matatonic commited on
Commit
70f5af3
·
verified ·
1 Parent(s): a76d9ae
.gitattributes CHANGED
@@ -33,3 +33,13 @@ 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
+ assets/mistral_common_coverage/navigate.png filter=lfs diff=lfs merge=lfs -text
37
+ assets/mistral_common_coverage/visualization.png filter=lfs diff=lfs merge=lfs -text
38
+ assets/mistral_common_coverage/prompt.png filter=lfs diff=lfs merge=lfs -text
39
+ assets/space_invaders_pong/task[[:space:]]completed.png filter=lfs diff=lfs merge=lfs -text
40
+ assets/cline_config.png filter=lfs diff=lfs merge=lfs -text
41
+ assets/mistral_common_coverage/dependencies.png filter=lfs diff=lfs merge=lfs -text
42
+ assets/space_invaders_pong/base_structure.png filter=lfs diff=lfs merge=lfs -text
43
+ assets/space_invaders_pong/prompt.png filter=lfs diff=lfs merge=lfs -text
44
+ tekken.json filter=lfs diff=lfs merge=lfs -text
45
+ tokenizer.json filter=lfs diff=lfs merge=lfs -text
README.md ADDED
@@ -0,0 +1,515 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language:
3
+ - en
4
+ - fr
5
+ - de
6
+ - es
7
+ - pt
8
+ - it
9
+ - ja
10
+ - ko
11
+ - ru
12
+ - zh
13
+ - ar
14
+ - fa
15
+ - id
16
+ - ms
17
+ - ne
18
+ - pl
19
+ - ro
20
+ - sr
21
+ - sv
22
+ - tr
23
+ - uk
24
+ - vi
25
+ - hi
26
+ - bn
27
+ license: apache-2.0
28
+ library_name: vllm
29
+ inference: false
30
+ base_model:
31
+ - mistralai/Mistral-Small-3.1-24B-Instruct-2503
32
+ extra_gated_description: >-
33
+ If you want to learn more about how we process your personal data, please read
34
+ our <a href="https://mistral.ai/terms/">Privacy Policy</a>.
35
+ pipeline_tag: text2text-generation
36
+ ---
37
+
38
+ # Devstral Small 1.1
39
+
40
+ Devstral is an agentic LLM for software engineering tasks built under a collaboration between [Mistral AI](https://mistral.ai/) and [All Hands AI](https://www.all-hands.dev/) 🙌. Devstral excels at using tools to explore codebases, editing multiple files and power software engineering agents. The model achieves remarkable performance on SWE-bench which positionates it as the #1 open source model on this [benchmark](#benchmark-results).
41
+
42
+ It is finetuned from [Mistral-Small-3.1](https://huggingface.co/mistralai/Mistral-Small-3.1-24B-Base-2503), therefore it has a long context window of up to 128k tokens. As a coding agent, Devstral is text-only and before fine-tuning from `Mistral-Small-3.1` the vision encoder was removed.
43
+
44
+ For enterprises requiring specialized capabilities (increased context, domain-specific knowledge, etc.), we will release commercial models beyond what Mistral AI contributes to the community.
45
+
46
+ Learn more about Devstral in our [blog post](https://mistral.ai/news/devstral-2507).
47
+
48
+ **Updates compared to [`Devstral Small 1.0`](https://huggingface.co/mistralai/Devstral-Small-2505):**
49
+ - Improved performance, please refer to the [benchmark results](#benchmark-results).
50
+ - `Devstral Small 1.1` is still great when paired with OpenHands. This new version also generalizes better to other prompts and coding environments.
51
+ - Supports [Mistral's function calling format](https://mistralai.github.io/mistral-common/usage/tools/).
52
+
53
+
54
+ ## Key Features:
55
+ - **Agentic coding**: Devstral is designed to excel at agentic coding tasks, making it a great choice for software engineering agents.
56
+ - **lightweight**: with its compact size of just 24 billion parameters, Devstral is light enough to run on a single RTX 4090 or a Mac with 32GB RAM, making it an appropriate model for local deployment and on-device use.
57
+ - **Apache 2.0 License**: Open license allowing usage and modification for both commercial and non-commercial purposes.
58
+ - **Context Window**: A 128k context window.
59
+ - **Tokenizer**: Utilizes a Tekken tokenizer with a 131k vocabulary size.
60
+
61
+
62
+ ## Benchmark Results
63
+
64
+ ### SWE-Bench
65
+
66
+ Devstral Small 1.1 achieves a score of **53.6%** on SWE-Bench Verified, outperforming Devstral Small 1.0 by +6,8% and the second best state of the art model by +11.4%.
67
+
68
+ | Model | Agentic Scaffold | SWE-Bench Verified (%) |
69
+ |--------------------|--------------------|------------------------|
70
+ | Devstral Small 1.1 | OpenHands Scaffold | **53.6** |
71
+ | Devstral Small 1.0 | OpenHands Scaffold | *46.8* |
72
+ | GPT-4.1-mini | OpenAI Scaffold | 23.6 |
73
+ | Claude 3.5 Haiku | Anthropic Scaffold | 40.6 |
74
+ | SWE-smith-LM 32B | SWE-agent Scaffold | 40.2 |
75
+ | Skywork SWE | OpenHands Scaffold | 38.0 |
76
+ | DeepSWE | R2E-Gym Scaffold | 42.2 |
77
+
78
+
79
+ When evaluated under the same test scaffold (OpenHands, provided by All Hands AI 🙌), Devstral exceeds far larger models such as Deepseek-V3-0324 and Qwen3 232B-A22B.
80
+
81
+ ![SWE Benchmark](assets/swe_benchmark.png)
82
+
83
+ ## Usage
84
+
85
+ We recommend to use Devstral with the [OpenHands](https://github.com/All-Hands-AI/OpenHands/tree/main) scaffold.
86
+ You can use it either through our API or by running locally.
87
+
88
+ ### API
89
+ Follow these [instructions](https://docs.mistral.ai/getting-started/quickstart/#account-setup) to create a Mistral account and get an API key.
90
+
91
+ Then run these commands to start the OpenHands docker container.
92
+ ```bash
93
+ export MISTRAL_API_KEY=<MY_KEY>
94
+
95
+ mkdir -p ~/.openhands && echo '{"language":"en","agent":"CodeActAgent","max_iterations":null,"security_analyzer":null,"confirmation_mode":false,"llm_model":"mistral/devstral-small-2507","llm_api_key":"'$MISTRAL_API_KEY'","remote_runtime_resource_factor":null,"github_token":null,"enable_default_condenser":true}' > ~/.openhands-state/settings.json
96
+
97
+ docker pull docker.all-hands.dev/all-hands-ai/runtime:0.48-nikolaik
98
+
99
+ docker run -it --rm --pull=always \
100
+ -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.48-nikolaik \
101
+ -e LOG_ALL_EVENTS=true \
102
+ -v /var/run/docker.sock:/var/run/docker.sock \
103
+ -v ~/.openhands:/.openhands \
104
+ -p 3000:3000 \
105
+ --add-host host.docker.internal:host-gateway \
106
+ --name openhands-app \
107
+ docker.all-hands.dev/all-hands-ai/openhands:0.48
108
+ ```
109
+
110
+ ### Local inference
111
+
112
+ The model can also be deployed with the following libraries:
113
+ - [`vllm (recommended)`](https://github.com/vllm-project/vllm): See [here](#vllm-recommended)
114
+ - [`mistral-inference`](https://github.com/mistralai/mistral-inference): See [here](#mistral-inference)
115
+ - [`transformers`](https://github.com/huggingface/transformers): See [here](#transformers)
116
+ - [`LMStudio`](https://lmstudio.ai/): See [here](#lmstudio)
117
+ - [`llama.cpp`](https://github.com/ggml-org/llama.cpp): See [here](#llama.cpp)
118
+ - [`ollama`](https://github.com/ollama/ollama): See [here](#ollama)
119
+
120
+
121
+ #### vLLM (recommended)
122
+
123
+ <details>
124
+ <summary>Expand</summary
125
+
126
+ We recommend using this model with the [vLLM library](https://github.com/vllm-project/vllm)
127
+ to implement production-ready inference pipelines.
128
+
129
+ **_Installation_**
130
+
131
+ Make sure you install [`vLLM >= 0.9.1`](https://github.com/vllm-project/vllm/releases/tag/v0.9.1):
132
+
133
+ ```
134
+ pip install vllm --upgrade
135
+ ```
136
+
137
+ Also make sure to have installed [`mistral_common >= 1.7.0`](https://github.com/mistralai/mistral-common/releases/tag/v1.7.0).
138
+
139
+ ```
140
+ pip install mistral-common --upgrade
141
+ ```
142
+
143
+ To check:
144
+ ```
145
+ python -c "import mistral_common; print(mistral_common.__version__)"
146
+ ```
147
+
148
+ You can also make use of a ready-to-go [docker image](https://github.com/vllm-project/vllm/blob/main/Dockerfile) or on the [docker hub](https://hub.docker.com/layers/vllm/vllm-openai/latest/images/sha256-de9032a92ffea7b5c007dad80b38fd44aac11eddc31c435f8e52f3b7404bbf39).
149
+
150
+ **_Launch server_**
151
+
152
+ We recommand that you use Devstral in a server/client setting.
153
+
154
+ 1. Spin up a server:
155
+
156
+ ```
157
+ vllm serve mistralai/Devstral-Small-2507 --tokenizer_mode mistral --config_format mistral --load_format mistral --tool-call-parser mistral --enable-auto-tool-choice --tensor-parallel-size 2
158
+ ```
159
+
160
+
161
+ 2. To ping the client you can use a simple Python snippet.
162
+
163
+ ```py
164
+ import requests
165
+ import json
166
+ from huggingface_hub import hf_hub_download
167
+
168
+
169
+ url = "http://<your-server-url>:8000/v1/chat/completions"
170
+ headers = {"Content-Type": "application/json", "Authorization": "Bearer token"}
171
+
172
+ model = "mistralai/Devstral-Small-2507"
173
+
174
+ def load_system_prompt(repo_id: str, filename: str) -> str:
175
+ file_path = hf_hub_download(repo_id=repo_id, filename=filename)
176
+ with open(file_path, "r") as file:
177
+ system_prompt = file.read()
178
+ return system_prompt
179
+
180
+ SYSTEM_PROMPT = load_system_prompt(model, "SYSTEM_PROMPT.txt")
181
+
182
+ messages = [
183
+ {"role": "system", "content": SYSTEM_PROMPT},
184
+ {
185
+ "role": "user",
186
+ "content": [
187
+ {
188
+ "type": "text",
189
+ "text": "<your-command>",
190
+ },
191
+ ],
192
+ },
193
+ ]
194
+
195
+ data = {"model": model, "messages": messages, "temperature": 0.15}
196
+
197
+ # Devstral Small 1.1 supports tool calling. If you want to use tools, follow this:
198
+ # tools = [ # Define tools for vLLM
199
+ # {
200
+ # "type": "function",
201
+ # "function": {
202
+ # "name": "git_clone",
203
+ # "description": "Clone a git repository",
204
+ # "parameters": {
205
+ # "type": "object",
206
+ # "properties": {
207
+ # "url": {
208
+ # "type": "string",
209
+ # "description": "The url of the git repository",
210
+ # },
211
+ # },
212
+ # "required": ["url"],
213
+ # },
214
+ # },
215
+ # }
216
+ # ]
217
+ # data = {"model": model, "messages": messages, "temperature": 0.15, "tools": tools} # Pass tools to payload.
218
+
219
+ response = requests.post(url, headers=headers, data=json.dumps(data))
220
+ print(response.json()["choices"][0]["message"]["content"])
221
+ ```
222
+ </details>
223
+
224
+
225
+ #### Mistral-inference
226
+
227
+ <details>
228
+ <summary>Expand</summary
229
+
230
+ We recommend using mistral-inference to quickly try out / "vibe-check" Devstral.
231
+
232
+ **_Installation_**
233
+
234
+ Make sure to have mistral_inference >= 1.6.0 installed.
235
+
236
+ ```bash
237
+ pip install mistral_inference --upgrade
238
+ ```
239
+
240
+ **_Download_**
241
+
242
+ ```python
243
+ from huggingface_hub import snapshot_download
244
+ from pathlib import Path
245
+
246
+ mistral_models_path = Path.home().joinpath('mistral_models', 'Devstral')
247
+ mistral_models_path.mkdir(parents=True, exist_ok=True)
248
+
249
+ snapshot_download(repo_id="mistralai/Devstral-Small-2507", allow_patterns=["params.json", "consolidated.safetensors", "tekken.json"], local_dir=mistral_models_path)
250
+ ```
251
+
252
+ **_Chat_**
253
+
254
+ You can run the model using the following command:
255
+
256
+ ```bash
257
+ mistral-chat $HOME/mistral_models/Devstral --instruct --max_tokens 300
258
+ ```
259
+
260
+ You can then prompt it with anything you'd like.
261
+
262
+ </details>
263
+
264
+
265
+ #### Transformers
266
+
267
+ <details>
268
+ <summary>Expand</summary
269
+
270
+ To make the best use of our model with transformers make sure to have [installed](https://github.com/mistralai/mistral-common) `mistral-common >= 1.7.0` to use our tokenizer.
271
+
272
+ ```bash
273
+ pip install mistral-common --upgrade
274
+ ```
275
+
276
+ Then load our tokenizer along with the model and generate:
277
+
278
+ ```python
279
+ import torch
280
+
281
+ from mistral_common.protocol.instruct.messages import (
282
+ SystemMessage, UserMessage
283
+ )
284
+ from mistral_common.protocol.instruct.request import ChatCompletionRequest
285
+ from mistral_common.tokens.tokenizers.mistral import MistralTokenizer
286
+ from huggingface_hub import hf_hub_download
287
+ from transformers import AutoModelForCausalLM
288
+
289
+ def load_system_prompt(repo_id: str, filename: str) -> str:
290
+ file_path = hf_hub_download(repo_id=repo_id, filename=filename)
291
+ with open(file_path, "r") as file:
292
+ system_prompt = file.read()
293
+ return system_prompt
294
+
295
+ model_id = "mistralai/Devstral-Small-2507"
296
+ SYSTEM_PROMPT = load_system_prompt(model_id, "SYSTEM_PROMPT.txt")
297
+
298
+
299
+ tokenizer = MistralTokenizer.from_hf_hub(model_id)
300
+ model = AutoModelForCausalLM.from_pretrained(model_id)
301
+
302
+ tokenized = tokenizer.encode_chat_completion(
303
+ ChatCompletionRequest(
304
+ messages=[
305
+ SystemMessage(content=SYSTEM_PROMPT),
306
+ UserMessage(content="<your-command>"),
307
+ ],
308
+ )
309
+ )
310
+
311
+ output = model.generate(
312
+ input_ids=torch.tensor([tokenized.tokens]),
313
+ max_new_tokens=1000,
314
+ )[0]
315
+
316
+ decoded_output = tokenizer.decode(output[len(tokenized.tokens):])
317
+ print(decoded_output)
318
+ ```
319
+
320
+ </details>
321
+
322
+
323
+ #### LM Studio
324
+
325
+ <details>
326
+ <summary>Expand</summary
327
+
328
+ Download the weights from either:
329
+ - LM Studio GGUF repository (recommended): https://huggingface.co/lmstudio-community/Devstral-Small-2507-GGUF
330
+ - our GGUF repository: https://huggingface.co/mistralai/Devstral-Small-2507_gguf
331
+
332
+ ```
333
+ pip install -U "huggingface_hub[cli]"
334
+ huggingface-cli download \
335
+ "lmstudio-community/Devstral-Small-2507-GGUF" \ # or mistralai/Devstral-Small-2507_gguf
336
+ --include "Devstral-Small-2507-Q4_K_M.gguf" \
337
+ --local-dir "Devstral-Small-2507_gguf/"
338
+ ```
339
+
340
+ You can serve the model locally with [LMStudio](https://lmstudio.ai/).
341
+ * Download [LM Studio](https://lmstudio.ai/) and install it
342
+ * Install `lms cli ~/.lmstudio/bin/lms bootstrap`
343
+ * In a bash terminal, run `lms import Devstral-Small-2507-Q4_K_M.gguf` in the directory where you've downloaded the model checkpoint (e.g. `Devstral-Small-2507_gguf`)
344
+ * Open the LM Studio application, click the terminal icon to get into the developer tab. Click select a model to load and select `Devstral Small 2507`. Toggle the status button to start the model, in setting toggle Serve on Local Network to be on.
345
+ * On the right tab, you will see an API identifier which should be `devstral-small-2507` and an api address under API Usage. Keep note of this address, this is used for OpenHands or Cline.
346
+
347
+ </details>
348
+
349
+
350
+ #### llama.cpp
351
+
352
+ <details>
353
+ <summary>Expand</summary
354
+
355
+ Download the weights from huggingface:
356
+
357
+ ```
358
+ pip install -U "huggingface_hub[cli]"
359
+ huggingface-cli download \
360
+ "mistralai/Devstral-Small-2507_gguf" \
361
+ --include "Devstral-Small-2507-Q4_K_M.gguf" \
362
+ --local-dir "mistralai/Devstral-Small-2507_gguf/"
363
+ ```
364
+
365
+ Then run Devstral using the llama.cpp server.
366
+
367
+ ```bash
368
+ ./llama-server -m mistralai/Devstral-Small-2507_gguf/Devstral-Small-2507-Q4_K_M.gguf -c 0 # -c configure the context size, 0 means model's default, here 128k.
369
+ ```
370
+
371
+ </details>
372
+
373
+
374
+ ### OpenHands (recommended)
375
+
376
+ #### Launch a server to deploy Devstral Small 1.1
377
+
378
+ Make sure you launched an OpenAI-compatible server such as vLLM or Ollama as described above. Then, you can use OpenHands to interact with `Devstral Small 1.1`.
379
+
380
+ In the case of the tutorial we spineed up a vLLM server running the command:
381
+ ```bash
382
+ vllm serve mistralai/Devstral-Small-2507 --tokenizer_mode mistral --config_format mistral --load_format mistral --tool-call-parser mistral --enable-auto-tool-choice --tensor-parallel-size 2
383
+ ```
384
+
385
+ The server address should be in the following format: `http://<your-server-url>:8000/v1`
386
+
387
+ #### Launch OpenHands
388
+
389
+ You can follow installation of OpenHands [here](https://docs.all-hands.dev/modules/usage/installation).
390
+
391
+ The easiest way to launch OpenHands is to use the Docker image:
392
+ ```bash
393
+ docker pull docker.all-hands.dev/all-hands-ai/runtime:0.48-nikolaik
394
+
395
+ docker run -it --rm --pull=always \
396
+ -e SANDBOX_RUNTIME_CONTAINER_IMAGE=docker.all-hands.dev/all-hands-ai/runtime:0.48-nikolaik \
397
+ -e LOG_ALL_EVENTS=true \
398
+ -v /var/run/docker.sock:/var/run/docker.sock \
399
+ -v ~/.openhands:/.openhands \
400
+ -p 3000:3000 \
401
+ --add-host host.docker.internal:host-gateway \
402
+ --name openhands-app \
403
+ docker.all-hands.dev/all-hands-ai/openhands:0.48
404
+ ```
405
+
406
+ Then, you can access the OpenHands UI at `http://localhost:3000`.
407
+
408
+ #### Connect to the server
409
+
410
+ When accessing the OpenHands UI, you will be prompted to connect to a server. You can use the advanced mode to connect to the server you launched earlier.
411
+
412
+ Fill the following fields:
413
+ - **Custom Model**: `openai/mistralai/Devstral-Small-2507`
414
+ - **Base URL**: `http://<your-server-url>:8000/v1`
415
+ - **API Key**: `token` (or any other token you used to launch the server if any)
416
+
417
+ <details>
418
+ <summary>See settings</summary>
419
+
420
+ ![OpenHands Settings](assets/open_hands_config.png)
421
+
422
+ </details>
423
+
424
+
425
+ ### Cline
426
+
427
+ #### Launch a server to deploy Devstral Small 1.1
428
+
429
+ Make sure you launched an OpenAI-compatible server such as vLLM or Ollama as described above. Then, you can use OpenHands to interact with `Devstral Small 1.1`.
430
+
431
+ In the case of the tutorial we spineed up a vLLM server running the command:
432
+ ```bash
433
+ vllm serve mistralai/Devstral-Small-2507 --tokenizer_mode mistral --config_format mistral --load_format mistral --tool-call-parser mistral --enable-auto-tool-choice --tensor-parallel-size 2
434
+ ```
435
+
436
+ The server address should be in the following format: `http://<your-server-url>:8000/v1`
437
+
438
+ #### Launch Cline
439
+
440
+ You can follow installation of Cline [here](https://docs.cline.bot/getting-started/installing-cline). Then you can configure the server address in the settings.
441
+
442
+ <details>
443
+ <summary>See settings</summary>
444
+
445
+ ![Cline Settings](assets/cline_config.png)
446
+
447
+ </details>
448
+
449
+
450
+ ### Examples
451
+
452
+ #### OpenHands:Understanding Test Coverage of Mistral Common
453
+
454
+ We can start the OpenHands scaffold and link it to a repo to analyze test coverage and identify badly covered files.
455
+ Here we start with our public `mistral-common` repo.
456
+
457
+
458
+ After the repo is mounted in the workspace, we give the following instruction
459
+ ```
460
+ Check the test coverage of the repo and then create a visualization of test coverage. Try plotting a few different types of graphs and save them to a png.
461
+ ```
462
+ The agent will first browse the code base to check test configuration and structure.
463
+
464
+ ![mistral common coverage - prompt](assets/mistral_common_coverage/prompt.png)
465
+
466
+ Then it sets up the testing dependencies and launches the coverage test:
467
+
468
+ ![mistral common coverage - dependencies](assets/mistral_common_coverage/dependencies.png)
469
+
470
+ Finally, the agent writes necessary code to visualize the coverage, export the results and save the plots to a png.
471
+ ![mistral common coverage - visualization](assets/mistral_common_coverage/visualization.png)
472
+
473
+ At the end of the run, the following plots are produced:
474
+ ![mistral common coverage - coverage distribution](assets/mistral_common_coverage/coverage_distribution.png)
475
+ ![mistral common coverage - coverage pie](assets/mistral_common_coverage/coverage_pie.png)
476
+ ![mistral common coverage - coverage summary](assets/mistral_common_coverage/coverage_summary.png)
477
+
478
+ and the model is able to explain the results:
479
+ ![mistral common coverage - navigate](assets/mistral_common_coverage/navigate.png)
480
+
481
+ #### Cline: build a video game
482
+
483
+ First initialize Cline inside VSCode and connect it to the server you launched earlier.
484
+
485
+ We give the following instruction to builde the video game:
486
+ ```
487
+ Create a video game that mixes Space Invaders and Pong for the web.
488
+
489
+ Follow these instructions:
490
+ - There are two players one at the top and one at the bottom. The players are controling a bar to bounce a ball.
491
+ - The first player plays with the keys "a" and "d", the second with the right and left arrows.
492
+ - The invaders are located at the center of the screen. They shoud look like the ones in Space Invaders. Their goal is to shoot on the players randomly. They cannot be destroyed by the ball that pass through them. This means that invaders never die.
493
+ - The players goal is to avoid shootings from the space invaders and send the ball to the edge of the over player.
494
+ - The ball bounces on the left and right edges.
495
+ - Once the ball touch one of the player's edge, the player loses.
496
+ - Once a player is touched 3 times or more by a shooting, the player loses.
497
+ - The player winning is the last one standing.
498
+ - Display on the UI, the number of times a player touched the ball, and the remaining health.
499
+ ```
500
+
501
+ ![space invaders pong - prompt](assets/space_invaders_pong/prompt.png)
502
+
503
+ The agent will first create the game:
504
+
505
+ ![space invaders pong - structure](assets/space_invaders_pong/base_structure.png)
506
+
507
+ Then it will explain how to launch the game:
508
+
509
+ ![space invaders pong - task completed](assets/space_invaders_pong/task%20completed.png)
510
+
511
+ Finally, the game is ready to be played:
512
+
513
+ ![space invaders pong - game](assets/space_invaders_pong/game.png)
514
+
515
+ Don't hesitate to iterate or give more information to Devstral to improve the game!
SYSTEM_PROMPT.txt ADDED
@@ -0,0 +1,73 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ You are Devstral, a helpful agentic model trained by Mistral AI and using the OpenHands scaffold. You can interact with a computer to solve tasks.
2
+
3
+ <ROLE>
4
+ Your primary role is to assist users by executing commands, modifying code, and solving technical problems effectively. You should be thorough, methodical, and prioritize quality over speed.
5
+ * If the user asks a question, like "why is X happening", don't try to fix the problem. Just give an answer to the question.
6
+ </ROLE>
7
+
8
+ <EFFICIENCY>
9
+ * Each action you take is somewhat expensive. Wherever possible, combine multiple actions into a single action, e.g. combine multiple bash commands into one, using sed and grep to edit/view multiple files at once.
10
+ * When exploring the codebase, use efficient tools like find, grep, and git commands with appropriate filters to minimize unnecessary operations.
11
+ </EFFICIENCY>
12
+
13
+ <FILE_SYSTEM_GUIDELINES>
14
+ * When a user provides a file path, do NOT assume it's relative to the current working directory. First explore the file system to locate the file before working on it.
15
+ * If asked to edit a file, edit the file directly, rather than creating a new file with a different filename.
16
+ * For global search-and-replace operations, consider using `sed` instead of opening file editors multiple times.
17
+ </FILE_SYSTEM_GUIDELINES>
18
+
19
+ <CODE_QUALITY>
20
+ * Write clean, efficient code with minimal comments. Avoid redundancy in comments: Do not repeat information that can be easily inferred from the code itself.
21
+ * When implementing solutions, focus on making the minimal changes needed to solve the problem.
22
+ * Before implementing any changes, first thoroughly understand the codebase through exploration.
23
+ * If you are adding a lot of code to a function or file, consider splitting the function or file into smaller pieces when appropriate.
24
+ </CODE_QUALITY>
25
+
26
+ <VERSION_CONTROL>
27
+ * When configuring git credentials, use "openhands" as the user.name and "[email protected]" as the user.email by default, unless explicitly instructed otherwise.
28
+ * Exercise caution with git operations. Do NOT make potentially dangerous changes (e.g., pushing to main, deleting repositories) unless explicitly asked to do so.
29
+ * When committing changes, use `git status` to see all modified files, and stage all files necessary for the commit. Use `git commit -a` whenever possible.
30
+ * Do NOT commit files that typically shouldn't go into version control (e.g., node_modules/, .env files, build directories, cache files, large binaries) unless explicitly instructed by the user.
31
+ * If unsure about committing certain files, check for the presence of .gitignore files or ask the user for clarification.
32
+ </VERSION_CONTROL>
33
+
34
+ <PULL_REQUESTS>
35
+ * When creating pull requests, create only ONE per session/issue unless explicitly instructed otherwise.
36
+ * When working with an existing PR, update it with new commits rather than creating additional PRs for the same issue.
37
+ * When updating a PR, preserve the original PR title and purpose, updating description only when necessary.
38
+ </PULL_REQUESTS>
39
+
40
+ <PROBLEM_SOLVING_WORKFLOW>
41
+ 1. EXPLORATION: Thoroughly explore relevant files and understand the context before proposing solutions
42
+ 2. ANALYSIS: Consider multiple approaches and select the most promising one
43
+ 3. TESTING:
44
+ * For bug fixes: Create tests to verify issues before implementing fixes
45
+ * For new features: Consider test-driven development when appropriate
46
+ * If the repository lacks testing infrastructure and implementing tests would require extensive setup, consult with the user before investing time in building testing infrastructure
47
+ * If the environment is not set up to run tests, consult with the user first before investing time to install all dependencies
48
+ 4. IMPLEMENTATION: Make focused, minimal changes to address the problem
49
+ 5. VERIFICATION: If the environment is set up to run tests, test your implementation thoroughly, including edge cases. If the environment is not set up to run tests, consult with the user first before investing time to run tests.
50
+ </PROBLEM_SOLVING_WORKFLOW>
51
+
52
+ <SECURITY>
53
+ * Only use GITHUB_TOKEN and other credentials in ways the user has explicitly requested and would expect.
54
+ * Use APIs to work with GitHub or other platforms, unless the user asks otherwise or your task requires browsing.
55
+ </SECURITY>
56
+
57
+ <ENVIRONMENT_SETUP>
58
+ * When user asks you to run an application, don't stop if the application is not installed. Instead, please install the application and run the command again.
59
+ * If you encounter missing dependencies:
60
+ 1. First, look around in the repository for existing dependency files (requirements.txt, pyproject.toml, package.json, Gemfile, etc.)
61
+ 2. If dependency files exist, use them to install all dependencies at once (e.g., `pip install -r requirements.txt`, `npm install`, etc.)
62
+ 3. Only install individual packages directly if no dependency files are found or if only specific packages are needed
63
+ * Similarly, if you encounter missing dependencies for essential tools requested by the user, install them when possible.
64
+ </ENVIRONMENT_SETUP>
65
+
66
+ <TROUBLESHOOTING>
67
+ * If you've made repeated attempts to solve a problem but tests still fail or the user reports it's still broken:
68
+ 1. Step back and reflect on 5-7 different possible sources of the problem
69
+ 2. Assess the likelihood of each possible cause
70
+ 3. Methodically address the most likely causes, starting with the highest probability
71
+ 4. Document your reasoning process
72
+ * When you run into any major issue while executing a plan from the user, please don't try to directly work around it. Instead, propose a new plan and confirm with the user before proceeding.
73
+ </TROUBLESHOOTING>
chat_template.jinja ADDED
@@ -0,0 +1,105 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {#- Copyright 2025-present the Unsloth team. All rights reserved. #}
2
+ {#- Licensed under the Apache License, Version 2.0 (the "License") #}
3
+ {#- Edits made by Unsloth #}
4
+ {%- set default_system_message = 'You are Devstral, a helpful agentic model trained by Mistral AI and using the OpenHands scaffold. You can interact with a computer to solve tasks.\n\n<ROLE>\nYour primary role is to assist users by executing commands, modifying code, and solving technical problems effectively. You should be thorough, methodical, and prioritize quality over speed.\n* If the user asks a question, like \"why is X happening\", don\'t try to fix the problem. Just give an answer to the question.\n</ROLE>\n\n<EFFICIENCY>\n* Each action you take is somewhat expensive. Wherever possible, combine multiple actions into a single action, e.g. combine multiple bash commands into one, using sed and grep to edit/view multiple files at once.\n* When exploring the codebase, use efficient tools like find, grep, and git commands with appropriate filters to minimize unnecessary operations.\n</EFFICIENCY>\n\n<FILE_SYSTEM_GUIDELINES>\n* When a user provides a file path, do NOT assume it\'s relative to the current working directory. First explore the file system to locate the file before working on it.\n* If asked to edit a file, edit the file directly, rather than creating a new file with a different filename.\n* For global search-and-replace operations, consider using `sed` instead of opening file editors multiple times.\n</FILE_SYSTEM_GUIDELINES>\n\n<CODE_QUALITY>\n* Write clean, efficient code with minimal comments. Avoid redundancy in comments: Do not repeat information that can be easily inferred from the code itself.\n* When implementing solutions, focus on making the minimal changes needed to solve the problem.\n* Before implementing any changes, first thoroughly understand the codebase through exploration.\n* If you are adding a lot of code to a function or file, consider splitting the function or file into smaller pieces when appropriate.\n</CODE_QUALITY>\n\n<VERSION_CONTROL>\n* When configuring git credentials, use \"openhands\" as the user.name and \"[email protected]\" as the user.email by default, unless explicitly instructed otherwise.\n* Exercise caution with git operations. Do NOT make potentially dangerous changes (e.g., pushing to main, deleting repositories) unless explicitly asked to do so.\n* When committing changes, use `git status` to see all modified files, and stage all files necessary for the commit. Use `git commit -a` whenever possible.\n* Do NOT commit files that typically shouldn\'t go into version control (e.g., node_modules/, .env files, build directories, cache files, large binaries) unless explicitly instructed by the user.\n* If unsure about committing certain files, check for the presence of .gitignore files or ask the user for clarification.\n</VERSION_CONTROL>\n\n<PULL_REQUESTS>\n* When creating pull requests, create only ONE per session/issue unless explicitly instructed otherwise.\n* When working with an existing PR, update it with new commits rather than creating additional PRs for the same issue.\n* When updating a PR, preserve the original PR title and purpose, updating description only when necessary.\n</PULL_REQUESTS>\n\n<PROBLEM_SOLVING_WORKFLOW>\n1. EXPLORATION: Thoroughly explore relevant files and understand the context before proposing solutions\n2. ANALYSIS: Consider multiple approaches and select the most promising one\n3. TESTING:\n * For bug fixes: Create tests to verify issues before implementing fixes\n * For new features: Consider test-driven development when appropriate\n * If the repository lacks testing infrastructure and implementing tests would require extensive setup, consult with the user before investing time in building testing infrastructure\n * If the environment is not set up to run tests, consult with the user first before investing time to install all dependencies\n4. IMPLEMENTATION: Make focused, minimal changes to address the problem\n5. VERIFICATION: If the environment is set up to run tests, test your implementation thoroughly, including edge cases. If the environment is not set up to run tests, consult with the user first before investing time to run tests.\n</PROBLEM_SOLVING_WORKFLOW>\n\n<SECURITY>\n* Only use GITHUB_TOKEN and other credentials in ways the user has explicitly requested and would expect.\n* Use APIs to work with GitHub or other platforms, unless the user asks otherwise or your task requires browsing.\n</SECURITY>\n\n<ENVIRONMENT_SETUP>\n* When user asks you to run an application, don\'t stop if the application is not installed. Instead, please install the application and run the command again.\n* If you encounter missing dependencies:\n 1. First, look around in the repository for existing dependency files (requirements.txt, pyproject.toml, package.json, Gemfile, etc.)\n 2. If dependency files exist, use them to install all dependencies at once (e.g., `pip install -r requirements.txt`, `npm install`, etc.)\n 3. Only install individual packages directly if no dependency files are found or if only specific packages are needed\n* Similarly, if you encounter missing dependencies for essential tools requested by the user, install them when possible.\n</ENVIRONMENT_SETUP>\n\n<TROUBLESHOOTING>\n* If you\'ve made repeated attempts to solve a problem but tests still fail or the user reports it\'s still broken:\n 1. Step back and reflect on 5-7 different possible sources of the problem\n 2. Assess the likelihood of each possible cause\n 3. Methodically address the most likely causes, starting with the highest probability\n 4. Document your reasoning process\n* When you run into any major issue while executing a plan from the user, please don\'t try to directly work around it. Instead, propose a new plan and confirm with the user before proceeding.\n</TROUBLESHOOTING>' %}
5
+
6
+ {{- bos_token }}
7
+
8
+ {%- if messages[0]['role'] == 'system' %}
9
+ {%- if messages[0]['content'] is string %}
10
+ {%- set system_message = messages[0]['content'] %}
11
+ {%- else %}
12
+ {%- set system_message = messages[0]['content'][0]['text'] %}
13
+ {%- endif %}
14
+ {%- set loop_messages = messages[1:] %}
15
+ {%- else %}
16
+ {%- set system_message = default_system_message %}
17
+ {%- set loop_messages = messages %}
18
+ {%- endif %}
19
+ {{- '[SYSTEM_PROMPT]' + system_message + '[/SYSTEM_PROMPT]' }}
20
+
21
+
22
+ {#- Tool description appended ONLY to last user message. Edits made by Unsloth #}
23
+ {#- Tool description appended also if last message is tool. Edits made by Unsloth #}
24
+ {%- set tools_description = "" %}
25
+ {%- set has_tools = false %}
26
+
27
+ {%- if tools is defined and tools is not none and tools|length > 0 %}
28
+
29
+ {%- set has_tools = true %}
30
+ {%- set tools_description = "[AVAILABLE_TOOLS]" + (tools | tojson) + "[/AVAILABLE_TOOLS]" %}
31
+
32
+ {{- tools_description }}
33
+
34
+ {%- endif %}
35
+
36
+ {%- for message in loop_messages %}
37
+ {%- if message['role'] == 'user' %}
38
+
39
+ {%- if message['content'] is string %}
40
+ {{- '[INST]' + message['content'] + '[/INST]' }}
41
+ {%- else %}
42
+ {{- '[INST]' }}
43
+ {%- for block in message['content'] %}
44
+ {%- if block['type'] == 'text' %}
45
+
46
+ {#- Original did not have content which is weird. Added by Un-sloth. #}
47
+ {%- if block['text'] is defined %}
48
+ {{- block['text'] }}
49
+ {%- else %}
50
+ {{- block['content'] }}
51
+ {%- endif %}
52
+
53
+ {%- elif block['type'] in ['image', 'image_url'] %}
54
+ {{- '[IMG]' }}
55
+ {%- else %}
56
+ {{- raise_exception('Only text and image blocks are supported in message content!') }}
57
+ {%- endif %}
58
+ {%- endfor %}
59
+ {{- '[/INST]' }}
60
+ {%- endif %}
61
+
62
+ {%- elif message['role'] == 'system' %}
63
+ {%- if message['content'] is string %}
64
+ {{- '[SYSTEM_PROMPT]' + message['content'] + '[/SYSTEM_PROMPT]' }}
65
+ {%- else %}
66
+ {{- '[SYSTEM_PROMPT]' + message['content'][0]['text'] + '[/SYSTEM_PROMPT]' }}
67
+ {%- endif %}
68
+
69
+
70
+ {%- elif message['role'] == 'assistant' %}
71
+ {%- if message['content'] is string %}
72
+ {{- message['content'] }}
73
+ {%- else %}
74
+ {{- message['content'][0]['text'] }}
75
+ {%- endif %}
76
+
77
+ {#- If User,Assistant,Tool,Tool we also need to append tools_description. Edits made by Unsloth #}
78
+
79
+ {%- if message['tool_calls'] is defined and message['tool_calls'] is not none %}
80
+ {%- for tool in message['tool_calls'] %}
81
+ {%- set arguments = tool['function']['arguments'] %}
82
+ {%- if arguments is not string %}
83
+ {%- set arguments = arguments|tojson %}
84
+ {%- endif %}
85
+ {#- Must list tool calls AFTER assistant. Edits made by Un-sloth #}
86
+ {{- "[TOOL_CALLS]" + tool['function']['name'] + "[ARGS]" + arguments }}
87
+ {%- endfor %}
88
+ {%- endif %}
89
+
90
+ {{- eos_token }}
91
+
92
+ {%- elif message["role"] == "tool_results" or message["role"] == "tool" %}
93
+ {%- if message.content is defined and message.content.content is defined %}
94
+ {%- set content = message.content.content %}
95
+ {%- else %}
96
+ {%- set content = message.content %}
97
+ {%- endif %}
98
+ {{- "[TOOL_RESULTS]" + content|string + "[/TOOL_RESULTS]" }}
99
+
100
+ {%- else %}
101
+ {{- raise_exception('Only user, systemm assistant and tool roles are supported in the custom template made by Unsloth!') }}
102
+ {%- endif %}
103
+ {%- endfor %}
104
+ {#- Copyright 2025-present the Unsloth team. All rights reserved. #}
105
+ {#- Licensed under the Apache License, Version 2.0 (the "License") #}
config.json ADDED
@@ -0,0 +1,39 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "architectures": [
3
+ "MistralForCausalLM"
4
+ ],
5
+ "attention_dropout": 0.0,
6
+ "bos_token_id": 1,
7
+ "eos_token_id": 2,
8
+ "head_dim": 128,
9
+ "hidden_act": "silu",
10
+ "hidden_size": 5120,
11
+ "initializer_range": 0.02,
12
+ "intermediate_size": 32768,
13
+ "max_position_embeddings": 131072,
14
+ "model_type": "mistral",
15
+ "num_attention_heads": 32,
16
+ "num_hidden_layers": 40,
17
+ "num_key_value_heads": 8,
18
+ "pad_token_id": 11,
19
+ "rms_norm_eps": 1e-05,
20
+ "rope_theta": 1000000000.0,
21
+ "sliding_window": null,
22
+ "tie_word_embeddings": false,
23
+ "torch_dtype": "bfloat16",
24
+ "transformers_version": "4.53.1",
25
+ "unsloth_fixed": true,
26
+ "use_cache": true,
27
+ "vocab_size": 131072,
28
+ "quantization_config": {
29
+ "quant_method": "exl2",
30
+ "version": "0.3.2",
31
+ "bits": 5.0,
32
+ "head_bits": 6,
33
+ "calibration": {
34
+ "rows": 115,
35
+ "length": 2048,
36
+ "dataset": "(default)"
37
+ }
38
+ }
39
+ }
generation_config.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_from_model_config": true,
3
+ "bos_token_id": 1,
4
+ "eos_token_id": 2,
5
+ "max_length": 131072,
6
+ "pad_token_id": 11,
7
+ "transformers_version": "4.53.1"
8
+ }
measurement.json ADDED
The diff for this file is too large to render. See raw diff
 
model.safetensors.index.json ADDED
@@ -0,0 +1,371 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "metadata": {
3
+ "total_parameters": 23572403200,
4
+ "total_size": 47144806400
5
+ },
6
+ "weight_map": {
7
+ "lm_head.weight": "model-00010-of-00010.safetensors",
8
+ "model.embed_tokens.weight": "model-00001-of-00010.safetensors",
9
+ "model.layers.0.input_layernorm.weight": "model-00001-of-00010.safetensors",
10
+ "model.layers.0.mlp.down_proj.weight": "model-00001-of-00010.safetensors",
11
+ "model.layers.0.mlp.gate_proj.weight": "model-00001-of-00010.safetensors",
12
+ "model.layers.0.mlp.up_proj.weight": "model-00001-of-00010.safetensors",
13
+ "model.layers.0.post_attention_layernorm.weight": "model-00001-of-00010.safetensors",
14
+ "model.layers.0.self_attn.k_proj.weight": "model-00001-of-00010.safetensors",
15
+ "model.layers.0.self_attn.o_proj.weight": "model-00001-of-00010.safetensors",
16
+ "model.layers.0.self_attn.q_proj.weight": "model-00001-of-00010.safetensors",
17
+ "model.layers.0.self_attn.v_proj.weight": "model-00001-of-00010.safetensors",
18
+ "model.layers.1.input_layernorm.weight": "model-00001-of-00010.safetensors",
19
+ "model.layers.1.mlp.down_proj.weight": "model-00001-of-00010.safetensors",
20
+ "model.layers.1.mlp.gate_proj.weight": "model-00001-of-00010.safetensors",
21
+ "model.layers.1.mlp.up_proj.weight": "model-00001-of-00010.safetensors",
22
+ "model.layers.1.post_attention_layernorm.weight": "model-00001-of-00010.safetensors",
23
+ "model.layers.1.self_attn.k_proj.weight": "model-00001-of-00010.safetensors",
24
+ "model.layers.1.self_attn.o_proj.weight": "model-00001-of-00010.safetensors",
25
+ "model.layers.1.self_attn.q_proj.weight": "model-00001-of-00010.safetensors",
26
+ "model.layers.1.self_attn.v_proj.weight": "model-00001-of-00010.safetensors",
27
+ "model.layers.10.input_layernorm.weight": "model-00003-of-00010.safetensors",
28
+ "model.layers.10.mlp.down_proj.weight": "model-00003-of-00010.safetensors",
29
+ "model.layers.10.mlp.gate_proj.weight": "model-00003-of-00010.safetensors",
30
+ "model.layers.10.mlp.up_proj.weight": "model-00003-of-00010.safetensors",
31
+ "model.layers.10.post_attention_layernorm.weight": "model-00003-of-00010.safetensors",
32
+ "model.layers.10.self_attn.k_proj.weight": "model-00003-of-00010.safetensors",
33
+ "model.layers.10.self_attn.o_proj.weight": "model-00003-of-00010.safetensors",
34
+ "model.layers.10.self_attn.q_proj.weight": "model-00003-of-00010.safetensors",
35
+ "model.layers.10.self_attn.v_proj.weight": "model-00003-of-00010.safetensors",
36
+ "model.layers.11.input_layernorm.weight": "model-00004-of-00010.safetensors",
37
+ "model.layers.11.mlp.down_proj.weight": "model-00004-of-00010.safetensors",
38
+ "model.layers.11.mlp.gate_proj.weight": "model-00003-of-00010.safetensors",
39
+ "model.layers.11.mlp.up_proj.weight": "model-00003-of-00010.safetensors",
40
+ "model.layers.11.post_attention_layernorm.weight": "model-00004-of-00010.safetensors",
41
+ "model.layers.11.self_attn.k_proj.weight": "model-00003-of-00010.safetensors",
42
+ "model.layers.11.self_attn.o_proj.weight": "model-00003-of-00010.safetensors",
43
+ "model.layers.11.self_attn.q_proj.weight": "model-00003-of-00010.safetensors",
44
+ "model.layers.11.self_attn.v_proj.weight": "model-00003-of-00010.safetensors",
45
+ "model.layers.12.input_layernorm.weight": "model-00004-of-00010.safetensors",
46
+ "model.layers.12.mlp.down_proj.weight": "model-00004-of-00010.safetensors",
47
+ "model.layers.12.mlp.gate_proj.weight": "model-00004-of-00010.safetensors",
48
+ "model.layers.12.mlp.up_proj.weight": "model-00004-of-00010.safetensors",
49
+ "model.layers.12.post_attention_layernorm.weight": "model-00004-of-00010.safetensors",
50
+ "model.layers.12.self_attn.k_proj.weight": "model-00004-of-00010.safetensors",
51
+ "model.layers.12.self_attn.o_proj.weight": "model-00004-of-00010.safetensors",
52
+ "model.layers.12.self_attn.q_proj.weight": "model-00004-of-00010.safetensors",
53
+ "model.layers.12.self_attn.v_proj.weight": "model-00004-of-00010.safetensors",
54
+ "model.layers.13.input_layernorm.weight": "model-00004-of-00010.safetensors",
55
+ "model.layers.13.mlp.down_proj.weight": "model-00004-of-00010.safetensors",
56
+ "model.layers.13.mlp.gate_proj.weight": "model-00004-of-00010.safetensors",
57
+ "model.layers.13.mlp.up_proj.weight": "model-00004-of-00010.safetensors",
58
+ "model.layers.13.post_attention_layernorm.weight": "model-00004-of-00010.safetensors",
59
+ "model.layers.13.self_attn.k_proj.weight": "model-00004-of-00010.safetensors",
60
+ "model.layers.13.self_attn.o_proj.weight": "model-00004-of-00010.safetensors",
61
+ "model.layers.13.self_attn.q_proj.weight": "model-00004-of-00010.safetensors",
62
+ "model.layers.13.self_attn.v_proj.weight": "model-00004-of-00010.safetensors",
63
+ "model.layers.14.input_layernorm.weight": "model-00004-of-00010.safetensors",
64
+ "model.layers.14.mlp.down_proj.weight": "model-00004-of-00010.safetensors",
65
+ "model.layers.14.mlp.gate_proj.weight": "model-00004-of-00010.safetensors",
66
+ "model.layers.14.mlp.up_proj.weight": "model-00004-of-00010.safetensors",
67
+ "model.layers.14.post_attention_layernorm.weight": "model-00004-of-00010.safetensors",
68
+ "model.layers.14.self_attn.k_proj.weight": "model-00004-of-00010.safetensors",
69
+ "model.layers.14.self_attn.o_proj.weight": "model-00004-of-00010.safetensors",
70
+ "model.layers.14.self_attn.q_proj.weight": "model-00004-of-00010.safetensors",
71
+ "model.layers.14.self_attn.v_proj.weight": "model-00004-of-00010.safetensors",
72
+ "model.layers.15.input_layernorm.weight": "model-00004-of-00010.safetensors",
73
+ "model.layers.15.mlp.down_proj.weight": "model-00004-of-00010.safetensors",
74
+ "model.layers.15.mlp.gate_proj.weight": "model-00004-of-00010.safetensors",
75
+ "model.layers.15.mlp.up_proj.weight": "model-00004-of-00010.safetensors",
76
+ "model.layers.15.post_attention_layernorm.weight": "model-00004-of-00010.safetensors",
77
+ "model.layers.15.self_attn.k_proj.weight": "model-00004-of-00010.safetensors",
78
+ "model.layers.15.self_attn.o_proj.weight": "model-00004-of-00010.safetensors",
79
+ "model.layers.15.self_attn.q_proj.weight": "model-00004-of-00010.safetensors",
80
+ "model.layers.15.self_attn.v_proj.weight": "model-00004-of-00010.safetensors",
81
+ "model.layers.16.input_layernorm.weight": "model-00005-of-00010.safetensors",
82
+ "model.layers.16.mlp.down_proj.weight": "model-00005-of-00010.safetensors",
83
+ "model.layers.16.mlp.gate_proj.weight": "model-00005-of-00010.safetensors",
84
+ "model.layers.16.mlp.up_proj.weight": "model-00005-of-00010.safetensors",
85
+ "model.layers.16.post_attention_layernorm.weight": "model-00005-of-00010.safetensors",
86
+ "model.layers.16.self_attn.k_proj.weight": "model-00004-of-00010.safetensors",
87
+ "model.layers.16.self_attn.o_proj.weight": "model-00004-of-00010.safetensors",
88
+ "model.layers.16.self_attn.q_proj.weight": "model-00004-of-00010.safetensors",
89
+ "model.layers.16.self_attn.v_proj.weight": "model-00004-of-00010.safetensors",
90
+ "model.layers.17.input_layernorm.weight": "model-00005-of-00010.safetensors",
91
+ "model.layers.17.mlp.down_proj.weight": "model-00005-of-00010.safetensors",
92
+ "model.layers.17.mlp.gate_proj.weight": "model-00005-of-00010.safetensors",
93
+ "model.layers.17.mlp.up_proj.weight": "model-00005-of-00010.safetensors",
94
+ "model.layers.17.post_attention_layernorm.weight": "model-00005-of-00010.safetensors",
95
+ "model.layers.17.self_attn.k_proj.weight": "model-00005-of-00010.safetensors",
96
+ "model.layers.17.self_attn.o_proj.weight": "model-00005-of-00010.safetensors",
97
+ "model.layers.17.self_attn.q_proj.weight": "model-00005-of-00010.safetensors",
98
+ "model.layers.17.self_attn.v_proj.weight": "model-00005-of-00010.safetensors",
99
+ "model.layers.18.input_layernorm.weight": "model-00005-of-00010.safetensors",
100
+ "model.layers.18.mlp.down_proj.weight": "model-00005-of-00010.safetensors",
101
+ "model.layers.18.mlp.gate_proj.weight": "model-00005-of-00010.safetensors",
102
+ "model.layers.18.mlp.up_proj.weight": "model-00005-of-00010.safetensors",
103
+ "model.layers.18.post_attention_layernorm.weight": "model-00005-of-00010.safetensors",
104
+ "model.layers.18.self_attn.k_proj.weight": "model-00005-of-00010.safetensors",
105
+ "model.layers.18.self_attn.o_proj.weight": "model-00005-of-00010.safetensors",
106
+ "model.layers.18.self_attn.q_proj.weight": "model-00005-of-00010.safetensors",
107
+ "model.layers.18.self_attn.v_proj.weight": "model-00005-of-00010.safetensors",
108
+ "model.layers.19.input_layernorm.weight": "model-00005-of-00010.safetensors",
109
+ "model.layers.19.mlp.down_proj.weight": "model-00005-of-00010.safetensors",
110
+ "model.layers.19.mlp.gate_proj.weight": "model-00005-of-00010.safetensors",
111
+ "model.layers.19.mlp.up_proj.weight": "model-00005-of-00010.safetensors",
112
+ "model.layers.19.post_attention_layernorm.weight": "model-00005-of-00010.safetensors",
113
+ "model.layers.19.self_attn.k_proj.weight": "model-00005-of-00010.safetensors",
114
+ "model.layers.19.self_attn.o_proj.weight": "model-00005-of-00010.safetensors",
115
+ "model.layers.19.self_attn.q_proj.weight": "model-00005-of-00010.safetensors",
116
+ "model.layers.19.self_attn.v_proj.weight": "model-00005-of-00010.safetensors",
117
+ "model.layers.2.input_layernorm.weight": "model-00001-of-00010.safetensors",
118
+ "model.layers.2.mlp.down_proj.weight": "model-00001-of-00010.safetensors",
119
+ "model.layers.2.mlp.gate_proj.weight": "model-00001-of-00010.safetensors",
120
+ "model.layers.2.mlp.up_proj.weight": "model-00001-of-00010.safetensors",
121
+ "model.layers.2.post_attention_layernorm.weight": "model-00001-of-00010.safetensors",
122
+ "model.layers.2.self_attn.k_proj.weight": "model-00001-of-00010.safetensors",
123
+ "model.layers.2.self_attn.o_proj.weight": "model-00001-of-00010.safetensors",
124
+ "model.layers.2.self_attn.q_proj.weight": "model-00001-of-00010.safetensors",
125
+ "model.layers.2.self_attn.v_proj.weight": "model-00001-of-00010.safetensors",
126
+ "model.layers.20.input_layernorm.weight": "model-00006-of-00010.safetensors",
127
+ "model.layers.20.mlp.down_proj.weight": "model-00006-of-00010.safetensors",
128
+ "model.layers.20.mlp.gate_proj.weight": "model-00005-of-00010.safetensors",
129
+ "model.layers.20.mlp.up_proj.weight": "model-00006-of-00010.safetensors",
130
+ "model.layers.20.post_attention_layernorm.weight": "model-00006-of-00010.safetensors",
131
+ "model.layers.20.self_attn.k_proj.weight": "model-00005-of-00010.safetensors",
132
+ "model.layers.20.self_attn.o_proj.weight": "model-00005-of-00010.safetensors",
133
+ "model.layers.20.self_attn.q_proj.weight": "model-00005-of-00010.safetensors",
134
+ "model.layers.20.self_attn.v_proj.weight": "model-00005-of-00010.safetensors",
135
+ "model.layers.21.input_layernorm.weight": "model-00006-of-00010.safetensors",
136
+ "model.layers.21.mlp.down_proj.weight": "model-00006-of-00010.safetensors",
137
+ "model.layers.21.mlp.gate_proj.weight": "model-00006-of-00010.safetensors",
138
+ "model.layers.21.mlp.up_proj.weight": "model-00006-of-00010.safetensors",
139
+ "model.layers.21.post_attention_layernorm.weight": "model-00006-of-00010.safetensors",
140
+ "model.layers.21.self_attn.k_proj.weight": "model-00006-of-00010.safetensors",
141
+ "model.layers.21.self_attn.o_proj.weight": "model-00006-of-00010.safetensors",
142
+ "model.layers.21.self_attn.q_proj.weight": "model-00006-of-00010.safetensors",
143
+ "model.layers.21.self_attn.v_proj.weight": "model-00006-of-00010.safetensors",
144
+ "model.layers.22.input_layernorm.weight": "model-00006-of-00010.safetensors",
145
+ "model.layers.22.mlp.down_proj.weight": "model-00006-of-00010.safetensors",
146
+ "model.layers.22.mlp.gate_proj.weight": "model-00006-of-00010.safetensors",
147
+ "model.layers.22.mlp.up_proj.weight": "model-00006-of-00010.safetensors",
148
+ "model.layers.22.post_attention_layernorm.weight": "model-00006-of-00010.safetensors",
149
+ "model.layers.22.self_attn.k_proj.weight": "model-00006-of-00010.safetensors",
150
+ "model.layers.22.self_attn.o_proj.weight": "model-00006-of-00010.safetensors",
151
+ "model.layers.22.self_attn.q_proj.weight": "model-00006-of-00010.safetensors",
152
+ "model.layers.22.self_attn.v_proj.weight": "model-00006-of-00010.safetensors",
153
+ "model.layers.23.input_layernorm.weight": "model-00006-of-00010.safetensors",
154
+ "model.layers.23.mlp.down_proj.weight": "model-00006-of-00010.safetensors",
155
+ "model.layers.23.mlp.gate_proj.weight": "model-00006-of-00010.safetensors",
156
+ "model.layers.23.mlp.up_proj.weight": "model-00006-of-00010.safetensors",
157
+ "model.layers.23.post_attention_layernorm.weight": "model-00006-of-00010.safetensors",
158
+ "model.layers.23.self_attn.k_proj.weight": "model-00006-of-00010.safetensors",
159
+ "model.layers.23.self_attn.o_proj.weight": "model-00006-of-00010.safetensors",
160
+ "model.layers.23.self_attn.q_proj.weight": "model-00006-of-00010.safetensors",
161
+ "model.layers.23.self_attn.v_proj.weight": "model-00006-of-00010.safetensors",
162
+ "model.layers.24.input_layernorm.weight": "model-00007-of-00010.safetensors",
163
+ "model.layers.24.mlp.down_proj.weight": "model-00007-of-00010.safetensors",
164
+ "model.layers.24.mlp.gate_proj.weight": "model-00006-of-00010.safetensors",
165
+ "model.layers.24.mlp.up_proj.weight": "model-00006-of-00010.safetensors",
166
+ "model.layers.24.post_attention_layernorm.weight": "model-00007-of-00010.safetensors",
167
+ "model.layers.24.self_attn.k_proj.weight": "model-00006-of-00010.safetensors",
168
+ "model.layers.24.self_attn.o_proj.weight": "model-00006-of-00010.safetensors",
169
+ "model.layers.24.self_attn.q_proj.weight": "model-00006-of-00010.safetensors",
170
+ "model.layers.24.self_attn.v_proj.weight": "model-00006-of-00010.safetensors",
171
+ "model.layers.25.input_layernorm.weight": "model-00007-of-00010.safetensors",
172
+ "model.layers.25.mlp.down_proj.weight": "model-00007-of-00010.safetensors",
173
+ "model.layers.25.mlp.gate_proj.weight": "model-00007-of-00010.safetensors",
174
+ "model.layers.25.mlp.up_proj.weight": "model-00007-of-00010.safetensors",
175
+ "model.layers.25.post_attention_layernorm.weight": "model-00007-of-00010.safetensors",
176
+ "model.layers.25.self_attn.k_proj.weight": "model-00007-of-00010.safetensors",
177
+ "model.layers.25.self_attn.o_proj.weight": "model-00007-of-00010.safetensors",
178
+ "model.layers.25.self_attn.q_proj.weight": "model-00007-of-00010.safetensors",
179
+ "model.layers.25.self_attn.v_proj.weight": "model-00007-of-00010.safetensors",
180
+ "model.layers.26.input_layernorm.weight": "model-00007-of-00010.safetensors",
181
+ "model.layers.26.mlp.down_proj.weight": "model-00007-of-00010.safetensors",
182
+ "model.layers.26.mlp.gate_proj.weight": "model-00007-of-00010.safetensors",
183
+ "model.layers.26.mlp.up_proj.weight": "model-00007-of-00010.safetensors",
184
+ "model.layers.26.post_attention_layernorm.weight": "model-00007-of-00010.safetensors",
185
+ "model.layers.26.self_attn.k_proj.weight": "model-00007-of-00010.safetensors",
186
+ "model.layers.26.self_attn.o_proj.weight": "model-00007-of-00010.safetensors",
187
+ "model.layers.26.self_attn.q_proj.weight": "model-00007-of-00010.safetensors",
188
+ "model.layers.26.self_attn.v_proj.weight": "model-00007-of-00010.safetensors",
189
+ "model.layers.27.input_layernorm.weight": "model-00007-of-00010.safetensors",
190
+ "model.layers.27.mlp.down_proj.weight": "model-00007-of-00010.safetensors",
191
+ "model.layers.27.mlp.gate_proj.weight": "model-00007-of-00010.safetensors",
192
+ "model.layers.27.mlp.up_proj.weight": "model-00007-of-00010.safetensors",
193
+ "model.layers.27.post_attention_layernorm.weight": "model-00007-of-00010.safetensors",
194
+ "model.layers.27.self_attn.k_proj.weight": "model-00007-of-00010.safetensors",
195
+ "model.layers.27.self_attn.o_proj.weight": "model-00007-of-00010.safetensors",
196
+ "model.layers.27.self_attn.q_proj.weight": "model-00007-of-00010.safetensors",
197
+ "model.layers.27.self_attn.v_proj.weight": "model-00007-of-00010.safetensors",
198
+ "model.layers.28.input_layernorm.weight": "model-00007-of-00010.safetensors",
199
+ "model.layers.28.mlp.down_proj.weight": "model-00007-of-00010.safetensors",
200
+ "model.layers.28.mlp.gate_proj.weight": "model-00007-of-00010.safetensors",
201
+ "model.layers.28.mlp.up_proj.weight": "model-00007-of-00010.safetensors",
202
+ "model.layers.28.post_attention_layernorm.weight": "model-00007-of-00010.safetensors",
203
+ "model.layers.28.self_attn.k_proj.weight": "model-00007-of-00010.safetensors",
204
+ "model.layers.28.self_attn.o_proj.weight": "model-00007-of-00010.safetensors",
205
+ "model.layers.28.self_attn.q_proj.weight": "model-00007-of-00010.safetensors",
206
+ "model.layers.28.self_attn.v_proj.weight": "model-00007-of-00010.safetensors",
207
+ "model.layers.29.input_layernorm.weight": "model-00008-of-00010.safetensors",
208
+ "model.layers.29.mlp.down_proj.weight": "model-00008-of-00010.safetensors",
209
+ "model.layers.29.mlp.gate_proj.weight": "model-00008-of-00010.safetensors",
210
+ "model.layers.29.mlp.up_proj.weight": "model-00008-of-00010.safetensors",
211
+ "model.layers.29.post_attention_layernorm.weight": "model-00008-of-00010.safetensors",
212
+ "model.layers.29.self_attn.k_proj.weight": "model-00007-of-00010.safetensors",
213
+ "model.layers.29.self_attn.o_proj.weight": "model-00007-of-00010.safetensors",
214
+ "model.layers.29.self_attn.q_proj.weight": "model-00007-of-00010.safetensors",
215
+ "model.layers.29.self_attn.v_proj.weight": "model-00007-of-00010.safetensors",
216
+ "model.layers.3.input_layernorm.weight": "model-00002-of-00010.safetensors",
217
+ "model.layers.3.mlp.down_proj.weight": "model-00002-of-00010.safetensors",
218
+ "model.layers.3.mlp.gate_proj.weight": "model-00002-of-00010.safetensors",
219
+ "model.layers.3.mlp.up_proj.weight": "model-00002-of-00010.safetensors",
220
+ "model.layers.3.post_attention_layernorm.weight": "model-00002-of-00010.safetensors",
221
+ "model.layers.3.self_attn.k_proj.weight": "model-00001-of-00010.safetensors",
222
+ "model.layers.3.self_attn.o_proj.weight": "model-00001-of-00010.safetensors",
223
+ "model.layers.3.self_attn.q_proj.weight": "model-00001-of-00010.safetensors",
224
+ "model.layers.3.self_attn.v_proj.weight": "model-00001-of-00010.safetensors",
225
+ "model.layers.30.input_layernorm.weight": "model-00008-of-00010.safetensors",
226
+ "model.layers.30.mlp.down_proj.weight": "model-00008-of-00010.safetensors",
227
+ "model.layers.30.mlp.gate_proj.weight": "model-00008-of-00010.safetensors",
228
+ "model.layers.30.mlp.up_proj.weight": "model-00008-of-00010.safetensors",
229
+ "model.layers.30.post_attention_layernorm.weight": "model-00008-of-00010.safetensors",
230
+ "model.layers.30.self_attn.k_proj.weight": "model-00008-of-00010.safetensors",
231
+ "model.layers.30.self_attn.o_proj.weight": "model-00008-of-00010.safetensors",
232
+ "model.layers.30.self_attn.q_proj.weight": "model-00008-of-00010.safetensors",
233
+ "model.layers.30.self_attn.v_proj.weight": "model-00008-of-00010.safetensors",
234
+ "model.layers.31.input_layernorm.weight": "model-00008-of-00010.safetensors",
235
+ "model.layers.31.mlp.down_proj.weight": "model-00008-of-00010.safetensors",
236
+ "model.layers.31.mlp.gate_proj.weight": "model-00008-of-00010.safetensors",
237
+ "model.layers.31.mlp.up_proj.weight": "model-00008-of-00010.safetensors",
238
+ "model.layers.31.post_attention_layernorm.weight": "model-00008-of-00010.safetensors",
239
+ "model.layers.31.self_attn.k_proj.weight": "model-00008-of-00010.safetensors",
240
+ "model.layers.31.self_attn.o_proj.weight": "model-00008-of-00010.safetensors",
241
+ "model.layers.31.self_attn.q_proj.weight": "model-00008-of-00010.safetensors",
242
+ "model.layers.31.self_attn.v_proj.weight": "model-00008-of-00010.safetensors",
243
+ "model.layers.32.input_layernorm.weight": "model-00008-of-00010.safetensors",
244
+ "model.layers.32.mlp.down_proj.weight": "model-00008-of-00010.safetensors",
245
+ "model.layers.32.mlp.gate_proj.weight": "model-00008-of-00010.safetensors",
246
+ "model.layers.32.mlp.up_proj.weight": "model-00008-of-00010.safetensors",
247
+ "model.layers.32.post_attention_layernorm.weight": "model-00008-of-00010.safetensors",
248
+ "model.layers.32.self_attn.k_proj.weight": "model-00008-of-00010.safetensors",
249
+ "model.layers.32.self_attn.o_proj.weight": "model-00008-of-00010.safetensors",
250
+ "model.layers.32.self_attn.q_proj.weight": "model-00008-of-00010.safetensors",
251
+ "model.layers.32.self_attn.v_proj.weight": "model-00008-of-00010.safetensors",
252
+ "model.layers.33.input_layernorm.weight": "model-00009-of-00010.safetensors",
253
+ "model.layers.33.mlp.down_proj.weight": "model-00009-of-00010.safetensors",
254
+ "model.layers.33.mlp.gate_proj.weight": "model-00008-of-00010.safetensors",
255
+ "model.layers.33.mlp.up_proj.weight": "model-00009-of-00010.safetensors",
256
+ "model.layers.33.post_attention_layernorm.weight": "model-00009-of-00010.safetensors",
257
+ "model.layers.33.self_attn.k_proj.weight": "model-00008-of-00010.safetensors",
258
+ "model.layers.33.self_attn.o_proj.weight": "model-00008-of-00010.safetensors",
259
+ "model.layers.33.self_attn.q_proj.weight": "model-00008-of-00010.safetensors",
260
+ "model.layers.33.self_attn.v_proj.weight": "model-00008-of-00010.safetensors",
261
+ "model.layers.34.input_layernorm.weight": "model-00009-of-00010.safetensors",
262
+ "model.layers.34.mlp.down_proj.weight": "model-00009-of-00010.safetensors",
263
+ "model.layers.34.mlp.gate_proj.weight": "model-00009-of-00010.safetensors",
264
+ "model.layers.34.mlp.up_proj.weight": "model-00009-of-00010.safetensors",
265
+ "model.layers.34.post_attention_layernorm.weight": "model-00009-of-00010.safetensors",
266
+ "model.layers.34.self_attn.k_proj.weight": "model-00009-of-00010.safetensors",
267
+ "model.layers.34.self_attn.o_proj.weight": "model-00009-of-00010.safetensors",
268
+ "model.layers.34.self_attn.q_proj.weight": "model-00009-of-00010.safetensors",
269
+ "model.layers.34.self_attn.v_proj.weight": "model-00009-of-00010.safetensors",
270
+ "model.layers.35.input_layernorm.weight": "model-00009-of-00010.safetensors",
271
+ "model.layers.35.mlp.down_proj.weight": "model-00009-of-00010.safetensors",
272
+ "model.layers.35.mlp.gate_proj.weight": "model-00009-of-00010.safetensors",
273
+ "model.layers.35.mlp.up_proj.weight": "model-00009-of-00010.safetensors",
274
+ "model.layers.35.post_attention_layernorm.weight": "model-00009-of-00010.safetensors",
275
+ "model.layers.35.self_attn.k_proj.weight": "model-00009-of-00010.safetensors",
276
+ "model.layers.35.self_attn.o_proj.weight": "model-00009-of-00010.safetensors",
277
+ "model.layers.35.self_attn.q_proj.weight": "model-00009-of-00010.safetensors",
278
+ "model.layers.35.self_attn.v_proj.weight": "model-00009-of-00010.safetensors",
279
+ "model.layers.36.input_layernorm.weight": "model-00009-of-00010.safetensors",
280
+ "model.layers.36.mlp.down_proj.weight": "model-00009-of-00010.safetensors",
281
+ "model.layers.36.mlp.gate_proj.weight": "model-00009-of-00010.safetensors",
282
+ "model.layers.36.mlp.up_proj.weight": "model-00009-of-00010.safetensors",
283
+ "model.layers.36.post_attention_layernorm.weight": "model-00009-of-00010.safetensors",
284
+ "model.layers.36.self_attn.k_proj.weight": "model-00009-of-00010.safetensors",
285
+ "model.layers.36.self_attn.o_proj.weight": "model-00009-of-00010.safetensors",
286
+ "model.layers.36.self_attn.q_proj.weight": "model-00009-of-00010.safetensors",
287
+ "model.layers.36.self_attn.v_proj.weight": "model-00009-of-00010.safetensors",
288
+ "model.layers.37.input_layernorm.weight": "model-00010-of-00010.safetensors",
289
+ "model.layers.37.mlp.down_proj.weight": "model-00010-of-00010.safetensors",
290
+ "model.layers.37.mlp.gate_proj.weight": "model-00009-of-00010.safetensors",
291
+ "model.layers.37.mlp.up_proj.weight": "model-00009-of-00010.safetensors",
292
+ "model.layers.37.post_attention_layernorm.weight": "model-00010-of-00010.safetensors",
293
+ "model.layers.37.self_attn.k_proj.weight": "model-00009-of-00010.safetensors",
294
+ "model.layers.37.self_attn.o_proj.weight": "model-00009-of-00010.safetensors",
295
+ "model.layers.37.self_attn.q_proj.weight": "model-00009-of-00010.safetensors",
296
+ "model.layers.37.self_attn.v_proj.weight": "model-00009-of-00010.safetensors",
297
+ "model.layers.38.input_layernorm.weight": "model-00010-of-00010.safetensors",
298
+ "model.layers.38.mlp.down_proj.weight": "model-00010-of-00010.safetensors",
299
+ "model.layers.38.mlp.gate_proj.weight": "model-00010-of-00010.safetensors",
300
+ "model.layers.38.mlp.up_proj.weight": "model-00010-of-00010.safetensors",
301
+ "model.layers.38.post_attention_layernorm.weight": "model-00010-of-00010.safetensors",
302
+ "model.layers.38.self_attn.k_proj.weight": "model-00010-of-00010.safetensors",
303
+ "model.layers.38.self_attn.o_proj.weight": "model-00010-of-00010.safetensors",
304
+ "model.layers.38.self_attn.q_proj.weight": "model-00010-of-00010.safetensors",
305
+ "model.layers.38.self_attn.v_proj.weight": "model-00010-of-00010.safetensors",
306
+ "model.layers.39.input_layernorm.weight": "model-00010-of-00010.safetensors",
307
+ "model.layers.39.mlp.down_proj.weight": "model-00010-of-00010.safetensors",
308
+ "model.layers.39.mlp.gate_proj.weight": "model-00010-of-00010.safetensors",
309
+ "model.layers.39.mlp.up_proj.weight": "model-00010-of-00010.safetensors",
310
+ "model.layers.39.post_attention_layernorm.weight": "model-00010-of-00010.safetensors",
311
+ "model.layers.39.self_attn.k_proj.weight": "model-00010-of-00010.safetensors",
312
+ "model.layers.39.self_attn.o_proj.weight": "model-00010-of-00010.safetensors",
313
+ "model.layers.39.self_attn.q_proj.weight": "model-00010-of-00010.safetensors",
314
+ "model.layers.39.self_attn.v_proj.weight": "model-00010-of-00010.safetensors",
315
+ "model.layers.4.input_layernorm.weight": "model-00002-of-00010.safetensors",
316
+ "model.layers.4.mlp.down_proj.weight": "model-00002-of-00010.safetensors",
317
+ "model.layers.4.mlp.gate_proj.weight": "model-00002-of-00010.safetensors",
318
+ "model.layers.4.mlp.up_proj.weight": "model-00002-of-00010.safetensors",
319
+ "model.layers.4.post_attention_layernorm.weight": "model-00002-of-00010.safetensors",
320
+ "model.layers.4.self_attn.k_proj.weight": "model-00002-of-00010.safetensors",
321
+ "model.layers.4.self_attn.o_proj.weight": "model-00002-of-00010.safetensors",
322
+ "model.layers.4.self_attn.q_proj.weight": "model-00002-of-00010.safetensors",
323
+ "model.layers.4.self_attn.v_proj.weight": "model-00002-of-00010.safetensors",
324
+ "model.layers.5.input_layernorm.weight": "model-00002-of-00010.safetensors",
325
+ "model.layers.5.mlp.down_proj.weight": "model-00002-of-00010.safetensors",
326
+ "model.layers.5.mlp.gate_proj.weight": "model-00002-of-00010.safetensors",
327
+ "model.layers.5.mlp.up_proj.weight": "model-00002-of-00010.safetensors",
328
+ "model.layers.5.post_attention_layernorm.weight": "model-00002-of-00010.safetensors",
329
+ "model.layers.5.self_attn.k_proj.weight": "model-00002-of-00010.safetensors",
330
+ "model.layers.5.self_attn.o_proj.weight": "model-00002-of-00010.safetensors",
331
+ "model.layers.5.self_attn.q_proj.weight": "model-00002-of-00010.safetensors",
332
+ "model.layers.5.self_attn.v_proj.weight": "model-00002-of-00010.safetensors",
333
+ "model.layers.6.input_layernorm.weight": "model-00002-of-00010.safetensors",
334
+ "model.layers.6.mlp.down_proj.weight": "model-00002-of-00010.safetensors",
335
+ "model.layers.6.mlp.gate_proj.weight": "model-00002-of-00010.safetensors",
336
+ "model.layers.6.mlp.up_proj.weight": "model-00002-of-00010.safetensors",
337
+ "model.layers.6.post_attention_layernorm.weight": "model-00002-of-00010.safetensors",
338
+ "model.layers.6.self_attn.k_proj.weight": "model-00002-of-00010.safetensors",
339
+ "model.layers.6.self_attn.o_proj.weight": "model-00002-of-00010.safetensors",
340
+ "model.layers.6.self_attn.q_proj.weight": "model-00002-of-00010.safetensors",
341
+ "model.layers.6.self_attn.v_proj.weight": "model-00002-of-00010.safetensors",
342
+ "model.layers.7.input_layernorm.weight": "model-00003-of-00010.safetensors",
343
+ "model.layers.7.mlp.down_proj.weight": "model-00003-of-00010.safetensors",
344
+ "model.layers.7.mlp.gate_proj.weight": "model-00002-of-00010.safetensors",
345
+ "model.layers.7.mlp.up_proj.weight": "model-00003-of-00010.safetensors",
346
+ "model.layers.7.post_attention_layernorm.weight": "model-00003-of-00010.safetensors",
347
+ "model.layers.7.self_attn.k_proj.weight": "model-00002-of-00010.safetensors",
348
+ "model.layers.7.self_attn.o_proj.weight": "model-00002-of-00010.safetensors",
349
+ "model.layers.7.self_attn.q_proj.weight": "model-00002-of-00010.safetensors",
350
+ "model.layers.7.self_attn.v_proj.weight": "model-00002-of-00010.safetensors",
351
+ "model.layers.8.input_layernorm.weight": "model-00003-of-00010.safetensors",
352
+ "model.layers.8.mlp.down_proj.weight": "model-00003-of-00010.safetensors",
353
+ "model.layers.8.mlp.gate_proj.weight": "model-00003-of-00010.safetensors",
354
+ "model.layers.8.mlp.up_proj.weight": "model-00003-of-00010.safetensors",
355
+ "model.layers.8.post_attention_layernorm.weight": "model-00003-of-00010.safetensors",
356
+ "model.layers.8.self_attn.k_proj.weight": "model-00003-of-00010.safetensors",
357
+ "model.layers.8.self_attn.o_proj.weight": "model-00003-of-00010.safetensors",
358
+ "model.layers.8.self_attn.q_proj.weight": "model-00003-of-00010.safetensors",
359
+ "model.layers.8.self_attn.v_proj.weight": "model-00003-of-00010.safetensors",
360
+ "model.layers.9.input_layernorm.weight": "model-00003-of-00010.safetensors",
361
+ "model.layers.9.mlp.down_proj.weight": "model-00003-of-00010.safetensors",
362
+ "model.layers.9.mlp.gate_proj.weight": "model-00003-of-00010.safetensors",
363
+ "model.layers.9.mlp.up_proj.weight": "model-00003-of-00010.safetensors",
364
+ "model.layers.9.post_attention_layernorm.weight": "model-00003-of-00010.safetensors",
365
+ "model.layers.9.self_attn.k_proj.weight": "model-00003-of-00010.safetensors",
366
+ "model.layers.9.self_attn.o_proj.weight": "model-00003-of-00010.safetensors",
367
+ "model.layers.9.self_attn.q_proj.weight": "model-00003-of-00010.safetensors",
368
+ "model.layers.9.self_attn.v_proj.weight": "model-00003-of-00010.safetensors",
369
+ "model.norm.weight": "model-00010-of-00010.safetensors"
370
+ }
371
+ }
output-00001-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1083fb6471fe17187d008e609614b11214cecff2df3db08b346a6c49999d0ba3
3
+ size 10708726542
output-00002-of-00002.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:65e25331ba30228b7afedc840783c86b3f10e3ffc807df2a5c7da0945b3aa516
3
+ size 5059089386
params.json ADDED
@@ -0,0 +1,11 @@
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "dim": 5120,
3
+ "n_layers": 40,
4
+ "head_dim": 128,
5
+ "hidden_dim": 32768,
6
+ "n_heads": 32,
7
+ "n_kv_heads": 8,
8
+ "rope_theta": 1000000000.0,
9
+ "norm_eps": 1e-05,
10
+ "vocab_size": 131072
11
+ }
special_tokens_map.json ADDED
@@ -0,0 +1,1032 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "additional_special_tokens": [
3
+ "<unk>",
4
+ "<s>",
5
+ "</s>",
6
+ "[INST]",
7
+ "[/INST]",
8
+ "[AVAILABLE_TOOLS]",
9
+ "[/AVAILABLE_TOOLS]",
10
+ "[TOOL_RESULTS]",
11
+ "[/TOOL_RESULTS]",
12
+ "[TOOL_CALLS]",
13
+ "[IMG]",
14
+ "<pad>",
15
+ "[IMG_BREAK]",
16
+ "[IMG_END]",
17
+ "[PREFIX]",
18
+ "[MIDDLE]",
19
+ "[SUFFIX]",
20
+ "[SYSTEM_PROMPT]",
21
+ "[/SYSTEM_PROMPT]",
22
+ "[TOOL_CONTENT]",
23
+ "<SPECIAL_20>",
24
+ "<SPECIAL_21>",
25
+ "<SPECIAL_22>",
26
+ "<SPECIAL_23>",
27
+ "<SPECIAL_24>",
28
+ "<SPECIAL_25>",
29
+ "<SPECIAL_26>",
30
+ "<SPECIAL_27>",
31
+ "<SPECIAL_28>",
32
+ "<SPECIAL_29>",
33
+ "<SPECIAL_30>",
34
+ "<SPECIAL_31>",
35
+ "[ARGS]",
36
+ "[CALL_ID]",
37
+ "<SPECIAL_34>",
38
+ "<SPECIAL_35>",
39
+ "<SPECIAL_36>",
40
+ "<SPECIAL_37>",
41
+ "<SPECIAL_38>",
42
+ "<SPECIAL_39>",
43
+ "<SPECIAL_40>",
44
+ "<SPECIAL_41>",
45
+ "<SPECIAL_42>",
46
+ "<SPECIAL_43>",
47
+ "<SPECIAL_44>",
48
+ "<SPECIAL_45>",
49
+ "<SPECIAL_46>",
50
+ "<SPECIAL_47>",
51
+ "<SPECIAL_48>",
52
+ "<SPECIAL_49>",
53
+ "<SPECIAL_50>",
54
+ "<SPECIAL_51>",
55
+ "<SPECIAL_52>",
56
+ "<SPECIAL_53>",
57
+ "<SPECIAL_54>",
58
+ "<SPECIAL_55>",
59
+ "<SPECIAL_56>",
60
+ "<SPECIAL_57>",
61
+ "<SPECIAL_58>",
62
+ "<SPECIAL_59>",
63
+ "<SPECIAL_60>",
64
+ "<SPECIAL_61>",
65
+ "<SPECIAL_62>",
66
+ "<SPECIAL_63>",
67
+ "<SPECIAL_64>",
68
+ "<SPECIAL_65>",
69
+ "<SPECIAL_66>",
70
+ "<SPECIAL_67>",
71
+ "<SPECIAL_68>",
72
+ "<SPECIAL_69>",
73
+ "<SPECIAL_70>",
74
+ "<SPECIAL_71>",
75
+ "<SPECIAL_72>",
76
+ "<SPECIAL_73>",
77
+ "<SPECIAL_74>",
78
+ "<SPECIAL_75>",
79
+ "<SPECIAL_76>",
80
+ "<SPECIAL_77>",
81
+ "<SPECIAL_78>",
82
+ "<SPECIAL_79>",
83
+ "<SPECIAL_80>",
84
+ "<SPECIAL_81>",
85
+ "<SPECIAL_82>",
86
+ "<SPECIAL_83>",
87
+ "<SPECIAL_84>",
88
+ "<SPECIAL_85>",
89
+ "<SPECIAL_86>",
90
+ "<SPECIAL_87>",
91
+ "<SPECIAL_88>",
92
+ "<SPECIAL_89>",
93
+ "<SPECIAL_90>",
94
+ "<SPECIAL_91>",
95
+ "<SPECIAL_92>",
96
+ "<SPECIAL_93>",
97
+ "<SPECIAL_94>",
98
+ "<SPECIAL_95>",
99
+ "<SPECIAL_96>",
100
+ "<SPECIAL_97>",
101
+ "<SPECIAL_98>",
102
+ "<SPECIAL_99>",
103
+ "<SPECIAL_100>",
104
+ "<SPECIAL_101>",
105
+ "<SPECIAL_102>",
106
+ "<SPECIAL_103>",
107
+ "<SPECIAL_104>",
108
+ "<SPECIAL_105>",
109
+ "<SPECIAL_106>",
110
+ "<SPECIAL_107>",
111
+ "<SPECIAL_108>",
112
+ "<SPECIAL_109>",
113
+ "<SPECIAL_110>",
114
+ "<SPECIAL_111>",
115
+ "<SPECIAL_112>",
116
+ "<SPECIAL_113>",
117
+ "<SPECIAL_114>",
118
+ "<SPECIAL_115>",
119
+ "<SPECIAL_116>",
120
+ "<SPECIAL_117>",
121
+ "<SPECIAL_118>",
122
+ "<SPECIAL_119>",
123
+ "<SPECIAL_120>",
124
+ "<SPECIAL_121>",
125
+ "<SPECIAL_122>",
126
+ "<SPECIAL_123>",
127
+ "<SPECIAL_124>",
128
+ "<SPECIAL_125>",
129
+ "<SPECIAL_126>",
130
+ "<SPECIAL_127>",
131
+ "<SPECIAL_128>",
132
+ "<SPECIAL_129>",
133
+ "<SPECIAL_130>",
134
+ "<SPECIAL_131>",
135
+ "<SPECIAL_132>",
136
+ "<SPECIAL_133>",
137
+ "<SPECIAL_134>",
138
+ "<SPECIAL_135>",
139
+ "<SPECIAL_136>",
140
+ "<SPECIAL_137>",
141
+ "<SPECIAL_138>",
142
+ "<SPECIAL_139>",
143
+ "<SPECIAL_140>",
144
+ "<SPECIAL_141>",
145
+ "<SPECIAL_142>",
146
+ "<SPECIAL_143>",
147
+ "<SPECIAL_144>",
148
+ "<SPECIAL_145>",
149
+ "<SPECIAL_146>",
150
+ "<SPECIAL_147>",
151
+ "<SPECIAL_148>",
152
+ "<SPECIAL_149>",
153
+ "<SPECIAL_150>",
154
+ "<SPECIAL_151>",
155
+ "<SPECIAL_152>",
156
+ "<SPECIAL_153>",
157
+ "<SPECIAL_154>",
158
+ "<SPECIAL_155>",
159
+ "<SPECIAL_156>",
160
+ "<SPECIAL_157>",
161
+ "<SPECIAL_158>",
162
+ "<SPECIAL_159>",
163
+ "<SPECIAL_160>",
164
+ "<SPECIAL_161>",
165
+ "<SPECIAL_162>",
166
+ "<SPECIAL_163>",
167
+ "<SPECIAL_164>",
168
+ "<SPECIAL_165>",
169
+ "<SPECIAL_166>",
170
+ "<SPECIAL_167>",
171
+ "<SPECIAL_168>",
172
+ "<SPECIAL_169>",
173
+ "<SPECIAL_170>",
174
+ "<SPECIAL_171>",
175
+ "<SPECIAL_172>",
176
+ "<SPECIAL_173>",
177
+ "<SPECIAL_174>",
178
+ "<SPECIAL_175>",
179
+ "<SPECIAL_176>",
180
+ "<SPECIAL_177>",
181
+ "<SPECIAL_178>",
182
+ "<SPECIAL_179>",
183
+ "<SPECIAL_180>",
184
+ "<SPECIAL_181>",
185
+ "<SPECIAL_182>",
186
+ "<SPECIAL_183>",
187
+ "<SPECIAL_184>",
188
+ "<SPECIAL_185>",
189
+ "<SPECIAL_186>",
190
+ "<SPECIAL_187>",
191
+ "<SPECIAL_188>",
192
+ "<SPECIAL_189>",
193
+ "<SPECIAL_190>",
194
+ "<SPECIAL_191>",
195
+ "<SPECIAL_192>",
196
+ "<SPECIAL_193>",
197
+ "<SPECIAL_194>",
198
+ "<SPECIAL_195>",
199
+ "<SPECIAL_196>",
200
+ "<SPECIAL_197>",
201
+ "<SPECIAL_198>",
202
+ "<SPECIAL_199>",
203
+ "<SPECIAL_200>",
204
+ "<SPECIAL_201>",
205
+ "<SPECIAL_202>",
206
+ "<SPECIAL_203>",
207
+ "<SPECIAL_204>",
208
+ "<SPECIAL_205>",
209
+ "<SPECIAL_206>",
210
+ "<SPECIAL_207>",
211
+ "<SPECIAL_208>",
212
+ "<SPECIAL_209>",
213
+ "<SPECIAL_210>",
214
+ "<SPECIAL_211>",
215
+ "<SPECIAL_212>",
216
+ "<SPECIAL_213>",
217
+ "<SPECIAL_214>",
218
+ "<SPECIAL_215>",
219
+ "<SPECIAL_216>",
220
+ "<SPECIAL_217>",
221
+ "<SPECIAL_218>",
222
+ "<SPECIAL_219>",
223
+ "<SPECIAL_220>",
224
+ "<SPECIAL_221>",
225
+ "<SPECIAL_222>",
226
+ "<SPECIAL_223>",
227
+ "<SPECIAL_224>",
228
+ "<SPECIAL_225>",
229
+ "<SPECIAL_226>",
230
+ "<SPECIAL_227>",
231
+ "<SPECIAL_228>",
232
+ "<SPECIAL_229>",
233
+ "<SPECIAL_230>",
234
+ "<SPECIAL_231>",
235
+ "<SPECIAL_232>",
236
+ "<SPECIAL_233>",
237
+ "<SPECIAL_234>",
238
+ "<SPECIAL_235>",
239
+ "<SPECIAL_236>",
240
+ "<SPECIAL_237>",
241
+ "<SPECIAL_238>",
242
+ "<SPECIAL_239>",
243
+ "<SPECIAL_240>",
244
+ "<SPECIAL_241>",
245
+ "<SPECIAL_242>",
246
+ "<SPECIAL_243>",
247
+ "<SPECIAL_244>",
248
+ "<SPECIAL_245>",
249
+ "<SPECIAL_246>",
250
+ "<SPECIAL_247>",
251
+ "<SPECIAL_248>",
252
+ "<SPECIAL_249>",
253
+ "<SPECIAL_250>",
254
+ "<SPECIAL_251>",
255
+ "<SPECIAL_252>",
256
+ "<SPECIAL_253>",
257
+ "<SPECIAL_254>",
258
+ "<SPECIAL_255>",
259
+ "<SPECIAL_256>",
260
+ "<SPECIAL_257>",
261
+ "<SPECIAL_258>",
262
+ "<SPECIAL_259>",
263
+ "<SPECIAL_260>",
264
+ "<SPECIAL_261>",
265
+ "<SPECIAL_262>",
266
+ "<SPECIAL_263>",
267
+ "<SPECIAL_264>",
268
+ "<SPECIAL_265>",
269
+ "<SPECIAL_266>",
270
+ "<SPECIAL_267>",
271
+ "<SPECIAL_268>",
272
+ "<SPECIAL_269>",
273
+ "<SPECIAL_270>",
274
+ "<SPECIAL_271>",
275
+ "<SPECIAL_272>",
276
+ "<SPECIAL_273>",
277
+ "<SPECIAL_274>",
278
+ "<SPECIAL_275>",
279
+ "<SPECIAL_276>",
280
+ "<SPECIAL_277>",
281
+ "<SPECIAL_278>",
282
+ "<SPECIAL_279>",
283
+ "<SPECIAL_280>",
284
+ "<SPECIAL_281>",
285
+ "<SPECIAL_282>",
286
+ "<SPECIAL_283>",
287
+ "<SPECIAL_284>",
288
+ "<SPECIAL_285>",
289
+ "<SPECIAL_286>",
290
+ "<SPECIAL_287>",
291
+ "<SPECIAL_288>",
292
+ "<SPECIAL_289>",
293
+ "<SPECIAL_290>",
294
+ "<SPECIAL_291>",
295
+ "<SPECIAL_292>",
296
+ "<SPECIAL_293>",
297
+ "<SPECIAL_294>",
298
+ "<SPECIAL_295>",
299
+ "<SPECIAL_296>",
300
+ "<SPECIAL_297>",
301
+ "<SPECIAL_298>",
302
+ "<SPECIAL_299>",
303
+ "<SPECIAL_300>",
304
+ "<SPECIAL_301>",
305
+ "<SPECIAL_302>",
306
+ "<SPECIAL_303>",
307
+ "<SPECIAL_304>",
308
+ "<SPECIAL_305>",
309
+ "<SPECIAL_306>",
310
+ "<SPECIAL_307>",
311
+ "<SPECIAL_308>",
312
+ "<SPECIAL_309>",
313
+ "<SPECIAL_310>",
314
+ "<SPECIAL_311>",
315
+ "<SPECIAL_312>",
316
+ "<SPECIAL_313>",
317
+ "<SPECIAL_314>",
318
+ "<SPECIAL_315>",
319
+ "<SPECIAL_316>",
320
+ "<SPECIAL_317>",
321
+ "<SPECIAL_318>",
322
+ "<SPECIAL_319>",
323
+ "<SPECIAL_320>",
324
+ "<SPECIAL_321>",
325
+ "<SPECIAL_322>",
326
+ "<SPECIAL_323>",
327
+ "<SPECIAL_324>",
328
+ "<SPECIAL_325>",
329
+ "<SPECIAL_326>",
330
+ "<SPECIAL_327>",
331
+ "<SPECIAL_328>",
332
+ "<SPECIAL_329>",
333
+ "<SPECIAL_330>",
334
+ "<SPECIAL_331>",
335
+ "<SPECIAL_332>",
336
+ "<SPECIAL_333>",
337
+ "<SPECIAL_334>",
338
+ "<SPECIAL_335>",
339
+ "<SPECIAL_336>",
340
+ "<SPECIAL_337>",
341
+ "<SPECIAL_338>",
342
+ "<SPECIAL_339>",
343
+ "<SPECIAL_340>",
344
+ "<SPECIAL_341>",
345
+ "<SPECIAL_342>",
346
+ "<SPECIAL_343>",
347
+ "<SPECIAL_344>",
348
+ "<SPECIAL_345>",
349
+ "<SPECIAL_346>",
350
+ "<SPECIAL_347>",
351
+ "<SPECIAL_348>",
352
+ "<SPECIAL_349>",
353
+ "<SPECIAL_350>",
354
+ "<SPECIAL_351>",
355
+ "<SPECIAL_352>",
356
+ "<SPECIAL_353>",
357
+ "<SPECIAL_354>",
358
+ "<SPECIAL_355>",
359
+ "<SPECIAL_356>",
360
+ "<SPECIAL_357>",
361
+ "<SPECIAL_358>",
362
+ "<SPECIAL_359>",
363
+ "<SPECIAL_360>",
364
+ "<SPECIAL_361>",
365
+ "<SPECIAL_362>",
366
+ "<SPECIAL_363>",
367
+ "<SPECIAL_364>",
368
+ "<SPECIAL_365>",
369
+ "<SPECIAL_366>",
370
+ "<SPECIAL_367>",
371
+ "<SPECIAL_368>",
372
+ "<SPECIAL_369>",
373
+ "<SPECIAL_370>",
374
+ "<SPECIAL_371>",
375
+ "<SPECIAL_372>",
376
+ "<SPECIAL_373>",
377
+ "<SPECIAL_374>",
378
+ "<SPECIAL_375>",
379
+ "<SPECIAL_376>",
380
+ "<SPECIAL_377>",
381
+ "<SPECIAL_378>",
382
+ "<SPECIAL_379>",
383
+ "<SPECIAL_380>",
384
+ "<SPECIAL_381>",
385
+ "<SPECIAL_382>",
386
+ "<SPECIAL_383>",
387
+ "<SPECIAL_384>",
388
+ "<SPECIAL_385>",
389
+ "<SPECIAL_386>",
390
+ "<SPECIAL_387>",
391
+ "<SPECIAL_388>",
392
+ "<SPECIAL_389>",
393
+ "<SPECIAL_390>",
394
+ "<SPECIAL_391>",
395
+ "<SPECIAL_392>",
396
+ "<SPECIAL_393>",
397
+ "<SPECIAL_394>",
398
+ "<SPECIAL_395>",
399
+ "<SPECIAL_396>",
400
+ "<SPECIAL_397>",
401
+ "<SPECIAL_398>",
402
+ "<SPECIAL_399>",
403
+ "<SPECIAL_400>",
404
+ "<SPECIAL_401>",
405
+ "<SPECIAL_402>",
406
+ "<SPECIAL_403>",
407
+ "<SPECIAL_404>",
408
+ "<SPECIAL_405>",
409
+ "<SPECIAL_406>",
410
+ "<SPECIAL_407>",
411
+ "<SPECIAL_408>",
412
+ "<SPECIAL_409>",
413
+ "<SPECIAL_410>",
414
+ "<SPECIAL_411>",
415
+ "<SPECIAL_412>",
416
+ "<SPECIAL_413>",
417
+ "<SPECIAL_414>",
418
+ "<SPECIAL_415>",
419
+ "<SPECIAL_416>",
420
+ "<SPECIAL_417>",
421
+ "<SPECIAL_418>",
422
+ "<SPECIAL_419>",
423
+ "<SPECIAL_420>",
424
+ "<SPECIAL_421>",
425
+ "<SPECIAL_422>",
426
+ "<SPECIAL_423>",
427
+ "<SPECIAL_424>",
428
+ "<SPECIAL_425>",
429
+ "<SPECIAL_426>",
430
+ "<SPECIAL_427>",
431
+ "<SPECIAL_428>",
432
+ "<SPECIAL_429>",
433
+ "<SPECIAL_430>",
434
+ "<SPECIAL_431>",
435
+ "<SPECIAL_432>",
436
+ "<SPECIAL_433>",
437
+ "<SPECIAL_434>",
438
+ "<SPECIAL_435>",
439
+ "<SPECIAL_436>",
440
+ "<SPECIAL_437>",
441
+ "<SPECIAL_438>",
442
+ "<SPECIAL_439>",
443
+ "<SPECIAL_440>",
444
+ "<SPECIAL_441>",
445
+ "<SPECIAL_442>",
446
+ "<SPECIAL_443>",
447
+ "<SPECIAL_444>",
448
+ "<SPECIAL_445>",
449
+ "<SPECIAL_446>",
450
+ "<SPECIAL_447>",
451
+ "<SPECIAL_448>",
452
+ "<SPECIAL_449>",
453
+ "<SPECIAL_450>",
454
+ "<SPECIAL_451>",
455
+ "<SPECIAL_452>",
456
+ "<SPECIAL_453>",
457
+ "<SPECIAL_454>",
458
+ "<SPECIAL_455>",
459
+ "<SPECIAL_456>",
460
+ "<SPECIAL_457>",
461
+ "<SPECIAL_458>",
462
+ "<SPECIAL_459>",
463
+ "<SPECIAL_460>",
464
+ "<SPECIAL_461>",
465
+ "<SPECIAL_462>",
466
+ "<SPECIAL_463>",
467
+ "<SPECIAL_464>",
468
+ "<SPECIAL_465>",
469
+ "<SPECIAL_466>",
470
+ "<SPECIAL_467>",
471
+ "<SPECIAL_468>",
472
+ "<SPECIAL_469>",
473
+ "<SPECIAL_470>",
474
+ "<SPECIAL_471>",
475
+ "<SPECIAL_472>",
476
+ "<SPECIAL_473>",
477
+ "<SPECIAL_474>",
478
+ "<SPECIAL_475>",
479
+ "<SPECIAL_476>",
480
+ "<SPECIAL_477>",
481
+ "<SPECIAL_478>",
482
+ "<SPECIAL_479>",
483
+ "<SPECIAL_480>",
484
+ "<SPECIAL_481>",
485
+ "<SPECIAL_482>",
486
+ "<SPECIAL_483>",
487
+ "<SPECIAL_484>",
488
+ "<SPECIAL_485>",
489
+ "<SPECIAL_486>",
490
+ "<SPECIAL_487>",
491
+ "<SPECIAL_488>",
492
+ "<SPECIAL_489>",
493
+ "<SPECIAL_490>",
494
+ "<SPECIAL_491>",
495
+ "<SPECIAL_492>",
496
+ "<SPECIAL_493>",
497
+ "<SPECIAL_494>",
498
+ "<SPECIAL_495>",
499
+ "<SPECIAL_496>",
500
+ "<SPECIAL_497>",
501
+ "<SPECIAL_498>",
502
+ "<SPECIAL_499>",
503
+ "<SPECIAL_500>",
504
+ "<SPECIAL_501>",
505
+ "<SPECIAL_502>",
506
+ "<SPECIAL_503>",
507
+ "<SPECIAL_504>",
508
+ "<SPECIAL_505>",
509
+ "<SPECIAL_506>",
510
+ "<SPECIAL_507>",
511
+ "<SPECIAL_508>",
512
+ "<SPECIAL_509>",
513
+ "<SPECIAL_510>",
514
+ "<SPECIAL_511>",
515
+ "<SPECIAL_512>",
516
+ "<SPECIAL_513>",
517
+ "<SPECIAL_514>",
518
+ "<SPECIAL_515>",
519
+ "<SPECIAL_516>",
520
+ "<SPECIAL_517>",
521
+ "<SPECIAL_518>",
522
+ "<SPECIAL_519>",
523
+ "<SPECIAL_520>",
524
+ "<SPECIAL_521>",
525
+ "<SPECIAL_522>",
526
+ "<SPECIAL_523>",
527
+ "<SPECIAL_524>",
528
+ "<SPECIAL_525>",
529
+ "<SPECIAL_526>",
530
+ "<SPECIAL_527>",
531
+ "<SPECIAL_528>",
532
+ "<SPECIAL_529>",
533
+ "<SPECIAL_530>",
534
+ "<SPECIAL_531>",
535
+ "<SPECIAL_532>",
536
+ "<SPECIAL_533>",
537
+ "<SPECIAL_534>",
538
+ "<SPECIAL_535>",
539
+ "<SPECIAL_536>",
540
+ "<SPECIAL_537>",
541
+ "<SPECIAL_538>",
542
+ "<SPECIAL_539>",
543
+ "<SPECIAL_540>",
544
+ "<SPECIAL_541>",
545
+ "<SPECIAL_542>",
546
+ "<SPECIAL_543>",
547
+ "<SPECIAL_544>",
548
+ "<SPECIAL_545>",
549
+ "<SPECIAL_546>",
550
+ "<SPECIAL_547>",
551
+ "<SPECIAL_548>",
552
+ "<SPECIAL_549>",
553
+ "<SPECIAL_550>",
554
+ "<SPECIAL_551>",
555
+ "<SPECIAL_552>",
556
+ "<SPECIAL_553>",
557
+ "<SPECIAL_554>",
558
+ "<SPECIAL_555>",
559
+ "<SPECIAL_556>",
560
+ "<SPECIAL_557>",
561
+ "<SPECIAL_558>",
562
+ "<SPECIAL_559>",
563
+ "<SPECIAL_560>",
564
+ "<SPECIAL_561>",
565
+ "<SPECIAL_562>",
566
+ "<SPECIAL_563>",
567
+ "<SPECIAL_564>",
568
+ "<SPECIAL_565>",
569
+ "<SPECIAL_566>",
570
+ "<SPECIAL_567>",
571
+ "<SPECIAL_568>",
572
+ "<SPECIAL_569>",
573
+ "<SPECIAL_570>",
574
+ "<SPECIAL_571>",
575
+ "<SPECIAL_572>",
576
+ "<SPECIAL_573>",
577
+ "<SPECIAL_574>",
578
+ "<SPECIAL_575>",
579
+ "<SPECIAL_576>",
580
+ "<SPECIAL_577>",
581
+ "<SPECIAL_578>",
582
+ "<SPECIAL_579>",
583
+ "<SPECIAL_580>",
584
+ "<SPECIAL_581>",
585
+ "<SPECIAL_582>",
586
+ "<SPECIAL_583>",
587
+ "<SPECIAL_584>",
588
+ "<SPECIAL_585>",
589
+ "<SPECIAL_586>",
590
+ "<SPECIAL_587>",
591
+ "<SPECIAL_588>",
592
+ "<SPECIAL_589>",
593
+ "<SPECIAL_590>",
594
+ "<SPECIAL_591>",
595
+ "<SPECIAL_592>",
596
+ "<SPECIAL_593>",
597
+ "<SPECIAL_594>",
598
+ "<SPECIAL_595>",
599
+ "<SPECIAL_596>",
600
+ "<SPECIAL_597>",
601
+ "<SPECIAL_598>",
602
+ "<SPECIAL_599>",
603
+ "<SPECIAL_600>",
604
+ "<SPECIAL_601>",
605
+ "<SPECIAL_602>",
606
+ "<SPECIAL_603>",
607
+ "<SPECIAL_604>",
608
+ "<SPECIAL_605>",
609
+ "<SPECIAL_606>",
610
+ "<SPECIAL_607>",
611
+ "<SPECIAL_608>",
612
+ "<SPECIAL_609>",
613
+ "<SPECIAL_610>",
614
+ "<SPECIAL_611>",
615
+ "<SPECIAL_612>",
616
+ "<SPECIAL_613>",
617
+ "<SPECIAL_614>",
618
+ "<SPECIAL_615>",
619
+ "<SPECIAL_616>",
620
+ "<SPECIAL_617>",
621
+ "<SPECIAL_618>",
622
+ "<SPECIAL_619>",
623
+ "<SPECIAL_620>",
624
+ "<SPECIAL_621>",
625
+ "<SPECIAL_622>",
626
+ "<SPECIAL_623>",
627
+ "<SPECIAL_624>",
628
+ "<SPECIAL_625>",
629
+ "<SPECIAL_626>",
630
+ "<SPECIAL_627>",
631
+ "<SPECIAL_628>",
632
+ "<SPECIAL_629>",
633
+ "<SPECIAL_630>",
634
+ "<SPECIAL_631>",
635
+ "<SPECIAL_632>",
636
+ "<SPECIAL_633>",
637
+ "<SPECIAL_634>",
638
+ "<SPECIAL_635>",
639
+ "<SPECIAL_636>",
640
+ "<SPECIAL_637>",
641
+ "<SPECIAL_638>",
642
+ "<SPECIAL_639>",
643
+ "<SPECIAL_640>",
644
+ "<SPECIAL_641>",
645
+ "<SPECIAL_642>",
646
+ "<SPECIAL_643>",
647
+ "<SPECIAL_644>",
648
+ "<SPECIAL_645>",
649
+ "<SPECIAL_646>",
650
+ "<SPECIAL_647>",
651
+ "<SPECIAL_648>",
652
+ "<SPECIAL_649>",
653
+ "<SPECIAL_650>",
654
+ "<SPECIAL_651>",
655
+ "<SPECIAL_652>",
656
+ "<SPECIAL_653>",
657
+ "<SPECIAL_654>",
658
+ "<SPECIAL_655>",
659
+ "<SPECIAL_656>",
660
+ "<SPECIAL_657>",
661
+ "<SPECIAL_658>",
662
+ "<SPECIAL_659>",
663
+ "<SPECIAL_660>",
664
+ "<SPECIAL_661>",
665
+ "<SPECIAL_662>",
666
+ "<SPECIAL_663>",
667
+ "<SPECIAL_664>",
668
+ "<SPECIAL_665>",
669
+ "<SPECIAL_666>",
670
+ "<SPECIAL_667>",
671
+ "<SPECIAL_668>",
672
+ "<SPECIAL_669>",
673
+ "<SPECIAL_670>",
674
+ "<SPECIAL_671>",
675
+ "<SPECIAL_672>",
676
+ "<SPECIAL_673>",
677
+ "<SPECIAL_674>",
678
+ "<SPECIAL_675>",
679
+ "<SPECIAL_676>",
680
+ "<SPECIAL_677>",
681
+ "<SPECIAL_678>",
682
+ "<SPECIAL_679>",
683
+ "<SPECIAL_680>",
684
+ "<SPECIAL_681>",
685
+ "<SPECIAL_682>",
686
+ "<SPECIAL_683>",
687
+ "<SPECIAL_684>",
688
+ "<SPECIAL_685>",
689
+ "<SPECIAL_686>",
690
+ "<SPECIAL_687>",
691
+ "<SPECIAL_688>",
692
+ "<SPECIAL_689>",
693
+ "<SPECIAL_690>",
694
+ "<SPECIAL_691>",
695
+ "<SPECIAL_692>",
696
+ "<SPECIAL_693>",
697
+ "<SPECIAL_694>",
698
+ "<SPECIAL_695>",
699
+ "<SPECIAL_696>",
700
+ "<SPECIAL_697>",
701
+ "<SPECIAL_698>",
702
+ "<SPECIAL_699>",
703
+ "<SPECIAL_700>",
704
+ "<SPECIAL_701>",
705
+ "<SPECIAL_702>",
706
+ "<SPECIAL_703>",
707
+ "<SPECIAL_704>",
708
+ "<SPECIAL_705>",
709
+ "<SPECIAL_706>",
710
+ "<SPECIAL_707>",
711
+ "<SPECIAL_708>",
712
+ "<SPECIAL_709>",
713
+ "<SPECIAL_710>",
714
+ "<SPECIAL_711>",
715
+ "<SPECIAL_712>",
716
+ "<SPECIAL_713>",
717
+ "<SPECIAL_714>",
718
+ "<SPECIAL_715>",
719
+ "<SPECIAL_716>",
720
+ "<SPECIAL_717>",
721
+ "<SPECIAL_718>",
722
+ "<SPECIAL_719>",
723
+ "<SPECIAL_720>",
724
+ "<SPECIAL_721>",
725
+ "<SPECIAL_722>",
726
+ "<SPECIAL_723>",
727
+ "<SPECIAL_724>",
728
+ "<SPECIAL_725>",
729
+ "<SPECIAL_726>",
730
+ "<SPECIAL_727>",
731
+ "<SPECIAL_728>",
732
+ "<SPECIAL_729>",
733
+ "<SPECIAL_730>",
734
+ "<SPECIAL_731>",
735
+ "<SPECIAL_732>",
736
+ "<SPECIAL_733>",
737
+ "<SPECIAL_734>",
738
+ "<SPECIAL_735>",
739
+ "<SPECIAL_736>",
740
+ "<SPECIAL_737>",
741
+ "<SPECIAL_738>",
742
+ "<SPECIAL_739>",
743
+ "<SPECIAL_740>",
744
+ "<SPECIAL_741>",
745
+ "<SPECIAL_742>",
746
+ "<SPECIAL_743>",
747
+ "<SPECIAL_744>",
748
+ "<SPECIAL_745>",
749
+ "<SPECIAL_746>",
750
+ "<SPECIAL_747>",
751
+ "<SPECIAL_748>",
752
+ "<SPECIAL_749>",
753
+ "<SPECIAL_750>",
754
+ "<SPECIAL_751>",
755
+ "<SPECIAL_752>",
756
+ "<SPECIAL_753>",
757
+ "<SPECIAL_754>",
758
+ "<SPECIAL_755>",
759
+ "<SPECIAL_756>",
760
+ "<SPECIAL_757>",
761
+ "<SPECIAL_758>",
762
+ "<SPECIAL_759>",
763
+ "<SPECIAL_760>",
764
+ "<SPECIAL_761>",
765
+ "<SPECIAL_762>",
766
+ "<SPECIAL_763>",
767
+ "<SPECIAL_764>",
768
+ "<SPECIAL_765>",
769
+ "<SPECIAL_766>",
770
+ "<SPECIAL_767>",
771
+ "<SPECIAL_768>",
772
+ "<SPECIAL_769>",
773
+ "<SPECIAL_770>",
774
+ "<SPECIAL_771>",
775
+ "<SPECIAL_772>",
776
+ "<SPECIAL_773>",
777
+ "<SPECIAL_774>",
778
+ "<SPECIAL_775>",
779
+ "<SPECIAL_776>",
780
+ "<SPECIAL_777>",
781
+ "<SPECIAL_778>",
782
+ "<SPECIAL_779>",
783
+ "<SPECIAL_780>",
784
+ "<SPECIAL_781>",
785
+ "<SPECIAL_782>",
786
+ "<SPECIAL_783>",
787
+ "<SPECIAL_784>",
788
+ "<SPECIAL_785>",
789
+ "<SPECIAL_786>",
790
+ "<SPECIAL_787>",
791
+ "<SPECIAL_788>",
792
+ "<SPECIAL_789>",
793
+ "<SPECIAL_790>",
794
+ "<SPECIAL_791>",
795
+ "<SPECIAL_792>",
796
+ "<SPECIAL_793>",
797
+ "<SPECIAL_794>",
798
+ "<SPECIAL_795>",
799
+ "<SPECIAL_796>",
800
+ "<SPECIAL_797>",
801
+ "<SPECIAL_798>",
802
+ "<SPECIAL_799>",
803
+ "<SPECIAL_800>",
804
+ "<SPECIAL_801>",
805
+ "<SPECIAL_802>",
806
+ "<SPECIAL_803>",
807
+ "<SPECIAL_804>",
808
+ "<SPECIAL_805>",
809
+ "<SPECIAL_806>",
810
+ "<SPECIAL_807>",
811
+ "<SPECIAL_808>",
812
+ "<SPECIAL_809>",
813
+ "<SPECIAL_810>",
814
+ "<SPECIAL_811>",
815
+ "<SPECIAL_812>",
816
+ "<SPECIAL_813>",
817
+ "<SPECIAL_814>",
818
+ "<SPECIAL_815>",
819
+ "<SPECIAL_816>",
820
+ "<SPECIAL_817>",
821
+ "<SPECIAL_818>",
822
+ "<SPECIAL_819>",
823
+ "<SPECIAL_820>",
824
+ "<SPECIAL_821>",
825
+ "<SPECIAL_822>",
826
+ "<SPECIAL_823>",
827
+ "<SPECIAL_824>",
828
+ "<SPECIAL_825>",
829
+ "<SPECIAL_826>",
830
+ "<SPECIAL_827>",
831
+ "<SPECIAL_828>",
832
+ "<SPECIAL_829>",
833
+ "<SPECIAL_830>",
834
+ "<SPECIAL_831>",
835
+ "<SPECIAL_832>",
836
+ "<SPECIAL_833>",
837
+ "<SPECIAL_834>",
838
+ "<SPECIAL_835>",
839
+ "<SPECIAL_836>",
840
+ "<SPECIAL_837>",
841
+ "<SPECIAL_838>",
842
+ "<SPECIAL_839>",
843
+ "<SPECIAL_840>",
844
+ "<SPECIAL_841>",
845
+ "<SPECIAL_842>",
846
+ "<SPECIAL_843>",
847
+ "<SPECIAL_844>",
848
+ "<SPECIAL_845>",
849
+ "<SPECIAL_846>",
850
+ "<SPECIAL_847>",
851
+ "<SPECIAL_848>",
852
+ "<SPECIAL_849>",
853
+ "<SPECIAL_850>",
854
+ "<SPECIAL_851>",
855
+ "<SPECIAL_852>",
856
+ "<SPECIAL_853>",
857
+ "<SPECIAL_854>",
858
+ "<SPECIAL_855>",
859
+ "<SPECIAL_856>",
860
+ "<SPECIAL_857>",
861
+ "<SPECIAL_858>",
862
+ "<SPECIAL_859>",
863
+ "<SPECIAL_860>",
864
+ "<SPECIAL_861>",
865
+ "<SPECIAL_862>",
866
+ "<SPECIAL_863>",
867
+ "<SPECIAL_864>",
868
+ "<SPECIAL_865>",
869
+ "<SPECIAL_866>",
870
+ "<SPECIAL_867>",
871
+ "<SPECIAL_868>",
872
+ "<SPECIAL_869>",
873
+ "<SPECIAL_870>",
874
+ "<SPECIAL_871>",
875
+ "<SPECIAL_872>",
876
+ "<SPECIAL_873>",
877
+ "<SPECIAL_874>",
878
+ "<SPECIAL_875>",
879
+ "<SPECIAL_876>",
880
+ "<SPECIAL_877>",
881
+ "<SPECIAL_878>",
882
+ "<SPECIAL_879>",
883
+ "<SPECIAL_880>",
884
+ "<SPECIAL_881>",
885
+ "<SPECIAL_882>",
886
+ "<SPECIAL_883>",
887
+ "<SPECIAL_884>",
888
+ "<SPECIAL_885>",
889
+ "<SPECIAL_886>",
890
+ "<SPECIAL_887>",
891
+ "<SPECIAL_888>",
892
+ "<SPECIAL_889>",
893
+ "<SPECIAL_890>",
894
+ "<SPECIAL_891>",
895
+ "<SPECIAL_892>",
896
+ "<SPECIAL_893>",
897
+ "<SPECIAL_894>",
898
+ "<SPECIAL_895>",
899
+ "<SPECIAL_896>",
900
+ "<SPECIAL_897>",
901
+ "<SPECIAL_898>",
902
+ "<SPECIAL_899>",
903
+ "<SPECIAL_900>",
904
+ "<SPECIAL_901>",
905
+ "<SPECIAL_902>",
906
+ "<SPECIAL_903>",
907
+ "<SPECIAL_904>",
908
+ "<SPECIAL_905>",
909
+ "<SPECIAL_906>",
910
+ "<SPECIAL_907>",
911
+ "<SPECIAL_908>",
912
+ "<SPECIAL_909>",
913
+ "<SPECIAL_910>",
914
+ "<SPECIAL_911>",
915
+ "<SPECIAL_912>",
916
+ "<SPECIAL_913>",
917
+ "<SPECIAL_914>",
918
+ "<SPECIAL_915>",
919
+ "<SPECIAL_916>",
920
+ "<SPECIAL_917>",
921
+ "<SPECIAL_918>",
922
+ "<SPECIAL_919>",
923
+ "<SPECIAL_920>",
924
+ "<SPECIAL_921>",
925
+ "<SPECIAL_922>",
926
+ "<SPECIAL_923>",
927
+ "<SPECIAL_924>",
928
+ "<SPECIAL_925>",
929
+ "<SPECIAL_926>",
930
+ "<SPECIAL_927>",
931
+ "<SPECIAL_928>",
932
+ "<SPECIAL_929>",
933
+ "<SPECIAL_930>",
934
+ "<SPECIAL_931>",
935
+ "<SPECIAL_932>",
936
+ "<SPECIAL_933>",
937
+ "<SPECIAL_934>",
938
+ "<SPECIAL_935>",
939
+ "<SPECIAL_936>",
940
+ "<SPECIAL_937>",
941
+ "<SPECIAL_938>",
942
+ "<SPECIAL_939>",
943
+ "<SPECIAL_940>",
944
+ "<SPECIAL_941>",
945
+ "<SPECIAL_942>",
946
+ "<SPECIAL_943>",
947
+ "<SPECIAL_944>",
948
+ "<SPECIAL_945>",
949
+ "<SPECIAL_946>",
950
+ "<SPECIAL_947>",
951
+ "<SPECIAL_948>",
952
+ "<SPECIAL_949>",
953
+ "<SPECIAL_950>",
954
+ "<SPECIAL_951>",
955
+ "<SPECIAL_952>",
956
+ "<SPECIAL_953>",
957
+ "<SPECIAL_954>",
958
+ "<SPECIAL_955>",
959
+ "<SPECIAL_956>",
960
+ "<SPECIAL_957>",
961
+ "<SPECIAL_958>",
962
+ "<SPECIAL_959>",
963
+ "<SPECIAL_960>",
964
+ "<SPECIAL_961>",
965
+ "<SPECIAL_962>",
966
+ "<SPECIAL_963>",
967
+ "<SPECIAL_964>",
968
+ "<SPECIAL_965>",
969
+ "<SPECIAL_966>",
970
+ "<SPECIAL_967>",
971
+ "<SPECIAL_968>",
972
+ "<SPECIAL_969>",
973
+ "<SPECIAL_970>",
974
+ "<SPECIAL_971>",
975
+ "<SPECIAL_972>",
976
+ "<SPECIAL_973>",
977
+ "<SPECIAL_974>",
978
+ "<SPECIAL_975>",
979
+ "<SPECIAL_976>",
980
+ "<SPECIAL_977>",
981
+ "<SPECIAL_978>",
982
+ "<SPECIAL_979>",
983
+ "<SPECIAL_980>",
984
+ "<SPECIAL_981>",
985
+ "<SPECIAL_982>",
986
+ "<SPECIAL_983>",
987
+ "<SPECIAL_984>",
988
+ "<SPECIAL_985>",
989
+ "<SPECIAL_986>",
990
+ "<SPECIAL_987>",
991
+ "<SPECIAL_988>",
992
+ "<SPECIAL_989>",
993
+ "<SPECIAL_990>",
994
+ "<SPECIAL_991>",
995
+ "<SPECIAL_992>",
996
+ "<SPECIAL_993>",
997
+ "<SPECIAL_994>",
998
+ "<SPECIAL_995>",
999
+ "<SPECIAL_996>",
1000
+ "<SPECIAL_997>",
1001
+ "<SPECIAL_998>",
1002
+ "<SPECIAL_999>"
1003
+ ],
1004
+ "bos_token": {
1005
+ "content": "<s>",
1006
+ "lstrip": false,
1007
+ "normalized": false,
1008
+ "rstrip": false,
1009
+ "single_word": false
1010
+ },
1011
+ "eos_token": {
1012
+ "content": "</s>",
1013
+ "lstrip": false,
1014
+ "normalized": false,
1015
+ "rstrip": false,
1016
+ "single_word": false
1017
+ },
1018
+ "pad_token": {
1019
+ "content": "<pad>",
1020
+ "lstrip": false,
1021
+ "normalized": false,
1022
+ "rstrip": false,
1023
+ "single_word": false
1024
+ },
1025
+ "unk_token": {
1026
+ "content": "<unk>",
1027
+ "lstrip": false,
1028
+ "normalized": false,
1029
+ "rstrip": false,
1030
+ "single_word": false
1031
+ }
1032
+ }
tekken.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:839c48629ff570bd664586800aa3ee17ee628f56efc7fd8e145cc01467a1c188
3
+ size 19399650
tokenizer.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:23ad081f384b2bdb3c97f6e5461dfce52c1174c7328854a55006988f0fef9da7
3
+ size 17078019
tokenizer_config.json ADDED
The diff for this file is too large to render. See raw diff