danielhanchen commited on
Commit
f887516
·
verified ·
1 Parent(s): 1bdee9e

Upload folder using huggingface_hub

Browse files
README.md ADDED
@@ -0,0 +1,640 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ base_model:
3
+ - ibm-granite/granite-4.0-micro
4
+ license: apache-2.0
5
+ library_name: transformers
6
+ tags:
7
+ - language
8
+ - unsloth
9
+ - granite-4.0
10
+ ---
11
+ <div>
12
+ <p style="margin-top: 0;margin-bottom: 0;">
13
+ <em><a href="https://docs.unsloth.ai/basics/unsloth-dynamic-v2.0-gguf">Unsloth Dynamic 2.0</a> achieves superior accuracy & outperforms other leading quants.</em>
14
+ </p>
15
+ <div style="display: flex; gap: 5px; align-items: center; ">
16
+ <a href="https://github.com/unslothai/unsloth/">
17
+ <img src="https://github.com/unslothai/unsloth/raw/main/images/unsloth%20new%20logo.png" width="133">
18
+ </a>
19
+ <a href="https://discord.gg/unsloth">
20
+ <img src="https://github.com/unslothai/unsloth/raw/main/images/Discord%20button.png" width="173">
21
+ </a>
22
+ <a href="https://docs.unsloth.ai/">
23
+ <img src="https://raw.githubusercontent.com/unslothai/unsloth/refs/heads/main/images/documentation%20green%20button.png" width="143">
24
+ </a>
25
+ </div>
26
+ </div>
27
+
28
+
29
+ # Granite-4.0-Micro
30
+
31
+ **Model Summary:**
32
+ Granite-4.0-Micro is a 3B parameter long-context instruct model finetuned from *Granite-4.0-Micro-Base* using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets. This model is developed using a diverse set of techniques with a structured chat format, including supervised finetuning, model alignment using reinforcement learning, and model merging. Granite 4.0 instruct models feature improved *instruction following (IF)* and *tool-calling* capabilities, making them more effective in enterprise applications.
33
+
34
+ - **Developers:** Granite Team, IBM
35
+ - **HF Collection:** [Granite 4.0 Language Models HF Collection](https://huggingface.co/collections/ibm-granite/granite-40-language-models-6811a18b820ef362d9e5a82c)
36
+ - **GitHub Repository:** [ibm-granite/granite-4.0-language-models](https://github.com/ibm-granite/granite-4.0-language-models)
37
+ - **Website**: [Granite Docs](https://www.ibm.com/granite/docs/)
38
+ - **Release Date**: October 2nd, 2025
39
+ - **License:** [Apache 2.0](https://www.apache.org/licenses/LICENSE-2.0)
40
+
41
+ **Supported Languages:**
42
+ English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, and Chinese. Users may finetune Granite 4.0 models for languages beyond these languages.
43
+
44
+ **Intended use:**
45
+ The model is designed to follow general instructions and can serve as the foundation for AI assistants across diverse domains, including business applications, as well as for LLM agents equipped with tool-use capabilities.
46
+
47
+ *Capabilities*
48
+ * Summarization
49
+ * Text classification
50
+ * Text extraction
51
+ * Question-answering
52
+ * Retrieval Augmented Generation (RAG)
53
+ * Code related tasks
54
+ * Function-calling tasks
55
+ * Multilingual dialog use cases
56
+ * Fill-In-the-Middle (FIM) code completions
57
+
58
+ <!-- <todo>Need to test the examples. (especially the tool calling and RAG ones)</todo>
59
+ -->
60
+
61
+ **Generation:**
62
+ This is a simple example of how to use Granite-4.0-Micro model.
63
+
64
+ Install the following libraries:
65
+
66
+ ```shell
67
+ pip install torch torchvision torchaudio
68
+ pip install accelerate
69
+ pip install transformers
70
+ ```
71
+ Then, copy the snippet from the section that is relevant for your use case.
72
+
73
+ ```python
74
+ import torch
75
+ from transformers import AutoModelForCausalLM, AutoTokenizer
76
+
77
+ device = "cuda"
78
+ model_path = "ibm-granite/granite-4.0-micro"
79
+ tokenizer = AutoTokenizer.from_pretrained(model_path)
80
+ # drop device_map if running on CPU
81
+ model = AutoModelForCausalLM.from_pretrained(model_path, device_map=device)
82
+ model.eval()
83
+ # change input text as desired
84
+ chat = [
85
+ { "role": "user", "content": "Please list one IBM Research laboratory located in the United States. You should only output its name and location." },
86
+ ]
87
+ chat = tokenizer.apply_chat_template(chat, tokenize=False, add_generation_prompt=True)
88
+ # tokenize the text
89
+ input_tokens = tokenizer(chat, return_tensors="pt").to(device)
90
+ # generate output tokens
91
+ output = model.generate(**input_tokens,
92
+ max_new_tokens=100)
93
+ # decode output tokens into text
94
+ output = tokenizer.batch_decode(output)
95
+ # print output
96
+ print(output[0])
97
+ ```
98
+
99
+ Expected output:
100
+ ```shell
101
+ <|start_of_role|>user<|end_of_role|>Please list one IBM Research laboratory located in the United States. You should only output its name and location.<|end_of_text|>
102
+ <|start_of_role|>assistant<|end_of_role|>Almaden Research Center, San Jose, California<|end_of_text|>
103
+ ```
104
+
105
+ **Tool-calling:**
106
+ Granite-4.0-Micro comes with enhanced tool calling capabilities, enabling seamless integration with external functions and APIs. To define a list of tools please follow OpenAI's function [definition schema](https://platform.openai.com/docs/guides/function-calling?api-mode=responses#defining-functions).
107
+
108
+ This is an example of how to use Granite-4.0-Micro model tool-calling ability:
109
+
110
+ ```python
111
+ import torch
112
+ from transformers import AutoModelForCausalLM, AutoTokenizer
113
+
114
+ device = "cuda"
115
+ model_path = "ibm-granite/granite-4.0-micro"
116
+ tokenizer = AutoTokenizer.from_pretrained(model_path)
117
+ # drop device_map if running on CPU
118
+ model = AutoModelForCausalLM.from_pretrained(model_path, device_map=device)
119
+ model.eval()
120
+
121
+ tools = [
122
+ {
123
+ "type": "function",
124
+ "function": {
125
+ "name": "get_current_weather",
126
+ "description": "Get the current weather for a specified city.",
127
+ "parameters": {
128
+ "type": "object",
129
+ "properties": {
130
+ "city": {
131
+ "type": "string",
132
+ "description": "Name of the city"
133
+ }
134
+ },
135
+ "required": ["city"]
136
+ }
137
+ }
138
+ }
139
+ ]
140
+
141
+ # change input text as desired
142
+ chat = [
143
+ { "role": "user", "content": "What's the weather like in Boston right now?" },
144
+ ]
145
+ chat = tokenizer.apply_chat_template(chat, \
146
+ tokenize=False, \
147
+ tools=tools, \
148
+ add_generation_prompt=True)
149
+ # tokenize the text
150
+ input_tokens = tokenizer(chat, return_tensors="pt").to(device)
151
+ # generate output tokens
152
+ output = model.generate(**input_tokens,
153
+ max_new_tokens=100)
154
+ # decode output tokens into text
155
+ output = tokenizer.batch_decode(output)
156
+ # print output
157
+ print(output[0])
158
+ ```
159
+
160
+ Expected output:
161
+ ```shell
162
+ <|start_of_role|>system<|end_of_role|>You are a helpful assistant with access to the following tools. You may call one or more tools to assist with the user query.
163
+
164
+ You are provided with function signatures within <tools></tools> XML tags:
165
+ - <tools>
166
+ - unsloth
167
+ {"type": "function", "function": {"name": "get_current_weather", "description": "Get the current weather for a specified city.", "parameters": {"type": "object", "properties": {"city": {"type": "string", "description": "Name of the city"}}, "required": ["city"]}}}
168
+ </tools>
169
+
170
+ For each tool call, return a json object with function name and arguments within <tool_call></tool_call> XML tags:
171
+ - <tool_call>
172
+ - unsloth
173
+ {"name": <function-name>, "arguments": <args-json-object>}
174
+ </tool_call>. If a tool does not exist in the provided list of tools, notify the user that you do not have the ability to fulfill the request.<|end_of_text|>
175
+ <|start_of_role|>user<|end_of_role|>What's the weather like in Boston right now?<|end_of_text|>
176
+ <|start_of_role|>assistant<|end_of_role|><tool_call>
177
+ {"name": "get_current_weather", "arguments": {"city": "Boston"}}
178
+ </tool_call><|end_of_text|>
179
+ ```
180
+
181
+ <!-- **Retrieval Augmented Generation:**
182
+ *Coming soon* -->
183
+
184
+ **Evaluation Results:**
185
+
186
+ <table>
187
+ <!-- <caption><b> All Results</b></caption> -->
188
+ <thead>
189
+ <tr>
190
+ <th style="text-align:left; background-color: #001d6c; color: white;">Benchmarks</th>
191
+ <th style="text-align:left; background-color: #001d6c; color: white;">Metric</th>
192
+ <th style="text-align:center; background-color: #001d6c; color: white;">Micro Dense</th>
193
+ <th style="text-align:center; background-color: #001d6c; color: white;">H Micro Dense</th>
194
+ <th style="text-align:center; background-color: #001d6c; color: white;">H Tiny MoE</th>
195
+ <th style="text-align:center; background-color: #001d6c; color: white;">H Small MoE</th>
196
+ </tr>
197
+ </thead>
198
+ <tbody>
199
+ <tr>
200
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
201
+ General Tasks
202
+ </td>
203
+ </tr>
204
+ <tr>
205
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MMLU</td>
206
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">5-shot</td>
207
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">65.98</td>
208
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">67.43</td>
209
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">68.65</td>
210
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">78.44</td>
211
+ </tr>
212
+ <tr>
213
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MMLU-Pro</td>
214
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">5-shot, CoT</td>
215
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">44.5</td>
216
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">43.48</td>
217
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">44.94</td>
218
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">55.47</td>
219
+ </tr>
220
+ <tr>
221
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">BBH</td>
222
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">3-shot, CoT</td>
223
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">72.48</td>
224
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">69.36</td>
225
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">66.34</td>
226
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">81.62</td>
227
+ </tr>
228
+ <tr>
229
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">AGI EVAL</td>
230
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">0-shot, CoT</td>
231
+ <td style="text-align:right; background-color:#DAE8FF; color: #2D2D2D;">64.29</td>
232
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">59</td>
233
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">62.15</td>
234
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">70.63</td>
235
+ </tr>
236
+ <tr>
237
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">GPQA</td>
238
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">0-shot, CoT</td>
239
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">30.14</td>
240
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">32.15</td>
241
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">32.59</td>
242
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">40.63</td>
243
+ </tr>
244
+ <tr>
245
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
246
+ Alignment Tasks
247
+ </td>
248
+ </tr>
249
+ <tr>
250
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">AlpacaEval 2.0</td>
251
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;"></td>
252
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">29.49</td>
253
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">31.49</td>
254
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">30.61</td>
255
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">42.48</td>
256
+ </tr>
257
+ <tr>
258
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">IFEval</td>
259
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">Instruct, Strict</td>
260
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">85.5</td>
261
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">86.94</td>
262
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">84.78</td>
263
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">89.87</td>
264
+ </tr>
265
+ <tr>
266
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">IFEval</td>
267
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">Prompt, Strict</td>
268
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">79.12</td>
269
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">81.71</td>
270
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">78.1</td>
271
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">85.22</td>
272
+ </tr>
273
+ <tr>
274
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">IFEval</td>
275
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">Average</td>
276
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">82.31</td>
277
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">84.32</td>
278
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">81.44</td>
279
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">87.55</td>
280
+ </tr>
281
+ <tr>
282
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">ArenaHard</td>
283
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;"></td>
284
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">25.84</td>
285
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">36.15</td>
286
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">35.75</td>
287
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">46.48</td>
288
+ </tr>
289
+ <tr>
290
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
291
+ Math Tasks
292
+ </td>
293
+ </tr>
294
+ <tr>
295
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">GSM8K</td>
296
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">8-shot</td>
297
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">85.45</td>
298
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">81.35</td>
299
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">84.69</td>
300
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">87.27</td>
301
+ </tr>
302
+ <tr>
303
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">GSM8K Symbolic</td>
304
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">8-shot</td>
305
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">79.82</td>
306
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">77.5</td>
307
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">81.1</td>
308
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">87.38</td>
309
+ </tr>
310
+ <tr>
311
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">Minerva Math</td>
312
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">0-shot, CoT</td>
313
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">62.06</td>
314
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">66.44</td>
315
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">69.64</td>
316
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">74</td>
317
+ </tr>
318
+ <tr>
319
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">DeepMind Math</td>
320
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">0-shot, CoT</td>
321
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">44.56</td>
322
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">43.83</td>
323
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">49.92</td>
324
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">59.33</td>
325
+ </tr>
326
+ <tr>
327
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
328
+ Code Tasks
329
+ </td>
330
+ </tr>
331
+ <tr>
332
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">HumanEval</td>
333
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
334
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">80</td>
335
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">81</td>
336
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">83</td>
337
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">88</td>
338
+ </tr>
339
+ <tr>
340
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">HumanEval+</td>
341
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
342
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">72</td>
343
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">75</td>
344
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">76</td>
345
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">83</td>
346
+ </tr>
347
+ <tr>
348
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MBPP</td>
349
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
350
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">72</td>
351
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">73</td>
352
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">80</td>
353
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">84</td>
354
+ </tr>
355
+ <tr>
356
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MBPP+</td>
357
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
358
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">64</td>
359
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">64</td>
360
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">69</td>
361
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">71</td>
362
+ </tr>
363
+ <tr>
364
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">CRUXEval-O</td>
365
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
366
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">41.5</td>
367
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">41.25</td>
368
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">39.63</td>
369
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">50.25</td>
370
+ </tr>
371
+ <tr>
372
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">BigCodeBench</td>
373
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
374
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">39.21</td>
375
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">37.9</td>
376
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">41.06</td>
377
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">46.23</td>
378
+ </tr>
379
+ <tr>
380
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
381
+ Tool Calling Tasks
382
+ </td>
383
+ </tr>
384
+ <tr>
385
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">BFCL v3</td>
386
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;"></td>
387
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">59.98</td>
388
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">57.56</td>
389
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">57.65</td>
390
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">64.69</td>
391
+ </tr>
392
+ <tr>
393
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
394
+ Multilingual Tasks
395
+ </td>
396
+ </tr>
397
+ <tr>
398
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MULTIPLE</td>
399
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">pass@1</td>
400
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">49.21</td>
401
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">49.46</td>
402
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">55.83</td>
403
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">57.37</td>
404
+ </tr>
405
+ <tr>
406
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MMMLU</td>
407
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">5-shot</td>
408
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">55.14</td>
409
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">55.19</td>
410
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">61.87</td>
411
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">69.69</td>
412
+ </tr>
413
+ <tr>
414
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">INCLUDE</td>
415
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">5-shot</td>
416
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">51.62</td>
417
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">50.51</td>
418
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">53.12</td>
419
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">63.97</td>
420
+ </tr>
421
+ <tr>
422
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MGSM</td>
423
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">8-shot</td>
424
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">28.56</td>
425
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">44.48</td>
426
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">45.36</td>
427
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">38.72</td>
428
+ </tr>
429
+ <tr>
430
+ <td colspan="6" style="text-align:center; background-color: #FFFFFF; color: #2D2D2D; font-style:italic;">
431
+ Safety
432
+ </td>
433
+ </tr>
434
+ <tr>
435
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">SALAD-Bench</td>
436
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;"></td>
437
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">97.06</td>
438
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">96.28</td>
439
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">97.77</td>
440
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">97.3</td>
441
+ </tr>
442
+ <tr>
443
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">AttaQ</td>
444
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;"></td>
445
+ <td style="text-align:right; background-color: #DAE8FF; color: #2D2D2D;">86.05</td>
446
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">84.44</td>
447
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">86.61</td>
448
+ <td style="text-align:right; background-color: #FFFFFF; color: #2D2D2D;">86.64</td>
449
+ </tr>
450
+ </tbody></table>
451
+
452
+
453
+ <table>
454
+ <caption><b>Multilingual Benchmarks and thr included languages:</b></caption>
455
+ <thead>
456
+ <tr>
457
+ <th style="text-align:left; background-color: #001d6c; color: white;">Benchmarks</th>
458
+ <th style="text-align:left; background-color: #001d6c; color: white;"># Langs</th>
459
+ <th style="text-align:center; background-color: #001d6c; color: white;">Languages</th>
460
+ </tr>
461
+ </thead>
462
+ <tbody>
463
+ <tr>
464
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MMMLU</td>
465
+ <td style="text-align:center; background-color: #FFFFFF; color: #2D2D2D;">11</td>
466
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">ar, de, en, es, fr, ja, ko, pt, zh, bn, hi</td>
467
+ </tr>
468
+ <tr>
469
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">INCLUDE</td>
470
+ <td style="text-align:center; background-color: #FFFFFF; color: #2D2D2D;">14</td>
471
+ <!-- <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">hindi, bengali, tamil, telugu, arabic, german, spanish, french, italian, japanese, korean, dutch, portuguese, chinese</td> -->
472
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">hi, bn, ta, te, ar, de, es, fr, it, ja, ko, nl, pt, zh</td>
473
+
474
+ </tr>
475
+ <tr>
476
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">MGSM</td>
477
+ <td style="text-align:center; background-color: #FFFFFF; color: #2D2D2D;">5</td>
478
+ <td style="text-align:left; background-color: #FFFFFF; color: #2D2D2D;">en, es, fr, ja, zh</td>
479
+ </tr>
480
+ </tbody>
481
+ </table>
482
+
483
+ **Model Architecture:**
484
+
485
+ Granite-4.0-Micro baseline is built on a decoder-only dense transformer architecture. Core components of this architecture are: GQA, RoPE, MLP with SwiGLU, RMSNorm, and shared input/output embeddings.
486
+
487
+ <table>
488
+ <thead>
489
+ <tr>
490
+ <th style="text-align:left; background-color: #001d6c; color: white;">Model</th>
491
+ <th style="text-align:center; background-color: #001d6c; color: white;">Micro Dense</th>
492
+ <th style="text-align:center; background-color: #001d6c; color: white;">H Micro Dense</th>
493
+ <th style="text-align:center; background-color: #001d6c; color: white;">H Tiny MoE</th>
494
+ <th style="text-align:center; background-color: #001d6c; color: white;">H Small MoE</th>
495
+ </tr></thead>
496
+ <tbody>
497
+ <tr>
498
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Embedding size</td>
499
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">2560</td>
500
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">2048</td>
501
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">1536</td>
502
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">4096</td>
503
+ </tr>
504
+ <tr>
505
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Number of layers</td>
506
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">40 attention</td>
507
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">4 attention / 36 Mamba2</td>
508
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">4 attention / 36 Mamba2</td>
509
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">4 attention / 36 Mamba2</td>
510
+ </tr>
511
+ <tr>
512
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Attention head size</td>
513
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">64</td>
514
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">64</td>
515
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128</td>
516
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128</td>
517
+ </tr>
518
+ <tr>
519
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Number of attention heads</td>
520
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">40</td>
521
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">32</td>
522
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">12</td>
523
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">32</td>
524
+ </tr>
525
+ <tr>
526
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Number of KV heads</td>
527
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">8</td>
528
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">8</td>
529
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">4</td>
530
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">8</td>
531
+ </tr>
532
+ <tr>
533
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Mamba2 state size</td>
534
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">-</td>
535
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128</td>
536
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128</td>
537
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128</td>
538
+ </tr>
539
+ <tr>
540
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Number of Mamba2 heads</td>
541
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">-</td>
542
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">64</td>
543
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">48</td>
544
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128</td>
545
+ </tr>
546
+
547
+ <tr>
548
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">MLP / Shared expert hidden size</td>
549
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">8192</td>
550
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">8192</td>
551
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">1024</td>
552
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">1536</td>
553
+ </tr>
554
+
555
+
556
+ <tr>
557
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Num. Experts</td>
558
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">-</td>
559
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">-</td>
560
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">64</td>
561
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">72</td>
562
+ </tr>
563
+ <tr>
564
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Num. active Experts</td>
565
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">-</td>
566
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">-</td>
567
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">6</td>
568
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">10</td>
569
+ </tr>
570
+ <tr>
571
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Expert hidden size</td>
572
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">-</td>
573
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">-</td>
574
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">512</td>
575
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">768</td>
576
+ </tr>
577
+
578
+ <tr>
579
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">MLP activation</td>
580
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">SwiGLU</td>
581
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">SwiGLU</td>
582
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">SwiGLU</td>
583
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">SwiGLU</td>
584
+ </tr>
585
+
586
+ <tr>
587
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Sequence length</td>
588
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">128K</td>
589
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128K</td>
590
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128K</td>
591
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">128K</td>
592
+ </tr>
593
+ <tr>
594
+ <td style="text-align:left; background-color: #FFFFFF; color: black;">Position embedding</td>
595
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">RoPE</td>
596
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">NoPE</td>
597
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">NoPE</td>
598
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">NoPE</td>
599
+ </tr>
600
+ <tr>
601
+ <td style="text-align:left; background-color: #FFFFFF; color: black;"># Parameters</td>
602
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">3B</td>
603
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">3B</td>
604
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">7B</td>
605
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">32B</td>
606
+ </tr>
607
+ <tr>
608
+ <td style="text-align:left; background-color: #FFFFFF; color: black;"># Active parameters</td>
609
+ <td style="text-align:center; background-color: #DAE8FF; color: black;">3B</td>
610
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">3B</td>
611
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">1B</td>
612
+ <td style="text-align:center; background-color: #FFFFFF; color: black;">9B</td>
613
+ </tr>
614
+ </tbody></table>
615
+
616
+ **Training Data:**
617
+ Overall, our SFT data is largely comprised of three key sources: (1) publicly available datasets with permissive license, (2) internal synthetic data targeting specific capabilities, and (3) a select set of human-curated data.
618
+
619
+ **Infrastructure:**
620
+ We trained the Granite 4.0 Language Models utilizing an NVIDIA GB200 NVL72 cluster hosted in CoreWeave. Intra-rack communication occurs via the 72-GPU NVLink domain, and a non-blocking, full Fat-Tree NDR 400 Gb/s InfiniBand network provides inter-rack communication. This cluster provides a scalable and efficient infrastructure for training our models over thousands of GPUs.
621
+
622
+ **Ethical Considerations and Limitations:**
623
+ Granite 4.0 Instruction Models are primarily finetuned using instruction-response pairs mostly in English, but also multilingual data covering multiple languages. Although this model can handle multilingual dialog use cases, its performance might not be similar to English tasks. In such case, introducing a small number of examples (few-shot) can help the model in generating more accurate outputs. While this model has been aligned by keeping safety in consideration, the model may in some cases produce inaccurate, biased, or unsafe responses to user prompts. So we urge the community to use this model with proper safety testing and tuning tailored for their specific tasks.
624
+
625
+ **Resources**
626
+ - ⭐️ Learn about the latest updates with Granite: https://www.ibm.com/granite
627
+ - 📄 Get started with tutorials, best practices, and prompt engineering advice: https://www.ibm.com/granite/docs/
628
+ - 💡 Learn about the latest Granite learning resources: https://ibm.biz/granite-learning-resources
629
+
630
+ <!-- ## Citation
631
+ ```
632
+ @misc{granite-models,
633
+ author = {author 1, author2, ...},
634
+ title = {},
635
+ journal = {},
636
+ volume = {},
637
+ year = {2024},
638
+ url = {https://arxiv.org/abs/0000.00000},
639
+ }
640
+ ``` -->
chat_template.jinja ADDED
@@ -0,0 +1,114 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {%- set tools_system_message_prefix = 'You are a helpful assistant with access to the following tools. You may call one or more tools to assist with the user query.\n\nYou are provided with function signatures within <tools></tools> XML tags:\n<tools>' %}
2
+ {%- set tools_system_message_suffix = '\n</tools>\n\nFor each tool 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>. If a tool does not exist in the provided list of tools, notify the user that you do not have the ability to fulfill the request.' %}
3
+ {%- set documents_system_message_prefix = 'You are a helpful assistant with access to the following documents. You may use one or more documents to assist with the user query.\n\nYou are given a list of documents within <documents></documents> XML tags:\n<documents>' %}
4
+ {%- set documents_system_message_suffix = '\n</documents>\n\nWrite the response to the user\'s input by strictly aligning with the facts in the provided documents. If the information needed to answer the question is not available in the documents, inform the user that the question cannot be answered based on the available data.' %}
5
+ {%- if available_tools is defined and available_tools %}
6
+ {%- set tools = available_tools %}
7
+ {%- endif %}
8
+ {%- set ns = namespace(tools_system_message=tools_system_message_prefix,
9
+ documents_system_message=documents_system_message_prefix,
10
+ system_message=''
11
+ ) %}
12
+ {%- if tools %}
13
+ {%- for tool in tools %}
14
+ {%- set ns.tools_system_message = ns.tools_system_message + '\n' + (tool | tojson) %}
15
+ {%- endfor %}
16
+ {%- set ns.tools_system_message = ns.tools_system_message + tools_system_message_suffix %}
17
+ {%- else %}
18
+ {%- set ns.tools_system_message = '' %}
19
+ {%- endif %}
20
+ {%- if documents %}
21
+ {%- for document in documents %}
22
+ {%- set ns.documents_system_message = ns.documents_system_message + '\n' + (document | tojson) %}
23
+ {%- endfor %}
24
+ {%- set ns.documents_system_message = ns.documents_system_message + documents_system_message_suffix %}
25
+ {%- else %}
26
+ {%- set ns.documents_system_message = '' %}
27
+ {%- endif %}
28
+ {%- if messages[0].role == 'system' %}
29
+ {%- if messages[0].content is string %}
30
+ {%- set ns.system_message = messages[0].content %}
31
+ {%- elif messages[0].content is iterable %}
32
+ {%- for entry in messages[0].content %}
33
+ {%- if entry.type== 'text' %}
34
+ {%- if ns.system_message != '' %}
35
+ {%- set ns.system_message = ns.system_message + '\n' %}
36
+ {%- endif %}
37
+ {%- set ns.system_message = ns.system_message + entry.text %}
38
+ {%- endif %}
39
+ {%- endfor %}
40
+ {%- endif %}
41
+ {%- if tools and documents %}
42
+ {%- set ns.system_message = ns.system_message + '\n\n' + ns.tools_system_message + '\n\n' + ns.documents_system_message %}
43
+ {%- elif tools %}
44
+ {%- set ns.system_message = ns.system_message + '\n\n' + ns.tools_system_message %}
45
+ {%- elif documents %}
46
+ {%- set ns.system_message = ns.system_message + '\n\n' + ns.documents_system_message %}
47
+ {%- endif %}
48
+ {%- else %}
49
+ {%- if tools and documents %}
50
+ {%- set ns.system_message = ns.tools_system_message + '\n\n' + ns.documents_system_message %}
51
+ {%- elif tools %}
52
+ {%- set ns.system_message = ns.tools_system_message %}
53
+ {%- elif documents %}
54
+ {%- set ns.system_message = ns.documents_system_message %}
55
+ {%- endif %}
56
+ {%- endif %}
57
+ {%- if ns.system_message %}
58
+ {{- '<|start_of_role|>system<|end_of_role|>' + ns.system_message + '<|end_of_text|>\n' }}
59
+ {%- endif %}
60
+ {%- for message in messages %}
61
+ {%- set content = namespace(val='') %}
62
+ {%- if message.content is string %}
63
+ {%- set content.val = message.content %}
64
+ {%- else %}
65
+ {%- if message.content is iterable %}
66
+ {%- for entry in message.content %}
67
+ {%- if entry.type== 'text' %}
68
+ {%- if content.val != '' %}
69
+ {%- set content.val = content.val + '\n' %}
70
+ {%- endif %}
71
+ {%- set content.val = content.val + entry.text %}
72
+ {%- endif %}
73
+ {%- endfor %}
74
+ {%- endif %}
75
+ {%- endif %}
76
+ {%- if (message.role == 'user') or (message.role == 'system' and not loop.first) %}
77
+ {{- '<|start_of_role|>' + message.role + '<|end_of_role|>' + content.val + '<|end_of_text|>\n' }}
78
+ {%- elif message.role == 'assistant' %}
79
+ {{- '<|start_of_role|>' + message.role + '<|end_of_role|>' + content.val }}
80
+ {%- if message.tool_calls %}
81
+ {%- for tool_call in message.tool_calls %}
82
+ {%- if (loop.first and content.val) or (not loop.first) %}
83
+ {{- '\n' }}
84
+ {%- endif %}
85
+ {%- if tool_call.function %}
86
+ {%- set tool_call = tool_call.function %}
87
+ {%- endif %}
88
+ {{- '<tool_call>\n{"name": "' }}
89
+ {{- tool_call.name }}
90
+ {{- '", "arguments": ' }}
91
+ {%- if tool_call.arguments is string %}
92
+ {{- tool_call.arguments }}
93
+ {%- else %}
94
+ {{- tool_call.arguments | tojson }}
95
+ {%- endif %}
96
+ {{- '}\n</tool_call>' }}
97
+ {%- endfor %}
98
+ {%- endif %}
99
+ {{- '<|end_of_text|>\n' }}
100
+ {%- elif message.role == 'tool' %}
101
+ {%- if loop.first or (messages[loop.index0 - 1].role != 'tool') %}
102
+ {{- '<|start_of_role|>user<|end_of_role|>' }}
103
+ {%- endif %}
104
+ {{- '\n<tool_response>\n' }}
105
+ {{- content.val }}
106
+ {{- '\n</tool_response>' }}
107
+ {%- if loop.last or (messages[loop.index0 + 1].role != 'tool') %}
108
+ {{- '<|end_of_text|>\n' }}
109
+ {%- endif %}
110
+ {%- endif %}
111
+ {%- endfor %}
112
+ {%- if add_generation_prompt %}
113
+ {{- '<|start_of_role|>assistant<|end_of_role|>' }}
114
+ {%- endif %}
config.json ADDED
@@ -0,0 +1,130 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "GraniteMoeHybridForCausalLM"
4
+ ],
5
+ "attention_bias": false,
6
+ "attention_dropout": 0.0,
7
+ "attention_multiplier": 0.015625,
8
+ "bos_token_id": 100257,
9
+ "torch_dtype": "bfloat16",
10
+ "embedding_multiplier": 12,
11
+ "eos_token_id": 100257,
12
+ "hidden_act": "silu",
13
+ "hidden_size": 2560,
14
+ "initializer_range": 0.1,
15
+ "intermediate_size": 8192,
16
+ "layer_types": [
17
+ "attention",
18
+ "attention",
19
+ "attention",
20
+ "attention",
21
+ "attention",
22
+ "attention",
23
+ "attention",
24
+ "attention",
25
+ "attention",
26
+ "attention",
27
+ "attention",
28
+ "attention",
29
+ "attention",
30
+ "attention",
31
+ "attention",
32
+ "attention",
33
+ "attention",
34
+ "attention",
35
+ "attention",
36
+ "attention",
37
+ "attention",
38
+ "attention",
39
+ "attention",
40
+ "attention",
41
+ "attention",
42
+ "attention",
43
+ "attention",
44
+ "attention",
45
+ "attention",
46
+ "attention",
47
+ "attention",
48
+ "attention",
49
+ "attention",
50
+ "attention",
51
+ "attention",
52
+ "attention",
53
+ "attention",
54
+ "attention",
55
+ "attention",
56
+ "attention"
57
+ ],
58
+ "logits_scaling": 10,
59
+ "mamba_chunk_size": 256,
60
+ "mamba_conv_bias": true,
61
+ "mamba_d_conv": 4,
62
+ "mamba_d_head": 40,
63
+ "mamba_d_state": 256,
64
+ "mamba_expand": 2,
65
+ "mamba_n_groups": 1,
66
+ "mamba_n_heads": 128,
67
+ "mamba_proj_bias": false,
68
+ "max_position_embeddings": 131072,
69
+ "model_type": "granitemoehybrid",
70
+ "normalization_function": "rmsnorm",
71
+ "num_attention_heads": 40,
72
+ "num_experts_per_tok": 0,
73
+ "num_hidden_layers": 40,
74
+ "num_key_value_heads": 8,
75
+ "num_local_experts": 0,
76
+ "output_router_logits": false,
77
+ "pad_token_id": 100256,
78
+ "position_embedding_type": "rope",
79
+ "quantization_config": {
80
+ "_load_in_4bit": true,
81
+ "_load_in_8bit": false,
82
+ "bnb_4bit_compute_dtype": "bfloat16",
83
+ "bnb_4bit_quant_storage": "uint8",
84
+ "bnb_4bit_quant_type": "nf4",
85
+ "bnb_4bit_use_double_quant": true,
86
+ "llm_int8_enable_fp32_cpu_offload": false,
87
+ "llm_int8_has_fp16_weight": false,
88
+ "llm_int8_skip_modules": [
89
+ "embed_tokens",
90
+ "embedding",
91
+ "lm_head",
92
+ "multi_modal_projector",
93
+ "merger",
94
+ "modality_projection",
95
+ "router",
96
+ "visual",
97
+ "vision_tower",
98
+ "model.layers.16.self_attn",
99
+ "model.layers.38.shared_mlp",
100
+ "model.layers.35.shared_mlp",
101
+ "model.layers.33.shared_mlp",
102
+ "model.layers.17.self_attn",
103
+ "model.layers.3.shared_mlp",
104
+ "model.layers.34.shared_mlp",
105
+ "model.layers.6.self_attn",
106
+ "model.layers.0.self_attn",
107
+ "model.layers.2.shared_mlp",
108
+ "model.layers.2.self_attn",
109
+ "model.layers.6.shared_mlp",
110
+ "model.layers.39.shared_mlp",
111
+ "model.layers.0.shared_mlp",
112
+ "model.layers.1.self_attn"
113
+ ],
114
+ "llm_int8_threshold": 6.0,
115
+ "load_in_4bit": true,
116
+ "load_in_8bit": false,
117
+ "quant_method": "bitsandbytes"
118
+ },
119
+ "residual_multiplier": 0.22,
120
+ "rms_norm_eps": 1e-05,
121
+ "rope_scaling": null,
122
+ "rope_theta": 10000000,
123
+ "router_aux_loss_coef": 0.01,
124
+ "shared_intermediate_size": 8192,
125
+ "tie_word_embeddings": true,
126
+ "transformers_version": "4.56.2",
127
+ "unsloth_fixed": true,
128
+ "use_cache": true,
129
+ "vocab_size": 100352
130
+ }
generation_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 100257,
4
+ "eos_token_id": 100257,
5
+ "max_length": 131072,
6
+ "pad_token_id": 100256,
7
+ "transformers_version": "4.56.2"
8
+ }
merges.txt ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ad1c9b2f74ff5c4ad58306fe27300095ea3a2e9d7663d3d792d009f31bb4983d
3
+ size 3117831948
special_tokens_map.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token": {
3
+ "content": "<|end_of_text|>",
4
+ "lstrip": false,
5
+ "normalized": false,
6
+ "rstrip": false,
7
+ "single_word": false
8
+ },
9
+ "eos_token": {
10
+ "content": "<|end_of_text|>",
11
+ "lstrip": false,
12
+ "normalized": false,
13
+ "rstrip": false,
14
+ "single_word": false
15
+ },
16
+ "pad_token": {
17
+ "content": "<|pad|>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<|unk|>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer_config.json ADDED
@@ -0,0 +1,784 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": false,
3
+ "add_prefix_space": false,
4
+ "added_tokens_decoder": {
5
+ "100256": {
6
+ "content": "<|pad|>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "100257": {
14
+ "content": "<|end_of_text|>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "100258": {
22
+ "content": "<|fim_prefix|>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": false
28
+ },
29
+ "100259": {
30
+ "content": "<|fim_middle|>",
31
+ "lstrip": false,
32
+ "normalized": false,
33
+ "rstrip": false,
34
+ "single_word": false,
35
+ "special": false
36
+ },
37
+ "100260": {
38
+ "content": "<|fim_suffix|>",
39
+ "lstrip": false,
40
+ "normalized": false,
41
+ "rstrip": false,
42
+ "single_word": false,
43
+ "special": false
44
+ },
45
+ "100261": {
46
+ "content": "<|fim_pad|>",
47
+ "lstrip": false,
48
+ "normalized": false,
49
+ "rstrip": false,
50
+ "single_word": false,
51
+ "special": false
52
+ },
53
+ "100262": {
54
+ "content": "<|filename|>",
55
+ "lstrip": false,
56
+ "normalized": false,
57
+ "rstrip": false,
58
+ "single_word": false,
59
+ "special": false
60
+ },
61
+ "100263": {
62
+ "content": "<|reponame|>",
63
+ "lstrip": false,
64
+ "normalized": false,
65
+ "rstrip": false,
66
+ "single_word": false,
67
+ "special": false
68
+ },
69
+ "100264": {
70
+ "content": "<|start_of_role|>",
71
+ "lstrip": false,
72
+ "normalized": false,
73
+ "rstrip": false,
74
+ "single_word": false,
75
+ "special": true
76
+ },
77
+ "100265": {
78
+ "content": "<|end_of_role|>",
79
+ "lstrip": false,
80
+ "normalized": false,
81
+ "rstrip": false,
82
+ "single_word": false,
83
+ "special": true
84
+ },
85
+ "100266": {
86
+ "content": "<|unused_1|>",
87
+ "lstrip": false,
88
+ "normalized": false,
89
+ "rstrip": false,
90
+ "single_word": false,
91
+ "special": true
92
+ },
93
+ "100267": {
94
+ "content": "<|start_of_plugin|>",
95
+ "lstrip": false,
96
+ "normalized": false,
97
+ "rstrip": false,
98
+ "single_word": false,
99
+ "special": true
100
+ },
101
+ "100268": {
102
+ "content": "<|end_of_plugin|>",
103
+ "lstrip": false,
104
+ "normalized": false,
105
+ "rstrip": false,
106
+ "single_word": false,
107
+ "special": true
108
+ },
109
+ "100269": {
110
+ "content": "<|unk|>",
111
+ "lstrip": false,
112
+ "normalized": false,
113
+ "rstrip": false,
114
+ "single_word": false,
115
+ "special": true
116
+ },
117
+ "100270": {
118
+ "content": "<tool_call>",
119
+ "lstrip": false,
120
+ "normalized": false,
121
+ "rstrip": false,
122
+ "single_word": false,
123
+ "special": false
124
+ },
125
+ "100271": {
126
+ "content": "</tool_call>",
127
+ "lstrip": false,
128
+ "normalized": false,
129
+ "rstrip": false,
130
+ "single_word": false,
131
+ "special": false
132
+ },
133
+ "100272": {
134
+ "content": "<tool_response>",
135
+ "lstrip": false,
136
+ "normalized": false,
137
+ "rstrip": false,
138
+ "single_word": false,
139
+ "special": false
140
+ },
141
+ "100273": {
142
+ "content": "</tool_response>",
143
+ "lstrip": false,
144
+ "normalized": false,
145
+ "rstrip": false,
146
+ "single_word": false,
147
+ "special": false
148
+ },
149
+ "100274": {
150
+ "content": "<think>",
151
+ "lstrip": false,
152
+ "normalized": false,
153
+ "rstrip": false,
154
+ "single_word": false,
155
+ "special": false
156
+ },
157
+ "100275": {
158
+ "content": "</think>",
159
+ "lstrip": false,
160
+ "normalized": false,
161
+ "rstrip": false,
162
+ "single_word": false,
163
+ "special": false
164
+ },
165
+ "100276": {
166
+ "content": "<think_on>",
167
+ "lstrip": false,
168
+ "normalized": false,
169
+ "rstrip": false,
170
+ "single_word": false,
171
+ "special": true
172
+ },
173
+ "100277": {
174
+ "content": "<think_off>",
175
+ "lstrip": false,
176
+ "normalized": false,
177
+ "rstrip": false,
178
+ "single_word": false,
179
+ "special": true
180
+ },
181
+ "100278": {
182
+ "content": "<schema>",
183
+ "lstrip": false,
184
+ "normalized": false,
185
+ "rstrip": false,
186
+ "single_word": false,
187
+ "special": true
188
+ },
189
+ "100279": {
190
+ "content": "</schema>",
191
+ "lstrip": false,
192
+ "normalized": false,
193
+ "rstrip": false,
194
+ "single_word": false,
195
+ "special": true
196
+ },
197
+ "100280": {
198
+ "content": "<tools>",
199
+ "lstrip": false,
200
+ "normalized": false,
201
+ "rstrip": false,
202
+ "single_word": false,
203
+ "special": true
204
+ },
205
+ "100281": {
206
+ "content": "</tools>",
207
+ "lstrip": false,
208
+ "normalized": false,
209
+ "rstrip": false,
210
+ "single_word": false,
211
+ "special": true
212
+ },
213
+ "100282": {
214
+ "content": "<documents>",
215
+ "lstrip": false,
216
+ "normalized": false,
217
+ "rstrip": false,
218
+ "single_word": false,
219
+ "special": true
220
+ },
221
+ "100283": {
222
+ "content": "</documents>",
223
+ "lstrip": false,
224
+ "normalized": false,
225
+ "rstrip": false,
226
+ "single_word": false,
227
+ "special": true
228
+ },
229
+ "100284": {
230
+ "content": "<|unused_15|>",
231
+ "lstrip": false,
232
+ "normalized": false,
233
+ "rstrip": false,
234
+ "single_word": false,
235
+ "special": true
236
+ },
237
+ "100285": {
238
+ "content": "<|unused_16|>",
239
+ "lstrip": false,
240
+ "normalized": false,
241
+ "rstrip": false,
242
+ "single_word": false,
243
+ "special": true
244
+ },
245
+ "100286": {
246
+ "content": "<|unused_17|>",
247
+ "lstrip": false,
248
+ "normalized": false,
249
+ "rstrip": false,
250
+ "single_word": false,
251
+ "special": true
252
+ },
253
+ "100287": {
254
+ "content": "<|unused_18|>",
255
+ "lstrip": false,
256
+ "normalized": false,
257
+ "rstrip": false,
258
+ "single_word": false,
259
+ "special": true
260
+ },
261
+ "100288": {
262
+ "content": "<|unused_19|>",
263
+ "lstrip": false,
264
+ "normalized": false,
265
+ "rstrip": false,
266
+ "single_word": false,
267
+ "special": true
268
+ },
269
+ "100289": {
270
+ "content": "<|unused_20|>",
271
+ "lstrip": false,
272
+ "normalized": false,
273
+ "rstrip": false,
274
+ "single_word": false,
275
+ "special": true
276
+ },
277
+ "100290": {
278
+ "content": "<|unused_21|>",
279
+ "lstrip": false,
280
+ "normalized": false,
281
+ "rstrip": false,
282
+ "single_word": false,
283
+ "special": true
284
+ },
285
+ "100291": {
286
+ "content": "<|unused_22|>",
287
+ "lstrip": false,
288
+ "normalized": false,
289
+ "rstrip": false,
290
+ "single_word": false,
291
+ "special": true
292
+ },
293
+ "100292": {
294
+ "content": "<|unused_23|>",
295
+ "lstrip": false,
296
+ "normalized": false,
297
+ "rstrip": false,
298
+ "single_word": false,
299
+ "special": true
300
+ },
301
+ "100293": {
302
+ "content": "<|unused_24|>",
303
+ "lstrip": false,
304
+ "normalized": false,
305
+ "rstrip": false,
306
+ "single_word": false,
307
+ "special": true
308
+ },
309
+ "100294": {
310
+ "content": "<|unused_25|>",
311
+ "lstrip": false,
312
+ "normalized": false,
313
+ "rstrip": false,
314
+ "single_word": false,
315
+ "special": true
316
+ },
317
+ "100295": {
318
+ "content": "<|unused_26|>",
319
+ "lstrip": false,
320
+ "normalized": false,
321
+ "rstrip": false,
322
+ "single_word": false,
323
+ "special": true
324
+ },
325
+ "100296": {
326
+ "content": "<|unused_27|>",
327
+ "lstrip": false,
328
+ "normalized": false,
329
+ "rstrip": false,
330
+ "single_word": false,
331
+ "special": true
332
+ },
333
+ "100297": {
334
+ "content": "<|unused_28|>",
335
+ "lstrip": false,
336
+ "normalized": false,
337
+ "rstrip": false,
338
+ "single_word": false,
339
+ "special": true
340
+ },
341
+ "100298": {
342
+ "content": "<|unused_29|>",
343
+ "lstrip": false,
344
+ "normalized": false,
345
+ "rstrip": false,
346
+ "single_word": false,
347
+ "special": true
348
+ },
349
+ "100299": {
350
+ "content": "<|unused_30|>",
351
+ "lstrip": false,
352
+ "normalized": false,
353
+ "rstrip": false,
354
+ "single_word": false,
355
+ "special": true
356
+ },
357
+ "100300": {
358
+ "content": "<|unused_31|>",
359
+ "lstrip": false,
360
+ "normalized": false,
361
+ "rstrip": false,
362
+ "single_word": false,
363
+ "special": true
364
+ },
365
+ "100301": {
366
+ "content": "<|unused_32|>",
367
+ "lstrip": false,
368
+ "normalized": false,
369
+ "rstrip": false,
370
+ "single_word": false,
371
+ "special": true
372
+ },
373
+ "100302": {
374
+ "content": "<|unused_33|>",
375
+ "lstrip": false,
376
+ "normalized": false,
377
+ "rstrip": false,
378
+ "single_word": false,
379
+ "special": true
380
+ },
381
+ "100303": {
382
+ "content": "<|unused_34|>",
383
+ "lstrip": false,
384
+ "normalized": false,
385
+ "rstrip": false,
386
+ "single_word": false,
387
+ "special": true
388
+ },
389
+ "100304": {
390
+ "content": "<|unused_35|>",
391
+ "lstrip": false,
392
+ "normalized": false,
393
+ "rstrip": false,
394
+ "single_word": false,
395
+ "special": true
396
+ },
397
+ "100305": {
398
+ "content": "<|unused_36|>",
399
+ "lstrip": false,
400
+ "normalized": false,
401
+ "rstrip": false,
402
+ "single_word": false,
403
+ "special": true
404
+ },
405
+ "100306": {
406
+ "content": "<|unused_37|>",
407
+ "lstrip": false,
408
+ "normalized": false,
409
+ "rstrip": false,
410
+ "single_word": false,
411
+ "special": true
412
+ },
413
+ "100307": {
414
+ "content": "<|unused_38|>",
415
+ "lstrip": false,
416
+ "normalized": false,
417
+ "rstrip": false,
418
+ "single_word": false,
419
+ "special": true
420
+ },
421
+ "100308": {
422
+ "content": "<|unused_39|>",
423
+ "lstrip": false,
424
+ "normalized": false,
425
+ "rstrip": false,
426
+ "single_word": false,
427
+ "special": true
428
+ },
429
+ "100309": {
430
+ "content": "<|unused_40|>",
431
+ "lstrip": false,
432
+ "normalized": false,
433
+ "rstrip": false,
434
+ "single_word": false,
435
+ "special": true
436
+ },
437
+ "100310": {
438
+ "content": "<|unused_41|>",
439
+ "lstrip": false,
440
+ "normalized": false,
441
+ "rstrip": false,
442
+ "single_word": false,
443
+ "special": true
444
+ },
445
+ "100311": {
446
+ "content": "<|unused_42|>",
447
+ "lstrip": false,
448
+ "normalized": false,
449
+ "rstrip": false,
450
+ "single_word": false,
451
+ "special": true
452
+ },
453
+ "100312": {
454
+ "content": "<|unused_43|>",
455
+ "lstrip": false,
456
+ "normalized": false,
457
+ "rstrip": false,
458
+ "single_word": false,
459
+ "special": true
460
+ },
461
+ "100313": {
462
+ "content": "<|unused_44|>",
463
+ "lstrip": false,
464
+ "normalized": false,
465
+ "rstrip": false,
466
+ "single_word": false,
467
+ "special": true
468
+ },
469
+ "100314": {
470
+ "content": "<|unused_45|>",
471
+ "lstrip": false,
472
+ "normalized": false,
473
+ "rstrip": false,
474
+ "single_word": false,
475
+ "special": true
476
+ },
477
+ "100315": {
478
+ "content": "<|unused_46|>",
479
+ "lstrip": false,
480
+ "normalized": false,
481
+ "rstrip": false,
482
+ "single_word": false,
483
+ "special": true
484
+ },
485
+ "100316": {
486
+ "content": "<|unused_47|>",
487
+ "lstrip": false,
488
+ "normalized": false,
489
+ "rstrip": false,
490
+ "single_word": false,
491
+ "special": true
492
+ },
493
+ "100317": {
494
+ "content": "<|unused_48|>",
495
+ "lstrip": false,
496
+ "normalized": false,
497
+ "rstrip": false,
498
+ "single_word": false,
499
+ "special": true
500
+ },
501
+ "100318": {
502
+ "content": "<|unused_49|>",
503
+ "lstrip": false,
504
+ "normalized": false,
505
+ "rstrip": false,
506
+ "single_word": false,
507
+ "special": true
508
+ },
509
+ "100319": {
510
+ "content": "<|unused_50|>",
511
+ "lstrip": false,
512
+ "normalized": false,
513
+ "rstrip": false,
514
+ "single_word": false,
515
+ "special": true
516
+ },
517
+ "100320": {
518
+ "content": "<|unused_51|>",
519
+ "lstrip": false,
520
+ "normalized": false,
521
+ "rstrip": false,
522
+ "single_word": false,
523
+ "special": true
524
+ },
525
+ "100321": {
526
+ "content": "<|unused_52|>",
527
+ "lstrip": false,
528
+ "normalized": false,
529
+ "rstrip": false,
530
+ "single_word": false,
531
+ "special": true
532
+ },
533
+ "100322": {
534
+ "content": "<|unused_53|>",
535
+ "lstrip": false,
536
+ "normalized": false,
537
+ "rstrip": false,
538
+ "single_word": false,
539
+ "special": true
540
+ },
541
+ "100323": {
542
+ "content": "<|unused_54|>",
543
+ "lstrip": false,
544
+ "normalized": false,
545
+ "rstrip": false,
546
+ "single_word": false,
547
+ "special": true
548
+ },
549
+ "100324": {
550
+ "content": "<|unused_55|>",
551
+ "lstrip": false,
552
+ "normalized": false,
553
+ "rstrip": false,
554
+ "single_word": false,
555
+ "special": true
556
+ },
557
+ "100325": {
558
+ "content": "<|unused_56|>",
559
+ "lstrip": false,
560
+ "normalized": false,
561
+ "rstrip": false,
562
+ "single_word": false,
563
+ "special": true
564
+ },
565
+ "100326": {
566
+ "content": "<|unused_57|>",
567
+ "lstrip": false,
568
+ "normalized": false,
569
+ "rstrip": false,
570
+ "single_word": false,
571
+ "special": true
572
+ },
573
+ "100327": {
574
+ "content": "<|unused_58|>",
575
+ "lstrip": false,
576
+ "normalized": false,
577
+ "rstrip": false,
578
+ "single_word": false,
579
+ "special": true
580
+ },
581
+ "100328": {
582
+ "content": "<|unused_59|>",
583
+ "lstrip": false,
584
+ "normalized": false,
585
+ "rstrip": false,
586
+ "single_word": false,
587
+ "special": true
588
+ },
589
+ "100329": {
590
+ "content": "<|unused_60|>",
591
+ "lstrip": false,
592
+ "normalized": false,
593
+ "rstrip": false,
594
+ "single_word": false,
595
+ "special": true
596
+ },
597
+ "100330": {
598
+ "content": "<|unused_61|>",
599
+ "lstrip": false,
600
+ "normalized": false,
601
+ "rstrip": false,
602
+ "single_word": false,
603
+ "special": true
604
+ },
605
+ "100331": {
606
+ "content": "<|unused_62|>",
607
+ "lstrip": false,
608
+ "normalized": false,
609
+ "rstrip": false,
610
+ "single_word": false,
611
+ "special": true
612
+ },
613
+ "100332": {
614
+ "content": "<|unused_63|>",
615
+ "lstrip": false,
616
+ "normalized": false,
617
+ "rstrip": false,
618
+ "single_word": false,
619
+ "special": true
620
+ },
621
+ "100333": {
622
+ "content": "<|unused_64|>",
623
+ "lstrip": false,
624
+ "normalized": false,
625
+ "rstrip": false,
626
+ "single_word": false,
627
+ "special": true
628
+ },
629
+ "100334": {
630
+ "content": "<|unused_65|>",
631
+ "lstrip": false,
632
+ "normalized": false,
633
+ "rstrip": false,
634
+ "single_word": false,
635
+ "special": true
636
+ },
637
+ "100335": {
638
+ "content": "<|unused_66|>",
639
+ "lstrip": false,
640
+ "normalized": false,
641
+ "rstrip": false,
642
+ "single_word": false,
643
+ "special": true
644
+ },
645
+ "100336": {
646
+ "content": "<|unused_67|>",
647
+ "lstrip": false,
648
+ "normalized": false,
649
+ "rstrip": false,
650
+ "single_word": false,
651
+ "special": true
652
+ },
653
+ "100337": {
654
+ "content": "<|unused_68|>",
655
+ "lstrip": false,
656
+ "normalized": false,
657
+ "rstrip": false,
658
+ "single_word": false,
659
+ "special": true
660
+ },
661
+ "100338": {
662
+ "content": "<|unused_69|>",
663
+ "lstrip": false,
664
+ "normalized": false,
665
+ "rstrip": false,
666
+ "single_word": false,
667
+ "special": true
668
+ },
669
+ "100339": {
670
+ "content": "<|unused_70|>",
671
+ "lstrip": false,
672
+ "normalized": false,
673
+ "rstrip": false,
674
+ "single_word": false,
675
+ "special": true
676
+ },
677
+ "100340": {
678
+ "content": "<|unused_71|>",
679
+ "lstrip": false,
680
+ "normalized": false,
681
+ "rstrip": false,
682
+ "single_word": false,
683
+ "special": true
684
+ },
685
+ "100341": {
686
+ "content": "<|unused_72|>",
687
+ "lstrip": false,
688
+ "normalized": false,
689
+ "rstrip": false,
690
+ "single_word": false,
691
+ "special": true
692
+ },
693
+ "100342": {
694
+ "content": "<|unused_73|>",
695
+ "lstrip": false,
696
+ "normalized": false,
697
+ "rstrip": false,
698
+ "single_word": false,
699
+ "special": true
700
+ },
701
+ "100343": {
702
+ "content": "<|unused_74|>",
703
+ "lstrip": false,
704
+ "normalized": false,
705
+ "rstrip": false,
706
+ "single_word": false,
707
+ "special": true
708
+ },
709
+ "100344": {
710
+ "content": "<|unused_75|>",
711
+ "lstrip": false,
712
+ "normalized": false,
713
+ "rstrip": false,
714
+ "single_word": false,
715
+ "special": true
716
+ },
717
+ "100345": {
718
+ "content": "<|unused_76|>",
719
+ "lstrip": false,
720
+ "normalized": false,
721
+ "rstrip": false,
722
+ "single_word": false,
723
+ "special": true
724
+ },
725
+ "100346": {
726
+ "content": "<|unused_77|>",
727
+ "lstrip": false,
728
+ "normalized": false,
729
+ "rstrip": false,
730
+ "single_word": false,
731
+ "special": true
732
+ },
733
+ "100347": {
734
+ "content": "<|unused_78|>",
735
+ "lstrip": false,
736
+ "normalized": false,
737
+ "rstrip": false,
738
+ "single_word": false,
739
+ "special": true
740
+ },
741
+ "100348": {
742
+ "content": "<|unused_79|>",
743
+ "lstrip": false,
744
+ "normalized": false,
745
+ "rstrip": false,
746
+ "single_word": false,
747
+ "special": true
748
+ },
749
+ "100349": {
750
+ "content": "<|unused_80|>",
751
+ "lstrip": false,
752
+ "normalized": false,
753
+ "rstrip": false,
754
+ "single_word": false,
755
+ "special": true
756
+ },
757
+ "100350": {
758
+ "content": "<|unused_81|>",
759
+ "lstrip": false,
760
+ "normalized": false,
761
+ "rstrip": false,
762
+ "single_word": false,
763
+ "special": true
764
+ },
765
+ "100351": {
766
+ "content": "<|unused_82|>",
767
+ "lstrip": false,
768
+ "normalized": false,
769
+ "rstrip": false,
770
+ "single_word": false,
771
+ "special": true
772
+ }
773
+ },
774
+ "bos_token": "<|end_of_text|>",
775
+ "clean_up_tokenization_spaces": false,
776
+ "eos_token": "<|end_of_text|>",
777
+ "extra_special_tokens": {},
778
+ "model_max_length": 131072,
779
+ "pad_token": "<|pad|>",
780
+ "padding_side": "left",
781
+ "tokenizer_class": "GPT2Tokenizer",
782
+ "unk_token": "<|unk|>",
783
+ "chat_template": "{%- set tools_system_message_prefix = 'You are a helpful assistant with access to the following tools. You may call one or more tools to assist with the user query.\\n\\nYou are provided with function signatures within <tools></tools> XML tags:\\n<tools>' %}\n{%- set tools_system_message_suffix = '\\n</tools>\\n\\nFor each tool 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>. If a tool does not exist in the provided list of tools, notify the user that you do not have the ability to fulfill the request.' %}\n{%- set documents_system_message_prefix = 'You are a helpful assistant with access to the following documents. You may use one or more documents to assist with the user query.\\n\\nYou are given a list of documents within <documents></documents> XML tags:\\n<documents>' %}\n{%- set documents_system_message_suffix = '\\n</documents>\\n\\nWrite the response to the user\\'s input by strictly aligning with the facts in the provided documents. If the information needed to answer the question is not available in the documents, inform the user that the question cannot be answered based on the available data.' %}\n{%- if available_tools is defined and available_tools %}\n {%- set tools = available_tools %}\n{%- endif %}\n{%- set ns = namespace(tools_system_message=tools_system_message_prefix,\n documents_system_message=documents_system_message_prefix,\n system_message=''\n ) %}\n{%- if tools %}\n {%- for tool in tools %}\n {%- set ns.tools_system_message = ns.tools_system_message + '\\n' + (tool | tojson) %}\n {%- endfor %}\n {%- set ns.tools_system_message = ns.tools_system_message + tools_system_message_suffix %}\n{%- else %}\n {%- set ns.tools_system_message = '' %}\n{%- endif %}\n{%- if documents %}\n {%- for document in documents %}\n {%- set ns.documents_system_message = ns.documents_system_message + '\\n' + (document | tojson) %}\n {%- endfor %}\n {%- set ns.documents_system_message = ns.documents_system_message + documents_system_message_suffix %}\n{%- else %}\n {%- set ns.documents_system_message = '' %}\n{%- endif %}\n{%- if messages[0].role == 'system' %}\n {%- if messages[0].content is string %}\n {%- set ns.system_message = messages[0].content %}\n {%- elif messages[0].content is iterable %}\n {%- for entry in messages[0].content %}\n {%- if entry.type== 'text' %}\n {%- if ns.system_message != '' %}\n {%- set ns.system_message = ns.system_message + '\\n' %}\n {%- endif %}\n {%- set ns.system_message = ns.system_message + entry.text %}\n {%- endif %}\n {%- endfor %}\n {%- endif %}\n {%- if tools and documents %}\n {%- set ns.system_message = ns.system_message + '\\n\\n' + ns.tools_system_message + '\\n\\n' + ns.documents_system_message %}\n {%- elif tools %}\n {%- set ns.system_message = ns.system_message + '\\n\\n' + ns.tools_system_message %}\n {%- elif documents %}\n {%- set ns.system_message = ns.system_message + '\\n\\n' + ns.documents_system_message %}\n {%- endif %}\n{%- else %}\n {%- if tools and documents %}\n {%- set ns.system_message = ns.tools_system_message + '\\n\\n' + ns.documents_system_message %}\n {%- elif tools %}\n {%- set ns.system_message = ns.tools_system_message %}\n {%- elif documents %}\n {%- set ns.system_message = ns.documents_system_message %}\n {%- endif %}\n{%- endif %}\n{%- if ns.system_message %}\n {{- '<|start_of_role|>system<|end_of_role|>' + ns.system_message + '<|end_of_text|>\\n' }}\n{%- endif %}\n{%- for message in messages %}\n {%- set content = namespace(val='') %}\n {%- if message.content is string %}\n {%- set content.val = message.content %}\n {%- else %}\n {%- if message.content is iterable %}\n {%- for entry in message.content %}\n {%- if entry.type== 'text' %}\n {%- if content.val != '' %}\n {%- set content.val = content.val + '\\n' %}\n {%- endif %}\n {%- set content.val = content.val + entry.text %}\n {%- endif %}\n {%- endfor %}\n {%- endif %}\n {%- endif %}\n {%- if (message.role == 'user') or (message.role == 'system' and not loop.first) %}\n {{- '<|start_of_role|>' + message.role + '<|end_of_role|>' + content.val + '<|end_of_text|>\\n' }}\n {%- elif message.role == 'assistant' %}\n {{- '<|start_of_role|>' + message.role + '<|end_of_role|>' + content.val }}\n {%- if message.tool_calls %}\n {%- for tool_call in message.tool_calls %}\n {%- if (loop.first and content.val) 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 {{- '<|end_of_text|>\\n' }}\n {%- elif message.role == 'tool' %}\n {%- if loop.first or (messages[loop.index0 - 1].role != 'tool') %}\n {{- '<|start_of_role|>user<|end_of_role|>' }}\n {%- endif %}\n {{- '\\n<tool_response>\\n' }}\n {{- content.val }}\n {{- '\\n</tool_response>' }}\n {%- if loop.last or (messages[loop.index0 + 1].role != 'tool') %}\n {{- '<|end_of_text|>\\n' }}\n {%- endif %}\n {%- endif %}\n{%- endfor %}\n{%- if add_generation_prompt %}\n {{- '<|start_of_role|>assistant<|end_of_role|>' }}\n{%- endif %}"
784
+ }
vocab.json ADDED
The diff for this file is too large to render. See raw diff