- elif param_spec.type is defined and param_spec.type is iterable and param_spec.type is not string and param_spec.type is not mapping and param_spec.type[0] is defined -%}
39
+ {#- Handle array of types like ["object", "object"] from Union[dict, list] #}
40
+ {%- if param_spec.type | length > 1 -%}
41
+ {{- param_spec.type | join(" | ") }}
42
+ {%- else -%}
43
+ {{- param_spec.type[0] }}
44
+ {%- endif -%}
45
+ {%- elif param_spec.oneOf -%}
46
+ {#- Handle oneOf schemas - check for complex unions and fallback to any #}
47
+ {%- set has_object_variants = false -%}
48
+ {%- for variant in param_spec.oneOf -%}
49
+ {%- if variant.type == "object" -%}
50
+ {%- set has_object_variants = true -%}
51
+ {%- endif -%}
52
+ {%- endfor -%}
53
+ {%- if has_object_variants and param_spec.oneOf|length > 1 -%}
54
+ {{- "any" }}
55
+ {%- else -%}
56
+ {%- for variant in param_spec.oneOf -%}
57
+ {{- render_typescript_type(variant, required_params) -}}
58
+ {%- if variant.description %}
59
+ {{- "// " + variant.description }}
60
+ {%- endif -%}
61
+ {%- if variant.default is defined %}
62
+ {{ "// default: " + variant.default|tojson }}
63
+ {%- endif -%}
64
+ {%- if not loop.last %}
65
+ {{- " | " }}
66
+ {% endif -%}
67
+ {%- endfor -%}
68
+ {%- endif -%}
69
+ {%- elif param_spec.type == "string" -%}
70
+ {%- if param_spec.enum -%}
71
+ {{- '"' + param_spec.enum|join('" | "') + '"' -}}
72
+ {%- else -%}
73
+ {{- "string" }}
74
+ {%- if param_spec.nullable %}
75
+ {{- " | null" }}
76
+ {%- endif -%}
77
+ {%- endif -%}
78
+ {%- elif param_spec.type == "number" -%}
79
+ {{- "number" }}
80
+ {%- elif param_spec.type == "integer" -%}
81
+ {{- "number" }}
82
+ {%- elif param_spec.type == "boolean" -%}
83
+ {{- "boolean" }}
84
+
85
+ {%- elif param_spec.type == "object" -%}
86
+ {%- if param_spec.properties -%}
87
+ {{- "{\n" }}
88
+ {%- for prop_name, prop_spec in param_spec.properties.items() -%}
89
+ {{- prop_name -}}
90
+ {%- if prop_name not in (param_spec.required or []) -%}
91
+ {{- "?" }}
92
+ {%- endif -%}
93
+ {{- ": " }}
94
+ {{ render_typescript_type(prop_spec, param_spec.required or []) }}
95
+ {%- if not loop.last -%}
96
+ {{-", " }}
97
+ {%- endif -%}
98
+ {%- endfor -%}
99
+ {{- "}" }}
100
+ {%- else -%}
101
+ {{- "object" }}
102
+ {%- endif -%}
103
+ {%- else -%}
104
+ {{- "any" }}
105
+ {%- endif -%}
106
+ {%- endmacro -%}
107
+
108
+ {%- macro render_tool_namespace(namespace_name, tools) -%}
109
+ {{- "## " + namespace_name + "\n\n" }}
110
+ {{- "namespace " + namespace_name + " {\n\n" }}
111
+ {%- for tool in tools %}
112
+ {%- set tool = tool.function %}
113
+ {{- "// " + tool.description + "\n" }}
114
+ {{- "type "+ tool.name + " = " }}
115
+ {%- if tool.parameters and tool.parameters.properties %}
116
+ {{- "(_: {\n" }}
117
+ {%- for param_name, param_spec in tool.parameters.properties.items() %}
118
+ {%- if param_spec.description %}
119
+ {{- "// " + param_spec.description + "\n" }}
120
+ {%- endif %}
121
+ {{- param_name }}
122
+ {%- if param_name not in (tool.parameters.required or []) -%}
123
+ {{- "?" }}
124
+ {%- endif -%}
125
+ {{- ": " }}
126
+ {{- render_typescript_type(param_spec, tool.parameters.required or []) }}
127
+ {%- if param_spec.default is defined -%}
128
+ {%- if param_spec.enum %}
129
+ {{- ", // default: " + param_spec.default }}
130
+ {%- elif param_spec.oneOf %}
131
+ {{- "// default: " + param_spec.default }}
132
+ {%- else %}
133
+ {{- ", // default: " + param_spec.default|tojson }}
134
+ {%- endif -%}
135
+ {%- endif -%}
136
+ {%- if not loop.last %}
137
+ {{- ",\n" }}
138
+ {%- else %}
139
+ {{- ",\n" }}
140
+ {%- endif -%}
141
+ {%- endfor %}
142
+ {{- "}) => any;\n\n" }}
143
+ {%- else -%}
144
+ {{- "() => any;\n\n" }}
145
+ {%- endif -%}
146
+ {%- endfor %}
147
+ {{- "} // namespace " + namespace_name }}
148
+ {%- endmacro -%}
149
+
150
+ {%- macro render_builtin_tools(browser_tool, python_tool) -%}
151
+ {%- if browser_tool %}
152
+ {{- "## browser\n\n" }}
153
+ {{- "// Tool for browsing.\n" }}
154
+ {{- "// The `cursor` appears in brackets before each browsing display: `[{cursor}]`.\n" }}
155
+ {{- "// Cite information from the tool using the following format:\n" }}
156
+ {{- "// `【{cursor}†L{line_start}(-L{line_end})?】`, for example: `【6†L9-L11】` or `【8†L3】`.\n" }}
157
+ {{- "// Do not quote more than 10 words directly from the tool output.\n" }}
158
+ {{- "// sources=web (default: web)\n" }}
159
+ {{- "namespace browser {\n\n" }}
160
+ {{- "// Searches for information related to `query` and displays `topn` results.\n" }}
161
+ {{- "type search = (_: {\n" }}
162
+ {{- "query: string,\n" }}
163
+ {{- "topn?: number, // default: 10\n" }}
164
+ {{- "source?: string,\n" }}
165
+ {{- "}) => any;\n\n" }}
166
+ {{- "// Opens the link `id` from the page indicated by `cursor` starting at line number `loc`, showing `num_lines` lines.\n" }}
167
+ {{- "// Valid link ids are displayed with the formatting: `【{id}†.*】`.\n" }}
168
+ {{- "// If `cursor` is not provided, the most recent page is implied.\n" }}
169
+ {{- "// If `id` is a string, it is treated as a fully qualified URL associated with `source`.\n" }}
170
+ {{- "// If `loc` is not provided, the viewport will be positioned at the beginning of the document or centered on the most relevant passage, if available.\n" }}
171
+ {{- "// Use this function without `id` to scroll to a new location of an opened page.\n" }}
172
+ {{- "type open = (_: {\n" }}
173
+ {{- "id?: number | string, // default: -1\n" }}
174
+ {{- "cursor?: number, // default: -1\n" }}
175
+ {{- "loc?: number, // default: -1\n" }}
176
+ {{- "num_lines?: number, // default: -1\n" }}
177
+ {{- "view_source?: boolean, // default: false\n" }}
178
+ {{- "source?: string,\n" }}
179
+ {{- "}) => any;\n\n" }}
180
+ {{- "// Finds exact matches of `pattern` in the current page, or the page given by `cursor`.\n" }}
181
+ {{- "type find = (_: {\n" }}
182
+ {{- "pattern: string,\n" }}
183
+ {{- "cursor?: number, // default: -1\n" }}
184
+ {{- "}) => any;\n\n" }}
185
+ {{- "} // namespace browser\n\n" }}
186
+ {%- endif -%}
187
+
188
+ {%- if python_tool %}
189
+ {{- "## python\n\n" }}
190
+ {{- "Use this tool to execute Python code in your chain of thought. The code will not be shown to the user. This tool should be used for internal reasoning, but not for code that is intended to be visible to the user (e.g. when creating plots, tables, or files).\n\n" }}
191
+ {{- "When you send a message containing Python code to python, it will be executed in a stateful Jupyter notebook environment. python will respond with the output of the execution or time out after 120.0 seconds. The drive at '/mnt/data' can be used to save and persist user files. Internet access for this session is UNKNOWN. Depends on the cluster.\n\n" }}
192
+ {%- endif -%}
193
+ {%- endmacro -%}
194
+
195
+ {#- System Message Construction ============================================ #}
196
+ {%- macro build_system_message() -%}
197
+ {%- if model_identity is not defined %}
198
+ {%- set model_identity = "You are ChatGPT, a large language model trained by OpenAI." %}
199
+ {%- endif %}
200
+ {{- model_identity + "\n" }}
201
+ {{- "Knowledge cutoff: 2024-06\n" }}
202
+ {{- "Current date: " + strftime_now("%Y-%m-%d") + "\n\n" }}
203
+ {%- if reasoning_effort is not defined %}
204
+ {%- set reasoning_effort = "low" %}
205
+ {%- endif %}
206
+ {{- "Reasoning: " + reasoning_effort + "\n\n" }}
207
+ {%- if builtin_tools %}
208
+ {{- "# Tools\n\n" }}
209
+ {%- set available_builtin_tools = namespace(browser=false, python=false) %}
210
+ {%- for tool in builtin_tools %}
211
+ {%- if tool == "browser" %}
212
+ {%- set available_builtin_tools.browser = true %}
213
+ {%- elif tool == "python" %}
214
+ {%- set available_builtin_tools.python = true %}
215
+ {%- endif %}
216
+ {%- endfor %}
217
+ {{- render_builtin_tools(available_builtin_tools.browser, available_builtin_tools.python) }}
218
+ {%- endif -%}
219
+ {{- "# Valid channels: analysis, commentary, final. Channel must be included for every message." }}
220
+ {%- if tools -%}
221
+ {{- "\nCalls to these tools must go to the commentary channel: 'functions'." }}
222
+ {%- endif -%}
223
+ {%- endmacro -%}
224
+
225
+ {#- Main Template Logic ================================================= #}
226
+ {#- Set defaults #}
227
+
228
+ {#- Render system message #}
229
+ {{- "<|start|>system<|message|>" }}
230
+ {{- build_system_message() }}
231
+ {{- "<|end|>" }}
232
+
233
+ {#- Extract developer message #}
234
+ {%- if messages[0].role == "developer" or messages[0].role == "system" %}
235
+ {%- set developer_message = messages[0].content %}
236
+ {%- set loop_messages = messages[1:] %}
237
+ {%- else %}
238
+ {%- set developer_message = "" %}
239
+ {%- set loop_messages = messages %}
240
+ {%- endif %}
241
+
242
+ {#- Render developer message #}
243
+ {%- if developer_message or tools %}
244
+ {{- "<|start|>developer<|message|>" }}
245
+ {%- if developer_message %}
246
+ {{- "# Instructions\n\n" }}
247
+ {{- developer_message }}
248
+ {{- "\n\n" }}
249
+ {%- endif %}
250
+ {%- if tools -%}
251
+ {{- "# Tools\n\n" }}
252
+ {{- render_tool_namespace("functions", tools) }}
253
+ {%- endif -%}
254
+ {{- "<|end|>" }}
255
+ {%- endif %}
256
+
257
+ {#- Render messages #}
258
+ {%- set last_tool_call = namespace(name=none) %}
259
+ {%- for message in loop_messages -%}
260
+ {#- At this point only assistant/user/tool messages should remain #}
261
+ {%- if message.role == 'assistant' -%}
262
+ {#- Checks to ensure the messages are being passed in the format we expect #}
263
+ {%- if "content" in message %}
264
+ {%- if "<|channel|>analysis<|message|>" in message.content or "<|channel|>final<|message|>" in message.content %}
265
+ {{- raise_exception("You have passed a message containing <|channel|> tags in the content field. Instead of doing this, you should pass analysis messages (the string between '<|message|>' and '<|end|>') in the 'thinking' field, and final messages (the string between '<|message|>' and '<|end|>') in the 'content' field.") }}
266
+ {%- endif %}
267
+ {%- endif %}
268
+ {%- if "thinking" in message %}
269
+ {%- if "<|channel|>analysis<|message|>" in message.thinking or "<|channel|>final<|message|>" in message.thinking %}
270
+ {{- raise_exception("You have passed a message containing <|channel|> tags in the thinking field. Instead of doing this, you should pass analysis messages (the string between '<|message|>' and '<|end|>') in the 'thinking' field, and final messages (the string between '<|message|>' and '<|end|>') in the 'content' field.") }}
271
+ {%- endif %}
272
+ {%- endif %}
273
+ {%- if "tool_calls" in message %}
274
+ {#- We need very careful handling here - we want to drop the tool call analysis message if the model #}
275
+ {#- has output a later <|final|> message, but otherwise we want to retain it. This is the only case #}
276
+ {#- when we render CoT/analysis messages in inference. #}
277
+ {%- set future_final_message = namespace(found=false) %}
278
+ {%- for future_message in loop_messages[loop.index:] %}
279
+ {%- if future_message.role == 'assistant' and "tool_calls" not in future_message %}
280
+ {%- set future_final_message.found = true %}
281
+ {%- endif %}
282
+ {%- endfor %}
283
+ {#- We assume max 1 tool call per message, and so we infer the tool call name #}
284
+ {#- in "tool" messages from the most recent assistant tool call name #}
285
+ {%- set tool_call = message.tool_calls[0] %}
286
+ {%- if tool_call.function %}
287
+ {%- set tool_call = tool_call.function %}
288
+ {%- endif %}
289
+ {%- if message.content and message.thinking %}
290
+ {{- raise_exception("Cannot pass both content and thinking in an assistant message with tool calls! Put the analysis message in one or the other, but not both.") }}
291
+ {%- elif message.content and not future_final_message.found %}
292
+ {{- "<|start|>assistant<|channel|>analysis<|message|>" + message.content + "<|end|>" }}
293
+ {%- elif message.thinking and not future_final_message.found %}
294
+ {{- "<|start|>assistant<|channel|>analysis<|message|>" + message.thinking + "<|end|>" }}
295
+ {%- endif %}
296
+ {{- "<|start|>assistant to=" }}
297
+ {{- "functions." + tool_call.name + "<|channel|>commentary " }}
298
+ {{- (tool_call.content_type if tool_call.content_type is defined else "json") + "<|message|>" }}
299
+ {{- tool_call.arguments|tojson }}
300
+ {{- "<|call|>" }}
301
+ {%- set last_tool_call.name = tool_call.name %}
302
+ {%- elif loop.last and not add_generation_prompt %}
303
+ {#- Only render the CoT if the final turn is an assistant turn and add_generation_prompt is false #}
304
+ {#- This is a situation that should only occur in training, never in inference. #}
305
+ {%- if "thinking" in message %}
306
+ {{- "<|start|>assistant<|channel|>analysis<|message|>" + message.thinking + "<|end|>" }}
307
+ {%- endif %}
308
+ {#- <|return|> indicates the end of generation, but <|end|> does not #}
309
+ {#- <|return|> should never be an input to the model, but we include it as the final token #}
310
+ {#- when training, so the model learns to emit it. #}
311
+ {{- "<|start|>assistant<|channel|>final<|message|>" + message.content + "<|return|>" }}
312
+ {%- else %}
313
+ {#- CoT is dropped during all previous turns, so we never render it for inference #}
314
+ {{- "<|start|>assistant<|channel|>final<|message|>" + message.content + "<|end|>" }}
315
+ {%- set last_tool_call.name = none %}
316
+ {%- endif %}
317
+ {%- elif message.role == 'tool' -%}
318
+ {%- if last_tool_call.name is none %}
319
+ {{- raise_exception("Message has tool role, but there was no previous assistant message with a tool call!") }}
320
+ {%- endif %}
321
+ {{- "<|start|>functions." + last_tool_call.name }}
322
+ {{- " to=assistant<|channel|>commentary<|message|>" + message.content|tojson + "<|end|>" }}
323
+ {%- elif message.role == 'user' -%}
324
+ {{- "<|start|>user<|message|>" + message.content + "<|end|>" }}
325
+ {%- endif -%}
326
+ {%- endfor -%}
327
+
328
+ {#- Generation prompt #}
329
+ {%- if add_generation_prompt -%}
330
+ <|start|>assistant
331
+ {%- endif -%}
config.json ADDED
@@ -0,0 +1,68 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GptOssForCausalLM"
4
+ ],
5
+ "attention_bias": true,
6
+ "attention_dropout": 0.0,
7
+ "eos_token_id": 200002,
8
+ "experts_per_token": 4,
9
+ "head_dim": 64,
10
+ "hidden_act": "silu",
11
+ "hidden_size": 2880,
12
+ "initial_context_length": 4096,
13
+ "initializer_range": 0.02,
14
+ "intermediate_size": 2880,
15
+ "layer_types": [
16
+ "sliding_attention",
17
+ "full_attention",
18
+ "sliding_attention",
19
+ "full_attention",
20
+ "sliding_attention",
21
+ "full_attention",
22
+ "sliding_attention",
23
+ "full_attention",
24
+ "sliding_attention",
25
+ "full_attention",
26
+ "sliding_attention",
27
+ "full_attention",
28
+ "sliding_attention",
29
+ "full_attention",
30
+ "sliding_attention",
31
+ "full_attention",
32
+ "sliding_attention",
33
+ "full_attention",
34
+ "sliding_attention",
35
+ "full_attention",
36
+ "sliding_attention",
37
+ "full_attention",
38
+ "sliding_attention",
39
+ "full_attention"
40
+ ],
41
+ "max_position_embeddings": 131072,
42
+ "model_type": "gpt_oss",
43
+ "num_attention_heads": 64,
44
+ "num_experts_per_tok": 4,
45
+ "num_hidden_layers": 24,
46
+ "num_key_value_heads": 8,
47
+ "num_local_experts": 32,
48
+ "output_router_logits": true,
49
+ "pad_token_id": 199999,
50
+ "rms_norm_eps": 1e-05,
51
+ "rope_scaling": {
52
+ "beta_fast": 32.0,
53
+ "beta_slow": 1.0,
54
+ "factor": 32.0,
55
+ "original_max_position_embeddings": 4096,
56
+ "rope_type": "yarn",
57
+ "truncate": false
58
+ },
59
+ "rope_theta": 150000,
60
+ "router_aux_loss_coef": 0.9,
61
+ "sliding_window": 128,
62
+ "swiglu_limit": 7.0,
63
+ "tie_word_embeddings": false,
64
+ "torch_dtype": "bfloat16",
65
+ "transformers_version": "4.55.0",
66
+ "use_cache": false,
67
+ "vocab_size": 201088
68
+ }
generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 199998,
3
+ "do_sample": true,
4
+ "eos_token_id": [
5
+ 200002,
6
+ 199999
7
+ ],
8
+ "pad_token_id": 199999,
9
+ "transformers_version": "4.55.0"
10
+ }
model-00001-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8a910800642d1bf94412b81f4c26f18f94e25f8e22ce4ae602228a6777abc5e8
3
+ size 4504304664
model-00002-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52e7fb820215dcf19ef2ed090c623721693f803b0a5b81c61abcb43e7b4530e9
3
+ size 4939127656
model-00003-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c40b92ecc1a0e36521d6f2f9f450e0ae6bebccb04e10c296cf9bbe60cc0a0c05
3
+ size 4939127656
model-00004-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1d8b30bf3c1196cec11db7b8f56b932949a7c35047b2c62d08bbb3d8b33d00bc
3
+ size 4939127680
model-00005-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7e6c358d6dcf59824f44104a1a8edec06814942a6209a3158d7612a617861950
3
+ size 4939127704
model-00006-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2218690deba44124c1f3f39263fbd9e85e56b5facad60e49ea46c5afed2a2717
3
+ size 4939127704
model-00007-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:72c872af98b10ec3aacc1c9b454bdede4342c80a023a090112a4dda548250554
3
+ size 4939127704
model-00008-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8c9956153bce7a8c884c29ef70adfbb04d9885e24a6ff837e89f0163b34c6c1a
3
+ size 4939127704
model-00009-of-00009.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:11ceccfde8c7df843274eb607aa7a2a66bd2721954d6d670f3b94ca0c3abdd97
3
+ size 2751362856
model.safetensors.index.json ADDED
@@ -0,0 +1,419 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 4759104,
4
+ "total_size": 41829514368
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00009-of-00009.safetensors",
8
+ "model.embed_tokens.weight": "model-00001-of-00009.safetensors",
9
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00009.safetensors",
10
+ "model.layers.0.mlp.experts.down_proj": "model-00001-of-00009.safetensors",
11
+ "model.layers.0.mlp.experts.down_proj_bias": "model-00001-of-00009.safetensors",
12
+ "model.layers.0.mlp.experts.gate_up_proj": "model-00001-of-00009.safetensors",
13
+ "model.layers.0.mlp.experts.gate_up_proj_bias": "model-00001-of-00009.safetensors",
14
+ "model.layers.0.mlp.router.bias": "model-00001-of-00009.safetensors",
15
+ "model.layers.0.mlp.router.weight": "model-00001-of-00009.safetensors",
16
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
17
+ "model.layers.0.self_attn.k_proj.bias": "model-00001-of-00009.safetensors",
18
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
19
+ "model.layers.0.self_attn.o_proj.bias": "model-00001-of-00009.safetensors",
20
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
21
+ "model.layers.0.self_attn.q_proj.bias": "model-00001-of-00009.safetensors",
22
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
23
+ "model.layers.0.self_attn.sinks": "model-00001-of-00009.safetensors",
24
+ "model.layers.0.self_attn.v_proj.bias": "model-00001-of-00009.safetensors",
25
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
26
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00009.safetensors",
27
+ "model.layers.1.mlp.experts.down_proj": "model-00001-of-00009.safetensors",
28
+ "model.layers.1.mlp.experts.down_proj_bias": "model-00001-of-00009.safetensors",
29
+ "model.layers.1.mlp.experts.gate_up_proj": "model-00001-of-00009.safetensors",
30
+ "model.layers.1.mlp.experts.gate_up_proj_bias": "model-00001-of-00009.safetensors",
31
+ "model.layers.1.mlp.router.bias": "model-00001-of-00009.safetensors",
32
+ "model.layers.1.mlp.router.weight": "model-00001-of-00009.safetensors",
33
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00009.safetensors",
34
+ "model.layers.1.self_attn.k_proj.bias": "model-00001-of-00009.safetensors",
35
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
36
+ "model.layers.1.self_attn.o_proj.bias": "model-00001-of-00009.safetensors",
37
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
38
+ "model.layers.1.self_attn.q_proj.bias": "model-00001-of-00009.safetensors",
39
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
40
+ "model.layers.1.self_attn.sinks": "model-00001-of-00009.safetensors",
41
+ "model.layers.1.self_attn.v_proj.bias": "model-00001-of-00009.safetensors",
42
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
43
+ "model.layers.10.input_layernorm.weight": "model-00004-of-00009.safetensors",
44
+ "model.layers.10.mlp.experts.down_proj": "model-00004-of-00009.safetensors",
45
+ "model.layers.10.mlp.experts.down_proj_bias": "model-00004-of-00009.safetensors",
46
+ "model.layers.10.mlp.experts.gate_up_proj": "model-00004-of-00009.safetensors",
47
+ "model.layers.10.mlp.experts.gate_up_proj_bias": "model-00004-of-00009.safetensors",
48
+ "model.layers.10.mlp.router.bias": "model-00004-of-00009.safetensors",
49
+ "model.layers.10.mlp.router.weight": "model-00004-of-00009.safetensors",
50
+ "model.layers.10.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
51
+ "model.layers.10.self_attn.k_proj.bias": "model-00004-of-00009.safetensors",
52
+ "model.layers.10.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
53
+ "model.layers.10.self_attn.o_proj.bias": "model-00004-of-00009.safetensors",
54
+ "model.layers.10.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
55
+ "model.layers.10.self_attn.q_proj.bias": "model-00004-of-00009.safetensors",
56
+ "model.layers.10.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
57
+ "model.layers.10.self_attn.sinks": "model-00004-of-00009.safetensors",
58
+ "model.layers.10.self_attn.v_proj.bias": "model-00004-of-00009.safetensors",
59
+ "model.layers.10.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
60
+ "model.layers.11.input_layernorm.weight": "model-00005-of-00009.safetensors",
61
+ "model.layers.11.mlp.experts.down_proj": "model-00005-of-00009.safetensors",
62
+ "model.layers.11.mlp.experts.down_proj_bias": "model-00005-of-00009.safetensors",
63
+ "model.layers.11.mlp.experts.gate_up_proj": "model-00005-of-00009.safetensors",
64
+ "model.layers.11.mlp.experts.gate_up_proj_bias": "model-00005-of-00009.safetensors",
65
+ "model.layers.11.mlp.router.bias": "model-00004-of-00009.safetensors",
66
+ "model.layers.11.mlp.router.weight": "model-00004-of-00009.safetensors",
67
+ "model.layers.11.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
68
+ "model.layers.11.self_attn.k_proj.bias": "model-00004-of-00009.safetensors",
69
+ "model.layers.11.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
70
+ "model.layers.11.self_attn.o_proj.bias": "model-00004-of-00009.safetensors",
71
+ "model.layers.11.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
72
+ "model.layers.11.self_attn.q_proj.bias": "model-00004-of-00009.safetensors",
73
+ "model.layers.11.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
74
+ "model.layers.11.self_attn.sinks": "model-00004-of-00009.safetensors",
75
+ "model.layers.11.self_attn.v_proj.bias": "model-00004-of-00009.safetensors",
76
+ "model.layers.11.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
77
+ "model.layers.12.input_layernorm.weight": "model-00005-of-00009.safetensors",
78
+ "model.layers.12.mlp.experts.down_proj": "model-00005-of-00009.safetensors",
79
+ "model.layers.12.mlp.experts.down_proj_bias": "model-00005-of-00009.safetensors",
80
+ "model.layers.12.mlp.experts.gate_up_proj": "model-00005-of-00009.safetensors",
81
+ "model.layers.12.mlp.experts.gate_up_proj_bias": "model-00005-of-00009.safetensors",
82
+ "model.layers.12.mlp.router.bias": "model-00005-of-00009.safetensors",
83
+ "model.layers.12.mlp.router.weight": "model-00005-of-00009.safetensors",
84
+ "model.layers.12.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
85
+ "model.layers.12.self_attn.k_proj.bias": "model-00005-of-00009.safetensors",
86
+ "model.layers.12.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
87
+ "model.layers.12.self_attn.o_proj.bias": "model-00005-of-00009.safetensors",
88
+ "model.layers.12.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
89
+ "model.layers.12.self_attn.q_proj.bias": "model-00005-of-00009.safetensors",
90
+ "model.layers.12.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
91
+ "model.layers.12.self_attn.sinks": "model-00005-of-00009.safetensors",
92
+ "model.layers.12.self_attn.v_proj.bias": "model-00005-of-00009.safetensors",
93
+ "model.layers.12.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
94
+ "model.layers.13.input_layernorm.weight": "model-00005-of-00009.safetensors",
95
+ "model.layers.13.mlp.experts.down_proj": "model-00005-of-00009.safetensors",
96
+ "model.layers.13.mlp.experts.down_proj_bias": "model-00005-of-00009.safetensors",
97
+ "model.layers.13.mlp.experts.gate_up_proj": "model-00005-of-00009.safetensors",
98
+ "model.layers.13.mlp.experts.gate_up_proj_bias": "model-00005-of-00009.safetensors",
99
+ "model.layers.13.mlp.router.bias": "model-00005-of-00009.safetensors",
100
+ "model.layers.13.mlp.router.weight": "model-00005-of-00009.safetensors",
101
+ "model.layers.13.post_attention_layernorm.weight": "model-00005-of-00009.safetensors",
102
+ "model.layers.13.self_attn.k_proj.bias": "model-00005-of-00009.safetensors",
103
+ "model.layers.13.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
104
+ "model.layers.13.self_attn.o_proj.bias": "model-00005-of-00009.safetensors",
105
+ "model.layers.13.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
106
+ "model.layers.13.self_attn.q_proj.bias": "model-00005-of-00009.safetensors",
107
+ "model.layers.13.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
108
+ "model.layers.13.self_attn.sinks": "model-00005-of-00009.safetensors",
109
+ "model.layers.13.self_attn.v_proj.bias": "model-00005-of-00009.safetensors",
110
+ "model.layers.13.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
111
+ "model.layers.14.input_layernorm.weight": "model-00006-of-00009.safetensors",
112
+ "model.layers.14.mlp.experts.down_proj": "model-00006-of-00009.safetensors",
113
+ "model.layers.14.mlp.experts.down_proj_bias": "model-00006-of-00009.safetensors",
114
+ "model.layers.14.mlp.experts.gate_up_proj": "model-00006-of-00009.safetensors",
115
+ "model.layers.14.mlp.experts.gate_up_proj_bias": "model-00006-of-00009.safetensors",
116
+ "model.layers.14.mlp.router.bias": "model-00005-of-00009.safetensors",
117
+ "model.layers.14.mlp.router.weight": "model-00005-of-00009.safetensors",
118
+ "model.layers.14.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
119
+ "model.layers.14.self_attn.k_proj.bias": "model-00005-of-00009.safetensors",
120
+ "model.layers.14.self_attn.k_proj.weight": "model-00005-of-00009.safetensors",
121
+ "model.layers.14.self_attn.o_proj.bias": "model-00005-of-00009.safetensors",
122
+ "model.layers.14.self_attn.o_proj.weight": "model-00005-of-00009.safetensors",
123
+ "model.layers.14.self_attn.q_proj.bias": "model-00005-of-00009.safetensors",
124
+ "model.layers.14.self_attn.q_proj.weight": "model-00005-of-00009.safetensors",
125
+ "model.layers.14.self_attn.sinks": "model-00005-of-00009.safetensors",
126
+ "model.layers.14.self_attn.v_proj.bias": "model-00005-of-00009.safetensors",
127
+ "model.layers.14.self_attn.v_proj.weight": "model-00005-of-00009.safetensors",
128
+ "model.layers.15.input_layernorm.weight": "model-00006-of-00009.safetensors",
129
+ "model.layers.15.mlp.experts.down_proj": "model-00006-of-00009.safetensors",
130
+ "model.layers.15.mlp.experts.down_proj_bias": "model-00006-of-00009.safetensors",
131
+ "model.layers.15.mlp.experts.gate_up_proj": "model-00006-of-00009.safetensors",
132
+ "model.layers.15.mlp.experts.gate_up_proj_bias": "model-00006-of-00009.safetensors",
133
+ "model.layers.15.mlp.router.bias": "model-00006-of-00009.safetensors",
134
+ "model.layers.15.mlp.router.weight": "model-00006-of-00009.safetensors",
135
+ "model.layers.15.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
136
+ "model.layers.15.self_attn.k_proj.bias": "model-00006-of-00009.safetensors",
137
+ "model.layers.15.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
138
+ "model.layers.15.self_attn.o_proj.bias": "model-00006-of-00009.safetensors",
139
+ "model.layers.15.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
140
+ "model.layers.15.self_attn.q_proj.bias": "model-00006-of-00009.safetensors",
141
+ "model.layers.15.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
142
+ "model.layers.15.self_attn.sinks": "model-00006-of-00009.safetensors",
143
+ "model.layers.15.self_attn.v_proj.bias": "model-00006-of-00009.safetensors",
144
+ "model.layers.15.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
145
+ "model.layers.16.input_layernorm.weight": "model-00006-of-00009.safetensors",
146
+ "model.layers.16.mlp.experts.down_proj": "model-00006-of-00009.safetensors",
147
+ "model.layers.16.mlp.experts.down_proj_bias": "model-00006-of-00009.safetensors",
148
+ "model.layers.16.mlp.experts.gate_up_proj": "model-00006-of-00009.safetensors",
149
+ "model.layers.16.mlp.experts.gate_up_proj_bias": "model-00006-of-00009.safetensors",
150
+ "model.layers.16.mlp.router.bias": "model-00006-of-00009.safetensors",
151
+ "model.layers.16.mlp.router.weight": "model-00006-of-00009.safetensors",
152
+ "model.layers.16.post_attention_layernorm.weight": "model-00006-of-00009.safetensors",
153
+ "model.layers.16.self_attn.k_proj.bias": "model-00006-of-00009.safetensors",
154
+ "model.layers.16.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
155
+ "model.layers.16.self_attn.o_proj.bias": "model-00006-of-00009.safetensors",
156
+ "model.layers.16.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
157
+ "model.layers.16.self_attn.q_proj.bias": "model-00006-of-00009.safetensors",
158
+ "model.layers.16.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
159
+ "model.layers.16.self_attn.sinks": "model-00006-of-00009.safetensors",
160
+ "model.layers.16.self_attn.v_proj.bias": "model-00006-of-00009.safetensors",
161
+ "model.layers.16.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
162
+ "model.layers.17.input_layernorm.weight": "model-00007-of-00009.safetensors",
163
+ "model.layers.17.mlp.experts.down_proj": "model-00007-of-00009.safetensors",
164
+ "model.layers.17.mlp.experts.down_proj_bias": "model-00007-of-00009.safetensors",
165
+ "model.layers.17.mlp.experts.gate_up_proj": "model-00007-of-00009.safetensors",
166
+ "model.layers.17.mlp.experts.gate_up_proj_bias": "model-00007-of-00009.safetensors",
167
+ "model.layers.17.mlp.router.bias": "model-00006-of-00009.safetensors",
168
+ "model.layers.17.mlp.router.weight": "model-00006-of-00009.safetensors",
169
+ "model.layers.17.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
170
+ "model.layers.17.self_attn.k_proj.bias": "model-00006-of-00009.safetensors",
171
+ "model.layers.17.self_attn.k_proj.weight": "model-00006-of-00009.safetensors",
172
+ "model.layers.17.self_attn.o_proj.bias": "model-00006-of-00009.safetensors",
173
+ "model.layers.17.self_attn.o_proj.weight": "model-00006-of-00009.safetensors",
174
+ "model.layers.17.self_attn.q_proj.bias": "model-00006-of-00009.safetensors",
175
+ "model.layers.17.self_attn.q_proj.weight": "model-00006-of-00009.safetensors",
176
+ "model.layers.17.self_attn.sinks": "model-00006-of-00009.safetensors",
177
+ "model.layers.17.self_attn.v_proj.bias": "model-00006-of-00009.safetensors",
178
+ "model.layers.17.self_attn.v_proj.weight": "model-00006-of-00009.safetensors",
179
+ "model.layers.18.input_layernorm.weight": "model-00007-of-00009.safetensors",
180
+ "model.layers.18.mlp.experts.down_proj": "model-00007-of-00009.safetensors",
181
+ "model.layers.18.mlp.experts.down_proj_bias": "model-00007-of-00009.safetensors",
182
+ "model.layers.18.mlp.experts.gate_up_proj": "model-00007-of-00009.safetensors",
183
+ "model.layers.18.mlp.experts.gate_up_proj_bias": "model-00007-of-00009.safetensors",
184
+ "model.layers.18.mlp.router.bias": "model-00007-of-00009.safetensors",
185
+ "model.layers.18.mlp.router.weight": "model-00007-of-00009.safetensors",
186
+ "model.layers.18.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
187
+ "model.layers.18.self_attn.k_proj.bias": "model-00007-of-00009.safetensors",
188
+ "model.layers.18.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
189
+ "model.layers.18.self_attn.o_proj.bias": "model-00007-of-00009.safetensors",
190
+ "model.layers.18.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
191
+ "model.layers.18.self_attn.q_proj.bias": "model-00007-of-00009.safetensors",
192
+ "model.layers.18.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
193
+ "model.layers.18.self_attn.sinks": "model-00007-of-00009.safetensors",
194
+ "model.layers.18.self_attn.v_proj.bias": "model-00007-of-00009.safetensors",
195
+ "model.layers.18.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
196
+ "model.layers.19.input_layernorm.weight": "model-00007-of-00009.safetensors",
197
+ "model.layers.19.mlp.experts.down_proj": "model-00007-of-00009.safetensors",
198
+ "model.layers.19.mlp.experts.down_proj_bias": "model-00007-of-00009.safetensors",
199
+ "model.layers.19.mlp.experts.gate_up_proj": "model-00007-of-00009.safetensors",
200
+ "model.layers.19.mlp.experts.gate_up_proj_bias": "model-00007-of-00009.safetensors",
201
+ "model.layers.19.mlp.router.bias": "model-00007-of-00009.safetensors",
202
+ "model.layers.19.mlp.router.weight": "model-00007-of-00009.safetensors",
203
+ "model.layers.19.post_attention_layernorm.weight": "model-00007-of-00009.safetensors",
204
+ "model.layers.19.self_attn.k_proj.bias": "model-00007-of-00009.safetensors",
205
+ "model.layers.19.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
206
+ "model.layers.19.self_attn.o_proj.bias": "model-00007-of-00009.safetensors",
207
+ "model.layers.19.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
208
+ "model.layers.19.self_attn.q_proj.bias": "model-00007-of-00009.safetensors",
209
+ "model.layers.19.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
210
+ "model.layers.19.self_attn.sinks": "model-00007-of-00009.safetensors",
211
+ "model.layers.19.self_attn.v_proj.bias": "model-00007-of-00009.safetensors",
212
+ "model.layers.19.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
213
+ "model.layers.2.input_layernorm.weight": "model-00002-of-00009.safetensors",
214
+ "model.layers.2.mlp.experts.down_proj": "model-00002-of-00009.safetensors",
215
+ "model.layers.2.mlp.experts.down_proj_bias": "model-00002-of-00009.safetensors",
216
+ "model.layers.2.mlp.experts.gate_up_proj": "model-00002-of-00009.safetensors",
217
+ "model.layers.2.mlp.experts.gate_up_proj_bias": "model-00002-of-00009.safetensors",
218
+ "model.layers.2.mlp.router.bias": "model-00001-of-00009.safetensors",
219
+ "model.layers.2.mlp.router.weight": "model-00001-of-00009.safetensors",
220
+ "model.layers.2.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
221
+ "model.layers.2.self_attn.k_proj.bias": "model-00001-of-00009.safetensors",
222
+ "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00009.safetensors",
223
+ "model.layers.2.self_attn.o_proj.bias": "model-00001-of-00009.safetensors",
224
+ "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00009.safetensors",
225
+ "model.layers.2.self_attn.q_proj.bias": "model-00001-of-00009.safetensors",
226
+ "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00009.safetensors",
227
+ "model.layers.2.self_attn.sinks": "model-00001-of-00009.safetensors",
228
+ "model.layers.2.self_attn.v_proj.bias": "model-00001-of-00009.safetensors",
229
+ "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00009.safetensors",
230
+ "model.layers.20.input_layernorm.weight": "model-00008-of-00009.safetensors",
231
+ "model.layers.20.mlp.experts.down_proj": "model-00008-of-00009.safetensors",
232
+ "model.layers.20.mlp.experts.down_proj_bias": "model-00008-of-00009.safetensors",
233
+ "model.layers.20.mlp.experts.gate_up_proj": "model-00008-of-00009.safetensors",
234
+ "model.layers.20.mlp.experts.gate_up_proj_bias": "model-00008-of-00009.safetensors",
235
+ "model.layers.20.mlp.router.bias": "model-00007-of-00009.safetensors",
236
+ "model.layers.20.mlp.router.weight": "model-00007-of-00009.safetensors",
237
+ "model.layers.20.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
238
+ "model.layers.20.self_attn.k_proj.bias": "model-00007-of-00009.safetensors",
239
+ "model.layers.20.self_attn.k_proj.weight": "model-00007-of-00009.safetensors",
240
+ "model.layers.20.self_attn.o_proj.bias": "model-00007-of-00009.safetensors",
241
+ "model.layers.20.self_attn.o_proj.weight": "model-00007-of-00009.safetensors",
242
+ "model.layers.20.self_attn.q_proj.bias": "model-00007-of-00009.safetensors",
243
+ "model.layers.20.self_attn.q_proj.weight": "model-00007-of-00009.safetensors",
244
+ "model.layers.20.self_attn.sinks": "model-00007-of-00009.safetensors",
245
+ "model.layers.20.self_attn.v_proj.bias": "model-00007-of-00009.safetensors",
246
+ "model.layers.20.self_attn.v_proj.weight": "model-00007-of-00009.safetensors",
247
+ "model.layers.21.input_layernorm.weight": "model-00008-of-00009.safetensors",
248
+ "model.layers.21.mlp.experts.down_proj": "model-00008-of-00009.safetensors",
249
+ "model.layers.21.mlp.experts.down_proj_bias": "model-00008-of-00009.safetensors",
250
+ "model.layers.21.mlp.experts.gate_up_proj": "model-00008-of-00009.safetensors",
251
+ "model.layers.21.mlp.experts.gate_up_proj_bias": "model-00008-of-00009.safetensors",
252
+ "model.layers.21.mlp.router.bias": "model-00008-of-00009.safetensors",
253
+ "model.layers.21.mlp.router.weight": "model-00008-of-00009.safetensors",
254
+ "model.layers.21.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
255
+ "model.layers.21.self_attn.k_proj.bias": "model-00008-of-00009.safetensors",
256
+ "model.layers.21.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
257
+ "model.layers.21.self_attn.o_proj.bias": "model-00008-of-00009.safetensors",
258
+ "model.layers.21.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
259
+ "model.layers.21.self_attn.q_proj.bias": "model-00008-of-00009.safetensors",
260
+ "model.layers.21.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
261
+ "model.layers.21.self_attn.sinks": "model-00008-of-00009.safetensors",
262
+ "model.layers.21.self_attn.v_proj.bias": "model-00008-of-00009.safetensors",
263
+ "model.layers.21.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
264
+ "model.layers.22.input_layernorm.weight": "model-00008-of-00009.safetensors",
265
+ "model.layers.22.mlp.experts.down_proj": "model-00008-of-00009.safetensors",
266
+ "model.layers.22.mlp.experts.down_proj_bias": "model-00008-of-00009.safetensors",
267
+ "model.layers.22.mlp.experts.gate_up_proj": "model-00008-of-00009.safetensors",
268
+ "model.layers.22.mlp.experts.gate_up_proj_bias": "model-00008-of-00009.safetensors",
269
+ "model.layers.22.mlp.router.bias": "model-00008-of-00009.safetensors",
270
+ "model.layers.22.mlp.router.weight": "model-00008-of-00009.safetensors",
271
+ "model.layers.22.post_attention_layernorm.weight": "model-00008-of-00009.safetensors",
272
+ "model.layers.22.self_attn.k_proj.bias": "model-00008-of-00009.safetensors",
273
+ "model.layers.22.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
274
+ "model.layers.22.self_attn.o_proj.bias": "model-00008-of-00009.safetensors",
275
+ "model.layers.22.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
276
+ "model.layers.22.self_attn.q_proj.bias": "model-00008-of-00009.safetensors",
277
+ "model.layers.22.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
278
+ "model.layers.22.self_attn.sinks": "model-00008-of-00009.safetensors",
279
+ "model.layers.22.self_attn.v_proj.bias": "model-00008-of-00009.safetensors",
280
+ "model.layers.22.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
281
+ "model.layers.23.input_layernorm.weight": "model-00009-of-00009.safetensors",
282
+ "model.layers.23.mlp.experts.down_proj": "model-00009-of-00009.safetensors",
283
+ "model.layers.23.mlp.experts.down_proj_bias": "model-00009-of-00009.safetensors",
284
+ "model.layers.23.mlp.experts.gate_up_proj": "model-00009-of-00009.safetensors",
285
+ "model.layers.23.mlp.experts.gate_up_proj_bias": "model-00009-of-00009.safetensors",
286
+ "model.layers.23.mlp.router.bias": "model-00008-of-00009.safetensors",
287
+ "model.layers.23.mlp.router.weight": "model-00008-of-00009.safetensors",
288
+ "model.layers.23.post_attention_layernorm.weight": "model-00009-of-00009.safetensors",
289
+ "model.layers.23.self_attn.k_proj.bias": "model-00008-of-00009.safetensors",
290
+ "model.layers.23.self_attn.k_proj.weight": "model-00008-of-00009.safetensors",
291
+ "model.layers.23.self_attn.o_proj.bias": "model-00008-of-00009.safetensors",
292
+ "model.layers.23.self_attn.o_proj.weight": "model-00008-of-00009.safetensors",
293
+ "model.layers.23.self_attn.q_proj.bias": "model-00008-of-00009.safetensors",
294
+ "model.layers.23.self_attn.q_proj.weight": "model-00008-of-00009.safetensors",
295
+ "model.layers.23.self_attn.sinks": "model-00008-of-00009.safetensors",
296
+ "model.layers.23.self_attn.v_proj.bias": "model-00008-of-00009.safetensors",
297
+ "model.layers.23.self_attn.v_proj.weight": "model-00008-of-00009.safetensors",
298
+ "model.layers.3.input_layernorm.weight": "model-00002-of-00009.safetensors",
299
+ "model.layers.3.mlp.experts.down_proj": "model-00002-of-00009.safetensors",
300
+ "model.layers.3.mlp.experts.down_proj_bias": "model-00002-of-00009.safetensors",
301
+ "model.layers.3.mlp.experts.gate_up_proj": "model-00002-of-00009.safetensors",
302
+ "model.layers.3.mlp.experts.gate_up_proj_bias": "model-00002-of-00009.safetensors",
303
+ "model.layers.3.mlp.router.bias": "model-00002-of-00009.safetensors",
304
+ "model.layers.3.mlp.router.weight": "model-00002-of-00009.safetensors",
305
+ "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
306
+ "model.layers.3.self_attn.k_proj.bias": "model-00002-of-00009.safetensors",
307
+ "model.layers.3.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
308
+ "model.layers.3.self_attn.o_proj.bias": "model-00002-of-00009.safetensors",
309
+ "model.layers.3.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
310
+ "model.layers.3.self_attn.q_proj.bias": "model-00002-of-00009.safetensors",
311
+ "model.layers.3.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
312
+ "model.layers.3.self_attn.sinks": "model-00002-of-00009.safetensors",
313
+ "model.layers.3.self_attn.v_proj.bias": "model-00002-of-00009.safetensors",
314
+ "model.layers.3.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
315
+ "model.layers.4.input_layernorm.weight": "model-00002-of-00009.safetensors",
316
+ "model.layers.4.mlp.experts.down_proj": "model-00002-of-00009.safetensors",
317
+ "model.layers.4.mlp.experts.down_proj_bias": "model-00002-of-00009.safetensors",
318
+ "model.layers.4.mlp.experts.gate_up_proj": "model-00002-of-00009.safetensors",
319
+ "model.layers.4.mlp.experts.gate_up_proj_bias": "model-00002-of-00009.safetensors",
320
+ "model.layers.4.mlp.router.bias": "model-00002-of-00009.safetensors",
321
+ "model.layers.4.mlp.router.weight": "model-00002-of-00009.safetensors",
322
+ "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00009.safetensors",
323
+ "model.layers.4.self_attn.k_proj.bias": "model-00002-of-00009.safetensors",
324
+ "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
325
+ "model.layers.4.self_attn.o_proj.bias": "model-00002-of-00009.safetensors",
326
+ "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
327
+ "model.layers.4.self_attn.q_proj.bias": "model-00002-of-00009.safetensors",
328
+ "model.layers.4.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
329
+ "model.layers.4.self_attn.sinks": "model-00002-of-00009.safetensors",
330
+ "model.layers.4.self_attn.v_proj.bias": "model-00002-of-00009.safetensors",
331
+ "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
332
+ "model.layers.5.input_layernorm.weight": "model-00003-of-00009.safetensors",
333
+ "model.layers.5.mlp.experts.down_proj": "model-00003-of-00009.safetensors",
334
+ "model.layers.5.mlp.experts.down_proj_bias": "model-00003-of-00009.safetensors",
335
+ "model.layers.5.mlp.experts.gate_up_proj": "model-00003-of-00009.safetensors",
336
+ "model.layers.5.mlp.experts.gate_up_proj_bias": "model-00003-of-00009.safetensors",
337
+ "model.layers.5.mlp.router.bias": "model-00002-of-00009.safetensors",
338
+ "model.layers.5.mlp.router.weight": "model-00002-of-00009.safetensors",
339
+ "model.layers.5.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
340
+ "model.layers.5.self_attn.k_proj.bias": "model-00002-of-00009.safetensors",
341
+ "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00009.safetensors",
342
+ "model.layers.5.self_attn.o_proj.bias": "model-00002-of-00009.safetensors",
343
+ "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00009.safetensors",
344
+ "model.layers.5.self_attn.q_proj.bias": "model-00002-of-00009.safetensors",
345
+ "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00009.safetensors",
346
+ "model.layers.5.self_attn.sinks": "model-00002-of-00009.safetensors",
347
+ "model.layers.5.self_attn.v_proj.bias": "model-00002-of-00009.safetensors",
348
+ "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00009.safetensors",
349
+ "model.layers.6.input_layernorm.weight": "model-00003-of-00009.safetensors",
350
+ "model.layers.6.mlp.experts.down_proj": "model-00003-of-00009.safetensors",
351
+ "model.layers.6.mlp.experts.down_proj_bias": "model-00003-of-00009.safetensors",
352
+ "model.layers.6.mlp.experts.gate_up_proj": "model-00003-of-00009.safetensors",
353
+ "model.layers.6.mlp.experts.gate_up_proj_bias": "model-00003-of-00009.safetensors",
354
+ "model.layers.6.mlp.router.bias": "model-00003-of-00009.safetensors",
355
+ "model.layers.6.mlp.router.weight": "model-00003-of-00009.safetensors",
356
+ "model.layers.6.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
357
+ "model.layers.6.self_attn.k_proj.bias": "model-00003-of-00009.safetensors",
358
+ "model.layers.6.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
359
+ "model.layers.6.self_attn.o_proj.bias": "model-00003-of-00009.safetensors",
360
+ "model.layers.6.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
361
+ "model.layers.6.self_attn.q_proj.bias": "model-00003-of-00009.safetensors",
362
+ "model.layers.6.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
363
+ "model.layers.6.self_attn.sinks": "model-00003-of-00009.safetensors",
364
+ "model.layers.6.self_attn.v_proj.bias": "model-00003-of-00009.safetensors",
365
+ "model.layers.6.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
366
+ "model.layers.7.input_layernorm.weight": "model-00003-of-00009.safetensors",
367
+ "model.layers.7.mlp.experts.down_proj": "model-00003-of-00009.safetensors",
368
+ "model.layers.7.mlp.experts.down_proj_bias": "model-00003-of-00009.safetensors",
369
+ "model.layers.7.mlp.experts.gate_up_proj": "model-00003-of-00009.safetensors",
370
+ "model.layers.7.mlp.experts.gate_up_proj_bias": "model-00003-of-00009.safetensors",
371
+ "model.layers.7.mlp.router.bias": "model-00003-of-00009.safetensors",
372
+ "model.layers.7.mlp.router.weight": "model-00003-of-00009.safetensors",
373
+ "model.layers.7.post_attention_layernorm.weight": "model-00003-of-00009.safetensors",
374
+ "model.layers.7.self_attn.k_proj.bias": "model-00003-of-00009.safetensors",
375
+ "model.layers.7.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
376
+ "model.layers.7.self_attn.o_proj.bias": "model-00003-of-00009.safetensors",
377
+ "model.layers.7.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
378
+ "model.layers.7.self_attn.q_proj.bias": "model-00003-of-00009.safetensors",
379
+ "model.layers.7.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
380
+ "model.layers.7.self_attn.sinks": "model-00003-of-00009.safetensors",
381
+ "model.layers.7.self_attn.v_proj.bias": "model-00003-of-00009.safetensors",
382
+ "model.layers.7.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
383
+ "model.layers.8.input_layernorm.weight": "model-00004-of-00009.safetensors",
384
+ "model.layers.8.mlp.experts.down_proj": "model-00004-of-00009.safetensors",
385
+ "model.layers.8.mlp.experts.down_proj_bias": "model-00004-of-00009.safetensors",
386
+ "model.layers.8.mlp.experts.gate_up_proj": "model-00004-of-00009.safetensors",
387
+ "model.layers.8.mlp.experts.gate_up_proj_bias": "model-00004-of-00009.safetensors",
388
+ "model.layers.8.mlp.router.bias": "model-00003-of-00009.safetensors",
389
+ "model.layers.8.mlp.router.weight": "model-00003-of-00009.safetensors",
390
+ "model.layers.8.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
391
+ "model.layers.8.self_attn.k_proj.bias": "model-00003-of-00009.safetensors",
392
+ "model.layers.8.self_attn.k_proj.weight": "model-00003-of-00009.safetensors",
393
+ "model.layers.8.self_attn.o_proj.bias": "model-00003-of-00009.safetensors",
394
+ "model.layers.8.self_attn.o_proj.weight": "model-00003-of-00009.safetensors",
395
+ "model.layers.8.self_attn.q_proj.bias": "model-00003-of-00009.safetensors",
396
+ "model.layers.8.self_attn.q_proj.weight": "model-00003-of-00009.safetensors",
397
+ "model.layers.8.self_attn.sinks": "model-00003-of-00009.safetensors",
398
+ "model.layers.8.self_attn.v_proj.bias": "model-00003-of-00009.safetensors",
399
+ "model.layers.8.self_attn.v_proj.weight": "model-00003-of-00009.safetensors",
400
+ "model.layers.9.input_layernorm.weight": "model-00004-of-00009.safetensors",
401
+ "model.layers.9.mlp.experts.down_proj": "model-00004-of-00009.safetensors",
402
+ "model.layers.9.mlp.experts.down_proj_bias": "model-00004-of-00009.safetensors",
403
+ "model.layers.9.mlp.experts.gate_up_proj": "model-00004-of-00009.safetensors",
404
+ "model.layers.9.mlp.experts.gate_up_proj_bias": "model-00004-of-00009.safetensors",
405
+ "model.layers.9.mlp.router.bias": "model-00004-of-00009.safetensors",
406
+ "model.layers.9.mlp.router.weight": "model-00004-of-00009.safetensors",
407
+ "model.layers.9.post_attention_layernorm.weight": "model-00004-of-00009.safetensors",
408
+ "model.layers.9.self_attn.k_proj.bias": "model-00004-of-00009.safetensors",
409
+ "model.layers.9.self_attn.k_proj.weight": "model-00004-of-00009.safetensors",
410
+ "model.layers.9.self_attn.o_proj.bias": "model-00004-of-00009.safetensors",
411
+ "model.layers.9.self_attn.o_proj.weight": "model-00004-of-00009.safetensors",
412
+ "model.layers.9.self_attn.q_proj.bias": "model-00004-of-00009.safetensors",
413
+ "model.layers.9.self_attn.q_proj.weight": "model-00004-of-00009.safetensors",
414
+ "model.layers.9.self_attn.sinks": "model-00004-of-00009.safetensors",
415
+ "model.layers.9.self_attn.v_proj.bias": "model-00004-of-00009.safetensors",
416
+ "model.layers.9.self_attn.v_proj.weight": "model-00004-of-00009.safetensors",
417
+ "model.norm.weight": "model-00009-of-00009.safetensors"
418
+ }
419
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,23 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<|startoftext|>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<|return|>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<|endoftext|>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ }
23
+ }
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f3ed4c75436e1a81847031f6f87da5716e6b3b5de3e9c7384380641fe7713024
3
+ size 27868274
tokenizer_config.json ADDED
@@ -0,0 +1,187 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "added_tokens_decoder": {
3
+ "199998": {
4
+ "content": "<|startoftext|>",
5
+ "lstrip": false,
6
+ "normalized": false,
7
+ "rstrip": false,
8
+ "single_word": false,
9
+ "special": true
10
+ },
11
+ "199999": {
12
+ "content": "<|endoftext|>",
13
+ "lstrip": false,
14
+ "normalized": false,
15
+ "rstrip": false,
16
+ "single_word": false,
17
+ "special": true
18
+ },
19
+ "200000": {
20
+ "content": "<|reserved_200000|>",
21
+ "lstrip": false,
22
+ "normalized": false,
23
+ "rstrip": false,
24
+ "single_word": false,
25
+ "special": true
26
+ },
27
+ "200001": {
28
+ "content": "<|reserved_200001|>",
29
+ "lstrip": false,
30
+ "normalized": false,
31
+ "rstrip": false,
32
+ "single_word": false,
33
+ "special": true
34
+ },
35
+ "200002": {
36
+ "content": "<|return|>",
37
+ "lstrip": false,
38
+ "normalized": false,
39
+ "rstrip": false,
40
+ "single_word": false,
41
+ "special": true
42
+ },
43
+ "200003": {
44
+ "content": "<|constrain|>",
45
+ "lstrip": false,
46
+ "normalized": false,
47
+ "rstrip": false,
48
+ "single_word": false,
49
+ "special": true
50
+ },
51
+ "200004": {
52
+ "content": "<|reserved_200004|>",
53
+ "lstrip": false,
54
+ "normalized": false,
55
+ "rstrip": false,
56
+ "single_word": false,
57
+ "special": true
58
+ },
59
+ "200005": {
60
+ "content": "<|channel|>",
61
+ "lstrip": false,
62
+ "normalized": false,
63
+ "rstrip": false,
64
+ "single_word": false,
65
+ "special": true
66
+ },
67
+ "200006": {
68
+ "content": "<|start|>",
69
+ "lstrip": false,
70
+ "normalized": false,
71
+ "rstrip": false,
72
+ "single_word": false,
73
+ "special": true
74
+ },
75
+ "200007": {
76
+ "content": "<|end|>",
77
+ "lstrip": false,
78
+ "normalized": false,
79
+ "rstrip": false,
80
+ "single_word": false,
81
+ "special": true
82
+ },
83
+ "200008": {
84
+ "content": "<|message|>",
85
+ "lstrip": false,
86
+ "normalized": false,
87
+ "rstrip": false,
88
+ "single_word": false,
89
+ "special": true
90
+ },
91
+ "200009": {
92
+ "content": "<|reserved_200009|>",
93
+ "lstrip": false,
94
+ "normalized": false,
95
+ "rstrip": false,
96
+ "single_word": false,
97
+ "special": true
98
+ },
99
+ "200010": {
100
+ "content": "<|reserved_200010|>",
101
+ "lstrip": false,
102
+ "normalized": false,
103
+ "rstrip": false,
104
+ "single_word": false,
105
+ "special": true
106
+ },
107
+ "200011": {
108
+ "content": "<|reserved_200011|>",
109
+ "lstrip": false,
110
+ "normalized": false,
111
+ "rstrip": false,
112
+ "single_word": false,
113
+ "special": true
114
+ },
115
+ "200012": {
116
+ "content": "<|call|>",
117
+ "lstrip": false,
118
+ "normalized": false,
119
+ "rstrip": false,
120
+ "single_word": false,
121
+ "special": true
122
+ },
123
+ "200013": {
124
+ "content": "<|reserved_200013|>",
125
+ "lstrip": false,
126
+ "normalized": false,
127
+ "rstrip": false,
128
+ "single_word": false,
129
+ "special": true
130
+ },
131
+ "200014": {
132
+ "content": "<|reserved_200014|>",
133
+ "lstrip": false,
134
+ "normalized": false,
135
+ "rstrip": false,
136
+ "single_word": false,
137
+ "special": true
138
+ },
139
+ "200015": {
140
+ "content": "<|reserved_200015|>",
141
+ "lstrip": false,
142
+ "normalized": false,
143
+ "rstrip": false,
144
+ "single_word": false,
145
+ "special": true
146
+ },
147
+ "200016": {
148
+ "content": "<|reserved_200016|>",
149
+ "lstrip": false,
150
+ "normalized": false,
151
+ "rstrip": false,
152
+ "single_word": false,
153
+ "special": true
154
+ },
155
+ "200017": {
156
+ "content": "<|reserved_200017|>",
157
+ "lstrip": false,
158
+ "normalized": false,
159
+ "rstrip": false,
160
+ "single_word": false,
161
+ "special": true
162
+ },
163
+ "200018": {
164
+ "content": "<|endofprompt|>",
165
+ "lstrip": false,
166
+ "normalized": false,
167
+ "rstrip": false,
168
+ "single_word": false,
169
+ "special": true
170
+ }
171
+ },
172
+ "bos_token": "<|startoftext|>",
173
+ "clean_up_tokenization_spaces": false,
174
+ "eos_token": "<|return|>",
175
+ "extra_special_tokens": {},
176
+ "max_length": 32768,
177
+ "model_input_names": [
178
+ "input_ids",
179
+ "attention_mask"
180
+ ],
181
+ "model_max_length": 1000000000000000019884624838656,
182
+ "pad_token": "<|endoftext|>",
183
+ "stride": 0,
184
+ "tokenizer_class": "PreTrainedTokenizerFast",
185
+ "truncation_side": "right",
186
+ "truncation_strategy": "longest_first"
187
+ }