| 1 | # SPDX-License-Identifier: Apache-2.0 |
| 2 | # SPDX-FileCopyrightText: Copyright contributors to the vLLM project |
| 3 | |
| 4 | import copy |
| 5 | from typing import Any |
| 6 | |
| 7 | import llguidance |
| 8 | import pytest |
| 9 | from mistral_common.exceptions import InvalidMessageStructureException |
| 10 | from mistral_common.guidance.grammar_factory import GrammarFactory |
| 11 | from mistral_common.tokens.tokenizers.base import SpecialTokenPolicy, SpecialTokens |
| 12 | |
| 13 | from vllm.entrypoints.chat_utils import ChatCompletionMessageParam |
| 14 | from vllm.entrypoints.openai.chat_completion.protocol import ChatCompletionRequest |
| 15 | from vllm.exceptions import VLLMValidationError |
| 16 | from vllm.tokenizers.mistral import ( |
| 17 | MistralTokenizer, |
| 18 | _validate_apply_chat_template_args, |
| 19 | validate_request_params, |
| 20 | ) |
| 21 | |
| 22 | |
| 23 | def test_validate_apply_chat_template_args(): |
| 24 | # add_generation_prompt with assistant last message → error |
| 25 | messages: list[ChatCompletionMessageParam] = [ |
| 26 | {"role": "assistant", "content": "Hello"} |
| 27 | ] |
| 28 | with pytest.raises(ValueError): |
| 29 | _validate_apply_chat_template_args(messages, add_generation_prompt=True) |
| 30 | |
| 31 | # add_generation_prompt with user last message → ok |
| 32 | messages = [{"role": "user", "content": "Hello"}] |
| 33 | _validate_apply_chat_template_args(messages, add_generation_prompt=True) |
| 34 | |
| 35 | # both add_generation_prompt and continue_final_message → error |
| 36 | with pytest.raises(ValueError): |
| 37 | _validate_apply_chat_template_args( |
| 38 | messages, |
| 39 | add_generation_prompt=True, |
| 40 | continue_final_message=True, |
| 41 | ) |
| 42 | |
| 43 | # continue_final_message with assistant last message → ok |
| 44 | messages = [{"role": "assistant", "content": "Hello"}] |
| 45 | _validate_apply_chat_template_args(messages, continue_final_message=True) |
| 46 | |
| 47 | # continue_final_message with user last message → error |
| 48 | messages = [{"role": "user", "content": "Hello"}] |
| 49 | with pytest.raises(ValueError): |
| 50 | _validate_apply_chat_template_args(messages, continue_final_message=True) |
| 51 | |
| 52 | |
| 53 | @pytest.fixture(scope="module") |
| 54 | def mistral_tokenizer(request) -> MistralTokenizer: |
| 55 | return MistralTokenizer.from_pretrained(request.param) |
| 56 | |
| 57 | |
| 58 | @pytest.mark.parametrize( |
| 59 | "mistral_tokenizer", |
| 60 | ["mistralai/Mistral-7B-Instruct-v0.3", "mistralai/Magistral-Small-2509"], |
| 61 | indirect=True, |
| 62 | ) |
| 63 | class TestMistralTokenizer: |
| 64 | def test_all_special_tokens(self, mistral_tokenizer: MistralTokenizer): |
| 65 | if mistral_tokenizer.is_tekken: |
| 66 | assert mistral_tokenizer.all_special_tokens == [ |
| 67 | "<unk>", |
| 68 | "<s>", |
| 69 | "</s>", |
| 70 | "[INST]", |
| 71 | "[/INST]", |
| 72 | "[AVAILABLE_TOOLS]", |
| 73 | "[/AVAILABLE_TOOLS]", |
| 74 | "[TOOL_RESULTS]", |
| 75 | "[/TOOL_RESULTS]", |
| 76 | "[TOOL_CALLS]", |
| 77 | "[IMG]", |
| 78 | "<pad>", |
| 79 | "[IMG_BREAK]", |
| 80 | "[IMG_END]", |
| 81 | "[PREFIX]", |
| 82 | "[MIDDLE]", |
| 83 | "[SUFFIX]", |
| 84 | "[SYSTEM_PROMPT]", |
| 85 | "[/SYSTEM_PROMPT]", |
| 86 | "[TOOL_CONTENT]", |
| 87 | ] + [f"<SPECIAL_{i}>" for i in range(20, 32)] + [ |
| 88 | "[ARGS]", |
| 89 | "[CALL_ID]", |
| 90 | "[THINK]", |
| 91 | "[/THINK]", |
| 92 | ] + [f"<SPECIAL_{i}>" for i in range(36, 1000)] |
| 93 | else: |
| 94 | assert mistral_tokenizer.all_special_tokens == [ |
| 95 | "<s>", |
| 96 | "</s>", |
| 97 | "[INST]", |
| 98 | "[/INST]", |
| 99 | "[TOOL_CALLS]", |
| 100 | "[AVAILABLE_TOOLS]", |
| 101 | "[/AVAILABLE_TOOLS]", |
| 102 | "[TOOL_RESULTS]", |
| 103 | "[/TOOL_RESULTS]", |
| 104 | ] + [f"[control_{i}]" for i in range(8, 769)] |
| 105 | |
| 106 | def get_vocab(self, mistral_tokenizer: MistralTokenizer): |
| 107 | assert ( |
| 108 | mistral_tokenizer.get_vocab() |
| 109 | == mistral_tokenizer.transformers_tokenizer.get_vocab() |
| 110 | ) |
| 111 | |
| 112 | def test_get_added_vocab(self, mistral_tokenizer: MistralTokenizer): |
| 113 | assert mistral_tokenizer.get_added_vocab() == {} |
| 114 | |
| 115 | def test_encode(self, mistral_tokenizer: MistralTokenizer): |
| 116 | token_ids = ( |
| 117 | [1, 22177, 4304, 2662] |
| 118 | if mistral_tokenizer.is_tekken |
| 119 | else [1, 23325, 2294, 1686] |
| 120 | ) |
| 121 | |
| 122 | assert mistral_tokenizer.encode("Hello world !") == token_ids |
| 123 | assert mistral_tokenizer.encode("Hello world !", max_length=3) == token_ids[:-1] |
| 124 | assert ( |
| 125 | mistral_tokenizer.encode("Hello world !", truncation=True, max_length=3) |
| 126 | == token_ids[:-1] |
| 127 | ) |
| 128 | assert ( |
| 129 | mistral_tokenizer.encode("Hello world !", truncation=False, max_length=3) |
| 130 | == token_ids |
| 131 | ) |
| 132 | |
| 133 | assert ( |
| 134 | mistral_tokenizer.encode("Hello world !", add_special_tokens=True) |
| 135 | == token_ids |
| 136 | ) |
| 137 | assert ( |
| 138 | mistral_tokenizer.encode( |
| 139 | "Hello world !", add_special_tokens=True, max_length=3 |
| 140 | ) |
| 141 | == token_ids[:-1] |
| 142 | ) |
| 143 | assert ( |
| 144 | mistral_tokenizer.encode( |
| 145 | "Hello world !", add_special_tokens=True, truncation=False, max_length=3 |
| 146 | ) |
| 147 | == token_ids |
| 148 | ) |
| 149 | assert ( |
| 150 | mistral_tokenizer.encode("Hello world !", add_special_tokens=False) |
| 151 | == token_ids[1:] |
| 152 | ) |
| 153 | assert mistral_tokenizer.encode("", add_special_tokens=False) == [] |
| 154 | |
| 155 | def test_call(self, mistral_tokenizer: MistralTokenizer): |
| 156 | token_ids = ( |
| 157 | [1, 22177, 4304, 2662] |
| 158 | if mistral_tokenizer.is_tekken |
| 159 | else [1, 23325, 2294, 1686] |
| 160 | ) |
| 161 | attn_mask = [1 for _ in range(len(token_ids))] |
| 162 | |
| 163 | # Test 1: no special tokens |
| 164 | assert mistral_tokenizer("Hello world !", add_special_tokens=False) == { |
| 165 | "attention_mask": attn_mask[1:], |
| 166 | "input_ids": token_ids[1:], |
| 167 | } |
| 168 | # Test 2: special tokens |
| 169 | assert mistral_tokenizer("Hello world !", add_special_tokens=True) == { |
| 170 | "attention_mask": attn_mask, |
| 171 | "input_ids": token_ids, |
| 172 | } |
| 173 | # Test 3: special tokens + truncation |
| 174 | assert mistral_tokenizer( |
| 175 | "Hello world !", add_special_tokens=True, truncation=True, max_length=3 |
| 176 | ) == { |
| 177 | "attention_mask": attn_mask[:-1], |
| 178 | "input_ids": token_ids[:-1], |
| 179 | } |
| 180 | # Test 4: special tokens + no truncation + max length |
| 181 | assert mistral_tokenizer( |
| 182 | "Hello world !", add_special_tokens=True, max_length=3 |
| 183 | ) == { |
| 184 | "attention_mask": attn_mask, |
| 185 | "input_ids": token_ids, |
| 186 | } |
| 187 | # Test 5: empty string |
| 188 | assert mistral_tokenizer("", add_special_tokens=False) == { |
| 189 | "attention_mask": [], |
| 190 | "input_ids": [], |
| 191 | } |
| 192 | |
| 193 | with pytest.raises( |
| 194 | ValueError, |
| 195 | match=(r"`text_pair` is not supported by `MistralTokenizer.__call__`."), |
| 196 | ): |
| 197 | mistral_tokenizer("Hello world !", "invalid pair") |
| 198 | |
| 199 | @pytest.mark.parametrize( |
| 200 | "openai_request,add_generation_prompt,continue_final_message,expected_output,decoded_expected_output", |
| 201 | [ |
| 202 | ( |
| 203 | { |
| 204 | "messages": [ |
| 205 | { |
| 206 | "role": "user", |
| 207 | "content": "Hello world !", |
| 208 | } |
| 209 | ], |
| 210 | }, |
| 211 | True, |
| 212 | False, |
| 213 | ([1, 3, 23325, 2294, 1686, 4], [1, 3, 22177, 4304, 2662, 4]), |
| 214 | ("<s>[INST]▁Hello▁world▁![/INST]", ("<s>[INST]Hello world ![/INST]")), |
| 215 | ), |
| 216 | ( |
| 217 | { |
| 218 | "messages": [ |
| 219 | { |
| 220 | "role": "system", |
| 221 | "content": "I am an AI", |
| 222 | }, |
| 223 | { |
| 224 | "role": "user", |
| 225 | "content": "Hello world !", |
| 226 | }, |
| 227 | ], |
| 228 | }, |
| 229 | True, |
| 230 | False, |
| 231 | ( |
| 232 | [1, 3, 1083, 1605, 1164, 16875, 781, 781, 16998, 2294, 1686, 4], |
| 233 | [1, 17, 1073, 1855, 1420, 26554, 18, 3, 22177, 4304, 2662, 4], |
| 234 | ), |
| 235 | ( |
| 236 | "<s>[INST]▁I▁am▁an▁AI<0x0A><0x0A>Hello▁world▁![/INST]", |
| 237 | ( |
| 238 | "<s>[SYSTEM_PROMPT]I am an AI[/SYSTEM_PROMPT][INST]Hello world ![/INST]" # noqa: E501 |
| 239 | ), |
| 240 | ), |
| 241 | ), |
| 242 | ( |
| 243 | { |
| 244 | "messages": [ |
| 245 | { |
| 246 | "role": "system", |
| 247 | "content": "I am an AI", |
| 248 | }, |
| 249 | { |
| 250 | "role": "user", |
| 251 | "content": "Hello world !", |
| 252 | }, |
| 253 | ], |
| 254 | "tools": [ |
| 255 | { |
| 256 | "type": "function", |
| 257 | "function": { |
| 258 | "name": "get_weather", |
| 259 | "description": "Gets the current weather in a city.", |
| 260 | "parameters": { |
| 261 | "type": "object", |
| 262 | "properties": { |
| 263 | "city": { |
| 264 | "type": "string", |
| 265 | "description": "The city name", |
| 266 | } |
| 267 | }, |
| 268 | "required": ["city"], |
| 269 | }, |
| 270 | }, |
| 271 | } |
| 272 | ], |
| 273 | }, |
| 274 | True, |
| 275 | False, |
| 276 | ( |
| 277 | [ |
| 278 | 1, |
| 279 | 6, |
| 280 | 1501, |
| 281 | 7567, |
| 282 | 1891, |
| 283 | 2032, |
| 284 | 1113, |
| 285 | 3396, |
| 286 | 1316, |
| 287 | 1113, |
| 288 | 3396, |
| 289 | 2032, |
| 290 | 10598, |
| 291 | 1629, |
| 292 | 2032, |
| 293 | 1113, |
| 294 | 1295, |
| 295 | 29498, |
| 296 | 1537, |
| 297 | 1991, |
| 298 | 1316, |
| 299 | 1113, |
| 300 | 7286, |
| 301 | 2032, |
| 302 | 1113, |
| 303 | 2226, |
| 304 | 29481, |
| 305 | 1040, |
| 306 | 2636, |
| 307 | 8854, |
| 308 | 1065, |
| 309 | 1032, |
| 310 | 3758, |
| 311 | 9959, |
| 312 | 1113, |
| 313 | 12206, |
| 314 | 2032, |
| 315 | 10598, |
| 316 | 1891, |
| 317 | 2032, |
| 318 | 1113, |
| 319 | 3582, |
| 320 | 1316, |
| 321 | 1113, |
| 322 | 11491, |
| 323 | 2032, |
| 324 | 10598, |
| 325 | 19141, |
| 326 | 2032, |
| 327 | 10598, |
| 328 | 1891, |
| 329 | 2032, |
| 330 | 1113, |
| 331 | 2195, |
| 332 | 1316, |
| 333 | 1113, |
| 334 | 7286, |
| 335 | 2032, |
| 336 | 1113, |
| 337 | 1782, |
| 338 | 3758, |
| 339 | 1909, |
| 340 | 29507, |
| 341 | 11549, |
| 342 | 1113, |
| 343 | 11661, |
| 344 | 2032, |
| 345 | 8135, |
| 346 | 19141, |
| 347 | 3010, |
| 348 | 1743, |
| 349 | 10925, |
| 350 | 7, |
| 351 | 3, |
| 352 | 1083, |
| 353 | 1605, |
| 354 | 1164, |
| 355 | 16875, |
| 356 | 781, |
| 357 | 781, |
| 358 | 16998, |
| 359 | 2294, |
| 360 | 1686, |
| 361 | 4, |
| 362 | ], |
| 363 | [ |
| 364 | 1, |
| 365 | 17, |
| 366 | 1073, |
| 367 | 1855, |
| 368 | 1420, |
| 369 | 26554, |
| 370 | 18, |
| 371 | 5, |
| 372 | 1091, |
| 373 | 19227, |
| 374 | 4994, |
| 375 | 2811, |
| 376 | 1429, |
| 377 | 5165, |
| 378 | 1897, |
| 379 | 1429, |
| 380 | 5165, |
| 381 | 2811, |
| 382 | 16753, |
| 383 | 2391, |
| 384 | 2811, |
| 385 | 1429, |
| 386 | 1689, |
| 387 | 1095, |
| 388 | 45629, |
| 389 | 1897, |
| 390 | 1429, |
| 391 | 14653, |
| 392 | 2811, |
| 393 | 1429, |
| 394 | 1071, |
| 395 | 3083, |
| 396 | 1278, |
| 397 | 3519, |
| 398 | 17253, |
| 399 | 1294, |
| 400 | 1261, |
| 401 | 5970, |
| 402 | 39249, |
| 403 | 1429, |
| 404 | 26204, |
| 405 | 2811, |
| 406 | 16753, |
| 407 | 4994, |
| 408 | 2811, |
| 409 | 1429, |
| 410 | 6371, |
| 411 | 1897, |
| 412 | 1429, |
| 413 | 48649, |
| 414 | 2811, |
| 415 | 16753, |
| 416 | 29363, |
| 417 | 2811, |
| 418 | 16753, |
| 419 | 4994, |
| 420 | 2811, |
| 421 | 1429, |
| 422 | 3607, |
| 423 | 1897, |
| 424 | 1429, |
| 425 | 14653, |
| 426 | 2811, |
| 427 | 1429, |
| 428 | 1784, |
| 429 | 5970, |
| 430 | 2564, |
| 431 | 1034, |
| 432 | 47579, |
| 433 | 1429, |
| 434 | 15760, |
| 435 | 2811, |
| 436 | 12161, |
| 437 | 29363, |
| 438 | 4964, |
| 439 | 2821, |
| 440 | 27028, |
| 441 | 6, |
| 442 | 3, |
| 443 | 22177, |
| 444 | 4304, |
| 445 | 2662, |
| 446 | 4, |
| 447 | ], |
| 448 | ), |
| 449 | ( |
| 450 | '<s>[AVAILABLE_TOOLS]▁[{"type":▁"function",▁"function":▁{"name":▁"get_weather",▁"description":▁"Gets▁the▁current▁weather▁in▁a▁city.",▁"parameters":▁{"type":▁"object",▁"properties":▁{"city":▁{"type":▁"string",▁"description":▁"The▁city▁name"}},▁"required":▁["city"]}}}][/AVAILABLE_TOOLS][INST]▁I▁am▁an▁AI<0x0A><0x0A>Hello▁world▁![/INST]', |
| 451 | ( |
| 452 | '<s>[SYSTEM_PROMPT]I am an AI[/SYSTEM_PROMPT][AVAILABLE_TOOLS][{"type": "function", "function": {"name": "get_weather", "description": "Gets the current weather in a city.", "parameters": {"type": "object", "properties": {"city": {"type": "string", "description": "The city name"}}, "required": ["city"]}}}][/AVAILABLE_TOOLS][INST]Hello world ![/INST]' # noqa: E501 |
| 453 | ), |
| 454 | ), |
| 455 | ), |
| 456 | ( |
| 457 | { |
| 458 | "messages": [ |
| 459 | { |
| 460 | "role": "system", |
| 461 | "content": "I am an AI", |
| 462 | }, |
| 463 | { |
| 464 | "role": "user", |
| 465 | "content": "Hello world !", |
| 466 | }, |
| 467 | { |
| 468 | "role": "assistant", |
| 469 | "content": "", |
| 470 | "tool_calls": [ |
| 471 | { |
| 472 | "id": "123456789", |
| 473 | "type": "function", |
| 474 | "function": { |
| 475 | "name": "get_weather", |
| 476 | "arguments": '{"city": "Paris"}', |
| 477 | }, |
| 478 | } |
| 479 | ], |
| 480 | }, |
| 481 | { |
| 482 | "role": "tool", |
| 483 | "tool_call_id": "123456789", |
| 484 | "content": '{"temperature": 20, "unit": "celsius"}', |
| 485 | }, |
| 486 | ], |
| 487 | "tools": [ |
| 488 | { |
| 489 | "type": "function", |
| 490 | "function": { |
| 491 | "name": "get_weather", |
| 492 | "description": "Gets the current weather in a city.", |
| 493 | "parameters": { |
| 494 | "type": "object", |
| 495 | "properties": { |
| 496 | "city": { |
| 497 | "type": "string", |
| 498 | "description": "The city name", |
| 499 | } |
| 500 | }, |
| 501 | "required": ["city"], |
| 502 | }, |
| 503 | }, |
| 504 | } |
| 505 | ], |
| 506 | }, |
| 507 | True, |
| 508 | False, |
| 509 | ( |
| 510 | [ |
| 511 | 1, |
| 512 | 6, |
| 513 | 1501, |
| 514 | 7567, |
| 515 | 1891, |
| 516 | 2032, |
| 517 | 1113, |
| 518 | 3396, |
| 519 | 1316, |
| 520 | 1113, |
| 521 | 3396, |
| 522 | 2032, |
| 523 | 10598, |
| 524 | 1629, |
| 525 | 2032, |
| 526 | 1113, |
| 527 | 1295, |
| 528 | 29498, |
| 529 | 1537, |
| 530 | 1991, |
| 531 | 1316, |
| 532 | 1113, |
| 533 | 7286, |
| 534 | 2032, |
| 535 | 1113, |
| 536 | 2226, |
| 537 | 29481, |
| 538 | 1040, |
| 539 | 2636, |
| 540 | 8854, |
| 541 | 1065, |
| 542 | 1032, |
| 543 | 3758, |
| 544 | 9959, |
| 545 | 1113, |
| 546 | 12206, |
| 547 | 2032, |
| 548 | 10598, |
| 549 | 1891, |
| 550 | 2032, |
| 551 | 1113, |
| 552 | 3582, |
| 553 | 1316, |
| 554 | 1113, |
| 555 | 11491, |
| 556 | 2032, |
| 557 | 10598, |
| 558 | 19141, |
| 559 | 2032, |
| 560 | 10598, |
| 561 | 1891, |
| 562 | 2032, |
| 563 | 1113, |
| 564 | 2195, |
| 565 | 1316, |
| 566 | 1113, |
| 567 | 7286, |
| 568 | 2032, |
| 569 | 1113, |
| 570 | 1782, |
| 571 | 3758, |
| 572 | 1909, |
| 573 | 29507, |
| 574 | 11549, |
| 575 | 1113, |
| 576 | 11661, |
| 577 | 2032, |
| 578 | 8135, |
| 579 | 19141, |
| 580 | 3010, |
| 581 | 1743, |
| 582 | 10925, |
| 583 | 7, |
| 584 | 3, |
| 585 | 1083, |
| 586 | 1605, |
| 587 | 1164, |
| 588 | 16875, |
| 589 | 781, |
| 590 | 781, |
| 591 | 16998, |
| 592 | 2294, |
| 593 | 1686, |
| 594 | 4, |
| 595 | 5, |
| 596 | 1501, |
| 597 | 7567, |
| 598 | 1629, |
| 599 | 2032, |
| 600 | 1113, |
| 601 | 1295, |
| 602 | 29498, |
| 603 | 1537, |
| 604 | 1991, |
| 605 | 1316, |
| 606 | 1113, |
| 607 | 17452, |
| 608 | 2032, |
| 609 | 10598, |
| 610 | 19141, |
| 611 | 2032, |
| 612 | 1113, |
| 613 | 4684, |
| 614 | 1046, |
| 615 | 8474, |
| 616 | 1113, |
| 617 | 1081, |
| 618 | 2032, |
| 619 | 1113, |
| 620 | 29508, |
| 621 | 29518, |
| 622 | 29538, |
| 623 | 29549, |
| 624 | 29550, |
| 625 | 29552, |
| 626 | 29555, |
| 627 | 29551, |
| 628 | 29542, |
| 629 | 29507, |
| 630 | 10925, |
| 631 | 2, |
| 632 | 8, |
| 633 | 10598, |
| 634 | 4557, |
| 635 | 2032, |
| 636 | 10598, |
| 637 | 29475, |
| 638 | 17329, |
| 639 | 2032, |
| 640 | 29473, |
| 641 | 29518, |
| 642 | 29502, |
| 643 | 29493, |
| 644 | 1113, |
| 645 | 6074, |
| 646 | 2032, |
| 647 | 1113, |
| 648 | 29485, |
| 649 | 1958, |
| 650 | 3938, |
| 651 | 8474, |
| 652 | 1113, |
| 653 | 3613, |
| 654 | 29498, |
| 655 | 1081, |
| 656 | 2032, |
| 657 | 1113, |
| 658 | 29508, |
| 659 | 29518, |
| 660 | 29538, |
| 661 | 29549, |
| 662 | 29550, |
| 663 | 29552, |
| 664 | 29555, |
| 665 | 29551, |
| 666 | 29542, |
| 667 | 18163, |
| 668 | 9, |
| 669 | ], |
| 670 | [ |
| 671 | 1, |
| 672 | 17, |
| 673 | 1073, |
| 674 | 1855, |
| 675 | 1420, |
| 676 | 26554, |
| 677 | 18, |
| 678 | 5, |
| 679 | 1091, |
| 680 | 19227, |
| 681 | 4994, |
| 682 | 2811, |
| 683 | 1429, |
| 684 | 5165, |
| 685 | 1897, |
| 686 | 1429, |
| 687 | 5165, |
| 688 | 2811, |
| 689 | 16753, |
| 690 | 2391, |
| 691 | 2811, |
| 692 | 1429, |
| 693 | 1689, |
| 694 | 1095, |
| 695 | 45629, |
| 696 | 1897, |
| 697 | 1429, |
| 698 | 14653, |
| 699 | 2811, |
| 700 | 1429, |
| 701 | 1071, |
| 702 | 3083, |
| 703 | 1278, |
| 704 | 3519, |
| 705 | 17253, |
| 706 | 1294, |
| 707 | 1261, |
| 708 | 5970, |
| 709 | 39249, |
| 710 | 1429, |
| 711 | 26204, |
| 712 | 2811, |
| 713 | 16753, |
| 714 | 4994, |
| 715 | 2811, |
| 716 | 1429, |
| 717 | 6371, |
| 718 | 1897, |
| 719 | 1429, |
| 720 | 48649, |
| 721 | 2811, |
| 722 | 16753, |
| 723 | 29363, |
| 724 | 2811, |
| 725 | 16753, |
| 726 | 4994, |
| 727 | 2811, |
| 728 | 1429, |
| 729 | 3607, |
| 730 | 1897, |
| 731 | 1429, |
| 732 | 14653, |
| 733 | 2811, |
| 734 | 1429, |
| 735 | 1784, |
| 736 | 5970, |
| 737 | 2564, |
| 738 | 1034, |
| 739 | 47579, |
| 740 | 1429, |
| 741 | 15760, |
| 742 | 2811, |
| 743 | 12161, |
| 744 | 29363, |
| 745 | 4964, |
| 746 | 2821, |
| 747 | 27028, |
| 748 | 6, |
| 749 | 3, |
| 750 | 22177, |
| 751 | 4304, |
| 752 | 2662, |
| 753 | 4, |
| 754 | 9, |
| 755 | 1689, |
| 756 | 1095, |
| 757 | 45629, |
| 758 | 32, |
| 759 | 19227, |
| 760 | 29363, |
| 761 | 2811, |
| 762 | 1429, |
| 763 | 42572, |
| 764 | 46005, |
| 765 | 2, |
| 766 | 7, |
| 767 | 19227, |
| 768 | 113824, |
| 769 | 2811, |
| 770 | 1032, |
| 771 | 1050, |
| 772 | 1048, |
| 773 | 1044, |
| 774 | 1429, |
| 775 | 8979, |
| 776 | 2811, |
| 777 | 1429, |
| 778 | 1099, |
| 779 | 79092, |
| 780 | 46005, |
| 781 | 8, |
| 782 | ], |
| 783 | ), |
| 784 | ( |
| 785 | '<s>[AVAILABLE_TOOLS]▁[{"type":▁"function",▁"function":▁{"name":▁"get_weather",▁"description":▁"Gets▁the▁current▁weather▁in▁a▁city.",▁"parameters":▁{"type":▁"object",▁"properties":▁{"city":▁{"type":▁"string",▁"description":▁"The▁city▁name"}},▁"required":▁["city"]}}}][/AVAILABLE_TOOLS][INST]▁I▁am▁an▁AI<0x0A><0x0A>Hello▁world▁![/INST][TOOL_CALLS]▁[{"name":▁"get_weather",▁"arguments":▁{"city":▁"Paris"},▁"id":▁"123456789"}]</s>[TOOL_RESULTS]▁{"content":▁{"temperature":▁20,▁"unit":▁"celsius"},▁"call_id":▁"123456789"}[/TOOL_RESULTS]', |
| 786 | ( |
| 787 | '<s>[SYSTEM_PROMPT]I am an AI[/SYSTEM_PROMPT][AVAILABLE_TOOLS][{"type": "function", "function": {"name": "get_weather", "description": "Gets the current weather in a city.", "parameters": {"type": "object", "properties": {"city": {"type": "string", "description": "The city name"}}, "required": ["city"]}}}][/AVAILABLE_TOOLS][INST]Hello world ![/INST][TOOL_CALLS]get_weather[ARGS]{"city": "Paris"}</s>[TOOL_RESULTS]{"temperature": 20, "unit": "celsius"}[/TOOL_RESULTS]' # noqa: E501 |
| 788 | ), |
| 789 | ), |
| 790 | ), |
| 791 | ( |
| 792 | { |
| 793 | "messages": [ |
| 794 | { |
| 795 | "role": "user", |
| 796 | "content": "Hello world !", |
| 797 | }, |
| 798 | { |
| 799 | "role": "assistant", |
| 800 | "content": "Hello ", |
| 801 | }, |
| 802 | ], |
| 803 | }, |
| 804 | False, |
| 805 | True, |
| 806 | ( |
| 807 | [1, 3, 23325, 2294, 1686, 4, 23325], |
| 808 | [1, 3, 22177, 4304, 2662, 4, 22177], |
| 809 | ), |
| 810 | ( |
| 811 | "<s>[INST]▁Hello▁world▁![/INST]▁Hello", |
| 812 | "<s>[INST]Hello world ![/INST]Hello", |
| 813 | ), |
| 814 | ), |
| 815 | ], |
| 816 | ) |
| 817 | def test_apply_chat_template( |
| 818 | self, |
| 819 | mistral_tokenizer: MistralTokenizer, |
| 820 | openai_request: dict[str, Any], |
| 821 | add_generation_prompt: bool, |
| 822 | continue_final_message: bool, |
| 823 | expected_output: tuple[list[int], list[int]], |
| 824 | decoded_expected_output: tuple[str, str], |
| 825 | ): |
| 826 | actual_output = mistral_tokenizer.apply_chat_template( |
| 827 | openai_request["messages"], |
| 828 | tools=openai_request.get("tools", []), |
| 829 | add_generation_prompt=add_generation_prompt, |
| 830 | continue_final_message=continue_final_message, |
| 831 | ) |
| 832 | decoded_actual_output = mistral_tokenizer.tokenizer.decode( |
| 833 | actual_output, SpecialTokenPolicy.KEEP |
| 834 | ) |
| 835 | |
| 836 | assert actual_output == expected_output[mistral_tokenizer.is_tekken] |
| 837 | assert ( |
| 838 | decoded_actual_output |
| 839 | == decoded_expected_output[mistral_tokenizer.is_tekken] |
| 840 | ) |
| 841 | |
| 842 | def test_apply_chat_template_error(self, mistral_tokenizer: MistralTokenizer): |
| 843 | messages: list[ChatCompletionMessageParam] = [ |
| 844 | {"role": "user", "content": "Hello world !"} |
| 845 | ] |
| 846 | |
| 847 | with pytest.raises(ValueError): |
| 848 | mistral_tokenizer.apply_chat_template( |
| 849 | messages, |
| 850 | tools=[], |
| 851 | add_generation_prompt=True, |
| 852 | continue_final_message=True, |
| 853 | ) |
| 854 | |
| 855 | with pytest.raises(ValueError): |
| 856 | mistral_tokenizer.apply_chat_template( |
| 857 | messages, |
| 858 | tools=[], |
| 859 | add_generation_prompt=False, |
| 860 | continue_final_message=True, |
| 861 | ) |
| 862 | |
| 863 | messages = [ |
| 864 | {"role": "user", "content": "Hello world !"}, |
| 865 | {"role": "assistant", "content": "Hello "}, |
| 866 | ] |
| 867 | with pytest.raises(ValueError): |
| 868 | mistral_tokenizer.apply_chat_template( |
| 869 | messages, |
| 870 | tools=[], |
| 871 | add_generation_prompt=True, |
| 872 | continue_final_message=False, |
| 873 | ) |
| 874 | |
| 875 | messages = [ |
| 876 | {"role": "user", "content": "Hello world !"}, |
| 877 | {"role": "assistant", "content": "Hello "}, |
| 878 | ] |
| 879 | with pytest.raises(InvalidMessageStructureException): |
| 880 | mistral_tokenizer.apply_chat_template( |
| 881 | messages, |
| 882 | tools=[], |
| 883 | add_generation_prompt=False, |
| 884 | continue_final_message=False, |
| 885 | ) |
| 886 | |
| 887 | @pytest.mark.parametrize( |
| 888 | "skip_special_tokens,expected_tokens", |
| 889 | ( |
| 890 | ( |
| 891 | False, |
| 892 | ( |
| 893 | "<s>[INST]▁Hello▁world▁![/INST]▁Hello</s>", |
| 894 | "<s>[INST]Hello world ![/INST]Hello</s>", |
| 895 | ), |
| 896 | ), |
| 897 | (True, ("Hello world ! Hello", "Hello world !Hello")), |
| 898 | ), |
| 899 | ) |
| 900 | def test_decode( |
| 901 | self, |
| 902 | mistral_tokenizer: MistralTokenizer, |
| 903 | skip_special_tokens: bool, |
| 904 | expected_tokens: tuple[str, str], |
| 905 | ): |
| 906 | ids = ( |
| 907 | [1, 3, 23325, 2294, 1686, 4, 23325, 2], |
| 908 | [1, 3, 22177, 4304, 2662, 4, 22177, 2], |
| 909 | ) |
| 910 | assert ( |
| 911 | mistral_tokenizer.decode( |
| 912 | ids[mistral_tokenizer.is_tekken], |
| 913 | skip_special_tokens=skip_special_tokens, |
| 914 | ) |
| 915 | == expected_tokens[mistral_tokenizer.is_tekken] |
| 916 | ) |
| 917 | |
| 918 | def test_decode_empty( |
| 919 | self, |
| 920 | mistral_tokenizer: MistralTokenizer, |
| 921 | ): |
| 922 | assert ( |
| 923 | mistral_tokenizer.decode( |
| 924 | [], |
| 925 | ) |
| 926 | == "" |
| 927 | ) |
| 928 | |
| 929 | def test_decode_int( |
| 930 | self, |
| 931 | mistral_tokenizer: MistralTokenizer, |
| 932 | ): |
| 933 | ids = 1 |
| 934 | assert ( |
| 935 | mistral_tokenizer.decode( |
| 936 | ids, |
| 937 | skip_special_tokens=False, |
| 938 | ) |
| 939 | == "<s>" |
| 940 | ) |
| 941 | |
| 942 | @pytest.mark.parametrize( |
| 943 | "skip_special_tokens,expected_tokens", |
| 944 | ( |
| 945 | ( |
| 946 | False, |
| 947 | ( |
| 948 | ["<s>[INST]▁Hello▁world▁![/INST]▁Hello</s>"], |
| 949 | ["<s>[INST]Hello world ![/INST]Hello</s>"], |
| 950 | ), |
| 951 | ), |
| 952 | (True, (["Hello world ! Hello"], ["Hello world !Hello"])), |
| 953 | ), |
| 954 | ) |
| 955 | def test_batch_decode( |
| 956 | self, |
| 957 | mistral_tokenizer: MistralTokenizer, |
| 958 | skip_special_tokens: bool, |
| 959 | expected_tokens: tuple[str, str], |
| 960 | ): |
| 961 | ids = ( |
| 962 | [[1, 3, 23325, 2294, 1686, 4, 23325, 2]], |
| 963 | [[1, 3, 22177, 4304, 2662, 4, 22177, 2]], |
| 964 | ) |
| 965 | assert ( |
| 966 | mistral_tokenizer.batch_decode( |
| 967 | ids[mistral_tokenizer.is_tekken], |
| 968 | skip_special_tokens=skip_special_tokens, |
| 969 | ) |
| 970 | == expected_tokens[mistral_tokenizer.is_tekken] |
| 971 | ) |
| 972 | |
| 973 | def test_batch_decode_empty( |
| 974 | self, |
| 975 | mistral_tokenizer: MistralTokenizer, |
| 976 | ): |
| 977 | assert mistral_tokenizer.batch_decode( |
| 978 | [[]], |
| 979 | ) == [""] |
| 980 | |
| 981 | def test_convert_tokens_to_string(self, mistral_tokenizer: MistralTokenizer): |
| 982 | tokens = ( |
| 983 | [ |
| 984 | "<s>", |
| 985 | "[AVAILABLE_TOOLS]", |
| 986 | "▁[", |
| 987 | '{"', |
| 988 | "type", |
| 989 | '":', |
| 990 | '▁"', |
| 991 | "function", |
| 992 | '",', |
| 993 | '▁"', |
| 994 | "function", |
| 995 | '":', |
| 996 | '▁{"', |
| 997 | "name", |
| 998 | '":', |
| 999 | '▁"', |
| 1000 | "get", |
| 1001 | "_", |
| 1002 | "we", |
| 1003 | "ather", |
| 1004 | '",', |
| 1005 | '▁"', |
| 1006 | "description", |
| 1007 | '":', |
| 1008 | '▁"', |
| 1009 | "Get", |
| 1010 | "s", |
| 1011 | "▁the", |
| 1012 | "▁current", |
| 1013 | "▁weather", |
| 1014 | "▁in", |
| 1015 | "▁a", |
| 1016 | "▁city", |
| 1017 | '.",', |
| 1018 | '▁"', |
| 1019 | "parameters", |
| 1020 | '":', |
| 1021 | '▁{"', |
| 1022 | "type", |
| 1023 | '":', |
| 1024 | '▁"', |
| 1025 | "object", |
| 1026 | '",', |
| 1027 | '▁"', |
| 1028 | "properties", |
| 1029 | '":', |
| 1030 | '▁{"', |
| 1031 | "city", |
| 1032 | '":', |
| 1033 | '▁{"', |
| 1034 | "type", |
| 1035 | '":', |
| 1036 | '▁"', |
| 1037 | "string", |
| 1038 | '",', |
| 1039 | '▁"', |
| 1040 | "description", |
| 1041 | '":', |
| 1042 | '▁"', |
| 1043 | "The", |
| 1044 | "▁city", |
| 1045 | "▁name", |
| 1046 | '"', |
| 1047 | "}},", |
| 1048 | '▁"', |
| 1049 | "required", |
| 1050 | '":', |
| 1051 | '▁["', |
| 1052 | "city", |
| 1053 | '"]', |
| 1054 | "}}", |
| 1055 | "}]", |
| 1056 | "[/AVAILABLE_TOOLS]", |
| 1057 | "[INST]", |
| 1058 | "▁I", |
| 1059 | "▁am", |
| 1060 | "▁an", |
| 1061 | "▁AI", |
| 1062 | "<0x0A>", |
| 1063 | "<0x0A>", |
| 1064 | "Hello", |
| 1065 | "▁world", |
| 1066 | "▁!", |
| 1067 | "[/INST]", |
| 1068 | "[TOOL_CALLS]", |
| 1069 | "▁[", |
| 1070 | '{"', |
| 1071 | "name", |
| 1072 | '":', |
| 1073 | '▁"', |
| 1074 | "get", |
| 1075 | "_", |
| 1076 | "we", |
| 1077 | "ather", |
| 1078 | '",', |
| 1079 | '▁"', |
| 1080 | "arguments", |
| 1081 | '":', |
| 1082 | '▁{"', |
| 1083 | "city", |
| 1084 | '":', |
| 1085 | '▁"', |
| 1086 | "Par", |
| 1087 | "is", |
| 1088 | '"},', |
| 1089 | '▁"', |
| 1090 | "id", |
| 1091 | '":', |
| 1092 | '▁"', |
| 1093 | "1", |
| 1094 | "2", |
| 1095 | "3", |
| 1096 | "4", |
| 1097 | "5", |
| 1098 | "6", |
| 1099 | "7", |
| 1100 | "8", |
| 1101 | "9", |
| 1102 | '"', |
| 1103 | "}]", |
| 1104 | "</s>", |
| 1105 | "[TOOL_RESULTS]", |
| 1106 | '▁{"', |
| 1107 | "content", |
| 1108 | '":', |
| 1109 | '▁{"', |
| 1110 | "t", |
| 1111 | "emperature", |
| 1112 | '":', |
| 1113 | "▁", |
| 1114 | "2", |
| 1115 | "0", |
| 1116 | ",", |
| 1117 | '▁"', |
| 1118 | "unit", |
| 1119 | '":', |
| 1120 | '▁"', |
| 1121 | "c", |
| 1122 | "els", |
| 1123 | "ius", |
| 1124 | '"},', |
| 1125 | '▁"', |
| 1126 | "call", |
| 1127 | "_", |
| 1128 | "id", |
| 1129 | '":', |
| 1130 | '▁"', |
| 1131 | "1", |
| 1132 | "2", |
| 1133 | "3", |
| 1134 | "4", |
| 1135 | "5", |
| 1136 | "6", |
| 1137 | "7", |
| 1138 | "8", |
| 1139 | "9", |
| 1140 | '"}', |
| 1141 | "[/TOOL_RESULTS]", |
| 1142 | ], |
| 1143 | [ |
| 1144 | "<s>", |
| 1145 | "[SYSTEM_PROMPT]", |
| 1146 | "I", |
| 1147 | " am", |
| 1148 | " an", |
| 1149 | " AI", |
| 1150 | "[/SYSTEM_PROMPT]", |
| 1151 | "[AVAILABLE_TOOLS]", |
| 1152 | "[", |
| 1153 | '{"', |
| 1154 | "type", |
| 1155 | '":', |
| 1156 | ' "', |
| 1157 | "function", |
| 1158 | '",', |
| 1159 | ' "', |
| 1160 | "function", |
| 1161 | '":', |
| 1162 | ' {"', |
| 1163 | "name", |
| 1164 | '":', |
| 1165 | ' "', |
| 1166 | "get", |
| 1167 | "_", |
| 1168 | "weather", |
| 1169 | '",', |
| 1170 | ' "', |
| 1171 | "description", |
| 1172 | '":', |
| 1173 | ' "', |
| 1174 | "G", |
| 1175 | "ets", |
| 1176 | " the", |
| 1177 | " current", |
| 1178 | " weather", |
| 1179 | " in", |
| 1180 | " a", |
| 1181 | " city", |
| 1182 | '.",', |
| 1183 | ' "', |
| 1184 | "parameters", |
| 1185 | '":', |
| 1186 | ' {"', |
| 1187 | "type", |
| 1188 | '":', |
| 1189 | ' "', |
| 1190 | "object", |
| 1191 | '",', |
| 1192 | ' "', |
| 1193 | "properties", |
| 1194 | '":', |
| 1195 | ' {"', |
| 1196 | "city", |
| 1197 | '":', |
| 1198 | ' {"', |
| 1199 | "type", |
| 1200 | '":', |
| 1201 | ' "', |
| 1202 | "string", |
| 1203 | '",', |
| 1204 | ' "', |
| 1205 | "description", |
| 1206 | '":', |
| 1207 | ' "', |
| 1208 | "The", |
| 1209 | " city", |
| 1210 | " name", |
| 1211 | '"', |
| 1212 | "}},", |
| 1213 | ' "', |
| 1214 | "required", |
| 1215 | '":', |
| 1216 | ' ["', |
| 1217 | "city", |
| 1218 | '"]', |
| 1219 | "}}", |
| 1220 | "}]", |
| 1221 | "[/AVAILABLE_TOOLS]", |
| 1222 | "[INST]", |
| 1223 | "Hello", |
| 1224 | " world", |
| 1225 | " !", |
| 1226 | "[/INST]", |
| 1227 | "[TOOL_CALLS]", |
| 1228 | "get", |
| 1229 | "_", |
| 1230 | "weather", |
| 1231 | "[ARGS]", |
| 1232 | '{"', |
| 1233 | "city", |
| 1234 | '":', |
| 1235 | ' "', |
| 1236 | "Paris", |
| 1237 | '"}', |
| 1238 | "</s>", |
| 1239 | "[TOOL_RESULTS]", |
| 1240 | '{"', |
| 1241 | "temperature", |
| 1242 | '":', |
| 1243 | " ", |
| 1244 | "2", |
| 1245 | "0", |
| 1246 | ",", |
| 1247 | ' "', |
| 1248 | "unit", |
| 1249 | '":', |
| 1250 | ' "', |
| 1251 | "c", |
| 1252 | "elsius", |
| 1253 | '"}', |
| 1254 | "[/TOOL_RESULTS]", |
| 1255 | ], |
| 1256 | ) |
| 1257 | |
| 1258 | expected_strings = ( |
| 1259 | '[{"type": "function", "function": {"name": "get_weather", "description": "Gets the current weather in a city.", "parameters": {"type": "object", "properties": {"city": {"type": "string", "description": "The city name"}}, "required": ["city"]}}}] I am an AI\n\nHello world ![TOOL_CALLS][{"name": "get_weather", "arguments": {"city": "Paris"}, "id": "123456789"}] {"content": {"temperature": 20, "unit": "celsius"}, "call_id": "123456789"}', # noqa: E501 |
| 1260 | # v11+ tool-call decode emits the explicit [ARGS] separator between |
| 1261 | # the function name and its JSON arguments (get_weather[ARGS]{...}). |
| 1262 | 'I am an AI[{"type": "function", "function": {"name": "get_weather", "description": "Gets the current weather in a city.", "parameters": {"type": "object", "properties": {"city": {"type": "string", "description": "The city name"}}, "required": ["city"]}}}]Hello world ![TOOL_CALLS]get_weather[ARGS]{"city": "Paris"}{"temperature": 20, "unit": "celsius"}', # noqa: E501 |
| 1263 | ) |
| 1264 | |
| 1265 | assert ( |
| 1266 | mistral_tokenizer.convert_tokens_to_string( |
| 1267 | tokens[mistral_tokenizer.is_tekken] |
| 1268 | ) |
| 1269 | == expected_strings[mistral_tokenizer.is_tekken] |
| 1270 | ) |
| 1271 | |
| 1272 | assert mistral_tokenizer.convert_tokens_to_string([]) == "" |
| 1273 | |
| 1274 | @pytest.mark.parametrize( |
| 1275 | "skip_special_tokens,tuple_expected_tokens", |
| 1276 | ( |
| 1277 | ( |
| 1278 | True, |
| 1279 | ( |
| 1280 | [ |
| 1281 | "▁[", |
| 1282 | '{"', |
| 1283 | "type", |
| 1284 | '":', |
| 1285 | '▁"', |
| 1286 | "function", |
| 1287 | '",', |
| 1288 | '▁"', |
| 1289 | "function", |
| 1290 | '":', |
| 1291 | '▁{"', |
| 1292 | "name", |
| 1293 | '":', |
| 1294 | '▁"', |
| 1295 | "get", |
| 1296 | "_", |
| 1297 | "we", |
| 1298 | "ather", |
| 1299 | '",', |
| 1300 | '▁"', |
| 1301 | "description", |
| 1302 | '":', |
| 1303 | '▁"', |
| 1304 | "Get", |
| 1305 | "s", |
| 1306 | "▁the", |
| 1307 | "▁current", |
| 1308 | "▁weather", |
| 1309 | "▁in", |
| 1310 | "▁a", |
| 1311 | "▁city", |
| 1312 | '.",', |
| 1313 | '▁"', |
| 1314 | "parameters", |
| 1315 | '":', |
| 1316 | '▁{"', |
| 1317 | "type", |
| 1318 | '":', |
| 1319 | '▁"', |
| 1320 | "object", |
| 1321 | '",', |
| 1322 | '▁"', |
| 1323 | "properties", |
| 1324 | '":', |
| 1325 | '▁{"', |
| 1326 | "city", |
| 1327 | '":', |
| 1328 | '▁{"', |
| 1329 | "type", |
| 1330 | '":', |
| 1331 | '▁"', |
| 1332 | "string", |
| 1333 | '",', |
| 1334 | '▁"', |
| 1335 | "description", |
| 1336 | '":', |
| 1337 | '▁"', |
| 1338 | "The", |
| 1339 | "▁city", |
| 1340 | "▁name", |
| 1341 | '"', |
| 1342 | "}},", |
| 1343 | '▁"', |
| 1344 | "required", |
| 1345 | '":', |
| 1346 | '▁["', |
| 1347 | "city", |
| 1348 | '"]', |
| 1349 | "}}", |
| 1350 | "}]", |
| 1351 | "▁I", |
| 1352 | "▁am", |
| 1353 | "▁an", |
| 1354 | "▁AI", |
| 1355 | "<0x0A>", |
| 1356 | "<0x0A>", |
| 1357 | "Hello", |
| 1358 | "▁world", |
| 1359 | "▁!", |
| 1360 | "[TOOL_CALLS]", |
| 1361 | "▁[", |
| 1362 | '{"', |
| 1363 | "name", |
| 1364 | '":', |
| 1365 | '▁"', |
| 1366 | "get", |
| 1367 | "_", |
| 1368 | "we", |
| 1369 | "ather", |
| 1370 | '",', |
| 1371 | '▁"', |
| 1372 | "arguments", |
| 1373 | '":', |
| 1374 | '▁{"', |
| 1375 | "city", |
| 1376 | '":', |
| 1377 | '▁"', |
| 1378 | "Par", |
| 1379 | "is", |
| 1380 | '"},', |
| 1381 | '▁"', |
| 1382 | "id", |
| 1383 | '":', |
| 1384 | '▁"', |
| 1385 | "1", |
| 1386 | "2", |
| 1387 | "3", |
| 1388 | "4", |
| 1389 | "5", |
| 1390 | "6", |
| 1391 | "7", |
| 1392 | "8", |
| 1393 | "9", |
| 1394 | '"', |
| 1395 | "}]", |
| 1396 | '▁{"', |
| 1397 | "content", |
| 1398 | '":', |
| 1399 | '▁{"', |
| 1400 | "t", |
| 1401 | "emperature", |
| 1402 | '":', |
| 1403 | "▁", |
| 1404 | "2", |
| 1405 | "0", |
| 1406 | ",", |
| 1407 | '▁"', |
| 1408 | "unit", |
| 1409 | '":', |
| 1410 | '▁"', |
| 1411 | "c", |
| 1412 | "els", |
| 1413 | "ius", |
| 1414 | '"},', |
| 1415 | '▁"', |
| 1416 | "call", |
| 1417 | "_", |
| 1418 | "id", |
| 1419 | '":', |
| 1420 | '▁"', |
| 1421 | "1", |
| 1422 | "2", |
| 1423 | "3", |
| 1424 | "4", |
| 1425 | "5", |
| 1426 | "6", |
| 1427 | "7", |
| 1428 | "8", |
| 1429 | "9", |
| 1430 | '"}', |
| 1431 | ], |
| 1432 | [ |
| 1433 | "I", |
| 1434 | " am", |
| 1435 | " an", |
| 1436 | " AI", |
| 1437 | "[", |
| 1438 | '{"', |
| 1439 | "type", |
| 1440 | '":', |
| 1441 | ' "', |
| 1442 | "function", |
| 1443 | '",', |
| 1444 | ' "', |
| 1445 | "function", |
| 1446 | '":', |
| 1447 | ' {"', |
| 1448 | "name", |
| 1449 | '":', |
| 1450 | ' "', |
| 1451 | "get", |
| 1452 | "_", |
| 1453 | "weather", |
| 1454 | '",', |
| 1455 | ' "', |
| 1456 | "description", |
| 1457 | '":', |
| 1458 | ' "', |
| 1459 | "G", |
| 1460 | "ets", |
| 1461 | " the", |
| 1462 | " current", |
| 1463 | " weather", |
| 1464 | " in", |
| 1465 | " a", |
| 1466 | " city", |
| 1467 | '.",', |
| 1468 | ' "', |
| 1469 | "parameters", |
| 1470 | '":', |
| 1471 | ' {"', |
| 1472 | "type", |
| 1473 | '":', |
| 1474 | ' "', |
| 1475 | "object", |
| 1476 | '",', |
| 1477 | ' "', |
| 1478 | "properties", |
| 1479 | '":', |
| 1480 | ' {"', |
| 1481 | "city", |
| 1482 | '":', |
| 1483 | ' {"', |
| 1484 | "type", |
| 1485 | '":', |
| 1486 | ' "', |
| 1487 | "string", |
| 1488 | '",', |
| 1489 | ' "', |
| 1490 | "description", |
| 1491 | '":', |
| 1492 | ' "', |
| 1493 | "The", |
| 1494 | " city", |
| 1495 | " name", |
| 1496 | '"', |
| 1497 | "}},", |
| 1498 | ' "', |
| 1499 | "required", |
| 1500 | '":', |
| 1501 | ' ["', |
| 1502 | "city", |
| 1503 | '"]', |
| 1504 | "}}", |
| 1505 | "}]", |
| 1506 | "Hello", |
| 1507 | " world", |
| 1508 | " !", |
| 1509 | "[TOOL_CALLS]", |
| 1510 | "get", |
| 1511 | "_", |
| 1512 | "weather", |
| 1513 | "[ARGS]", |
| 1514 | '{"', |
| 1515 | "city", |
| 1516 | '":', |
| 1517 | ' "', |
| 1518 | "Paris", |
| 1519 | '"}', |
| 1520 | '{"', |
| 1521 | "temperature", |
| 1522 | '":', |
| 1523 | " ", |
| 1524 | "2", |
| 1525 | "0", |
| 1526 | ",", |
| 1527 | ' "', |
| 1528 | "unit", |
| 1529 | '":', |
| 1530 | ' "', |
| 1531 | "c", |
| 1532 | "elsius", |
| 1533 | '"}', |
| 1534 | ], |
| 1535 | ), |
| 1536 | ), |
| 1537 | ( |
| 1538 | False, |
| 1539 | ( |
| 1540 | [ |
| 1541 | "<s>", |
| 1542 | "[AVAILABLE_TOOLS]", |
| 1543 | "▁[", |
| 1544 | '{"', |
| 1545 | "type", |
| 1546 | '":', |
| 1547 | '▁"', |
| 1548 | "function", |
| 1549 | '",', |
| 1550 | '▁"', |
| 1551 | "function", |
| 1552 | '":', |
| 1553 | '▁{"', |
| 1554 | "name", |
| 1555 | '":', |
| 1556 | '▁"', |
| 1557 | "get", |
| 1558 | "_", |
| 1559 | "we", |
| 1560 | "ather", |
| 1561 | '",', |
| 1562 | '▁"', |
| 1563 | "description", |
| 1564 | '":', |
| 1565 | '▁"', |
| 1566 | "Get", |
| 1567 | "s", |
| 1568 | "▁the", |
| 1569 | "▁current", |
| 1570 | "▁weather", |
| 1571 | "▁in", |
| 1572 | "▁a", |
| 1573 | "▁city", |
| 1574 | '.",', |
| 1575 | '▁"', |
| 1576 | "parameters", |
| 1577 | '":', |
| 1578 | '▁{"', |
| 1579 | "type", |
| 1580 | '":', |
| 1581 | '▁"', |
| 1582 | "object", |
| 1583 | '",', |
| 1584 | '▁"', |
| 1585 | "properties", |
| 1586 | '":', |
| 1587 | '▁{"', |
| 1588 | "city", |
| 1589 | '":', |
| 1590 | '▁{"', |
| 1591 | "type", |
| 1592 | '":', |
| 1593 | '▁"', |
| 1594 | "string", |
| 1595 | '",', |
| 1596 | '▁"', |
| 1597 | "description", |
| 1598 | '":', |
| 1599 | '▁"', |
| 1600 | "The", |
| 1601 | "▁city", |
| 1602 | "▁name", |
| 1603 | '"', |
| 1604 | "}},", |
| 1605 | '▁"', |
| 1606 | "required", |
| 1607 | '":', |
| 1608 | '▁["', |
| 1609 | "city", |
| 1610 | '"]', |
| 1611 | "}}", |
| 1612 | "}]", |
| 1613 | "[/AVAILABLE_TOOLS]", |
| 1614 | "[INST]", |
| 1615 | "▁I", |
| 1616 | "▁am", |
| 1617 | "▁an", |
| 1618 | "▁AI", |
| 1619 | "<0x0A>", |
| 1620 | "<0x0A>", |
| 1621 | "Hello", |
| 1622 | "▁world", |
| 1623 | "▁!", |
| 1624 | "[/INST]", |
| 1625 | "[TOOL_CALLS]", |
| 1626 | "▁[", |
| 1627 | '{"', |
| 1628 | "name", |
| 1629 | '":', |
| 1630 | '▁"', |
| 1631 | "get", |
| 1632 | "_", |
| 1633 | "we", |
| 1634 | "ather", |
| 1635 | '",', |
| 1636 | '▁"', |
| 1637 | "arguments", |
| 1638 | '":', |
| 1639 | '▁{"', |
| 1640 | "city", |
| 1641 | '":', |
| 1642 | '▁"', |
| 1643 | "Par", |
| 1644 | "is", |
| 1645 | '"},', |
| 1646 | '▁"', |
| 1647 | "id", |
| 1648 | '":', |
| 1649 | '▁"', |
| 1650 | "1", |
| 1651 | "2", |
| 1652 | "3", |
| 1653 | "4", |
| 1654 | "5", |
| 1655 | "6", |
| 1656 | "7", |
| 1657 | "8", |
| 1658 | "9", |
| 1659 | '"', |
| 1660 | "}]", |
| 1661 | "</s>", |
| 1662 | "[TOOL_RESULTS]", |
| 1663 | '▁{"', |
| 1664 | "content", |
| 1665 | '":', |
| 1666 | '▁{"', |
| 1667 | "t", |
| 1668 | "emperature", |
| 1669 | '":', |
| 1670 | "▁", |
| 1671 | "2", |
| 1672 | "0", |
| 1673 | ",", |
| 1674 | '▁"', |
| 1675 | "unit", |
| 1676 | '":', |
| 1677 | '▁"', |
| 1678 | "c", |
| 1679 | "els", |
| 1680 | "ius", |
| 1681 | '"},', |
| 1682 | '▁"', |
| 1683 | "call", |
| 1684 | "_", |
| 1685 | "id", |
| 1686 | '":', |
| 1687 | '▁"', |
| 1688 | "1", |
| 1689 | "2", |
| 1690 | "3", |
| 1691 | "4", |
| 1692 | "5", |
| 1693 | "6", |
| 1694 | "7", |
| 1695 | "8", |
| 1696 | "9", |
| 1697 | '"}', |
| 1698 | "[/TOOL_RESULTS]", |
| 1699 | ], |
| 1700 | [ |
| 1701 | "<s>", |
| 1702 | "[SYSTEM_PROMPT]", |
| 1703 | "I", |
| 1704 | " am", |
| 1705 | " an", |
| 1706 | " AI", |
| 1707 | "[/SYSTEM_PROMPT]", |
| 1708 | "[AVAILABLE_TOOLS]", |
| 1709 | "[", |
| 1710 | '{"', |
| 1711 | "type", |
| 1712 | '":', |
| 1713 | ' "', |
| 1714 | "function", |
| 1715 | '",', |
| 1716 | ' "', |
| 1717 | "function", |
| 1718 | '":', |
| 1719 | ' {"', |
| 1720 | "name", |
| 1721 | '":', |
| 1722 | ' "', |
| 1723 | "get", |
| 1724 | "_", |
| 1725 | "weather", |
| 1726 | '",', |
| 1727 | ' "', |
| 1728 | "description", |
| 1729 | '":', |
| 1730 | ' "', |
| 1731 | "G", |
| 1732 | "ets", |
| 1733 | " the", |
| 1734 | " current", |
| 1735 | " weather", |
| 1736 | " in", |
| 1737 | " a", |
| 1738 | " city", |
| 1739 | '.",', |
| 1740 | ' "', |
| 1741 | "parameters", |
| 1742 | '":', |
| 1743 | ' {"', |
| 1744 | "type", |
| 1745 | '":', |
| 1746 | ' "', |
| 1747 | "object", |
| 1748 | '",', |
| 1749 | ' "', |
| 1750 | "properties", |
| 1751 | '":', |
| 1752 | ' {"', |
| 1753 | "city", |
| 1754 | '":', |
| 1755 | ' {"', |
| 1756 | "type", |
| 1757 | '":', |
| 1758 | ' "', |
| 1759 | "string", |
| 1760 | '",', |
| 1761 | ' "', |
| 1762 | "description", |
| 1763 | '":', |
| 1764 | ' "', |
| 1765 | "The", |
| 1766 | " city", |
| 1767 | " name", |
| 1768 | '"', |
| 1769 | "}},", |
| 1770 | ' "', |
| 1771 | "required", |
| 1772 | '":', |
| 1773 | ' ["', |
| 1774 | "city", |
| 1775 | '"]', |
| 1776 | "}}", |
| 1777 | "}]", |
| 1778 | "[/AVAILABLE_TOOLS]", |
| 1779 | "[INST]", |
| 1780 | "Hello", |
| 1781 | " world", |
| 1782 | " !", |
| 1783 | "[/INST]", |
| 1784 | "[TOOL_CALLS]", |
| 1785 | "get", |
| 1786 | "_", |
| 1787 | "weather", |
| 1788 | "[ARGS]", |
| 1789 | '{"', |
| 1790 | "city", |
| 1791 | '":', |
| 1792 | ' "', |
| 1793 | "Paris", |
| 1794 | '"}', |
| 1795 | "</s>", |
| 1796 | "[TOOL_RESULTS]", |
| 1797 | '{"', |
| 1798 | "temperature", |
| 1799 | '":', |
| 1800 | " ", |
| 1801 | "2", |
| 1802 | "0", |
| 1803 | ",", |
| 1804 | ' "', |
| 1805 | "unit", |
| 1806 | '":', |
| 1807 | ' "', |
| 1808 | "c", |
| 1809 | "elsius", |
| 1810 | '"}', |
| 1811 | "[/TOOL_RESULTS]", |
| 1812 | ], |
| 1813 | ), |
| 1814 | ), |
| 1815 | ), |
| 1816 | ) |
| 1817 | def test_convert_ids_to_tokens( |
| 1818 | self, |
| 1819 | mistral_tokenizer: MistralTokenizer, |
| 1820 | skip_special_tokens: bool, |
| 1821 | tuple_expected_tokens: tuple[list[str], list[str]], |
| 1822 | ): |
| 1823 | tuple_ids = ( |
| 1824 | [ |
| 1825 | 1, |
| 1826 | 6, |
| 1827 | 1501, |
| 1828 | 7567, |
| 1829 | 1891, |
| 1830 | 2032, |
| 1831 | 1113, |
| 1832 | 3396, |
| 1833 | 1316, |
| 1834 | 1113, |
| 1835 | 3396, |
| 1836 | 2032, |
| 1837 | 10598, |
| 1838 | 1629, |
| 1839 | 2032, |
| 1840 | 1113, |
| 1841 | 1295, |
| 1842 | 29498, |
| 1843 | 1537, |
| 1844 | 1991, |
| 1845 | 1316, |
| 1846 | 1113, |
| 1847 | 7286, |
| 1848 | 2032, |
| 1849 | 1113, |
| 1850 | 2226, |
| 1851 | 29481, |
| 1852 | 1040, |
| 1853 | 2636, |
| 1854 | 8854, |
| 1855 | 1065, |
| 1856 | 1032, |
| 1857 | 3758, |
| 1858 | 9959, |
| 1859 | 1113, |
| 1860 | 12206, |
| 1861 | 2032, |
| 1862 | 10598, |
| 1863 | 1891, |
| 1864 | 2032, |
| 1865 | 1113, |
| 1866 | 3582, |
| 1867 | 1316, |
| 1868 | 1113, |
| 1869 | 11491, |
| 1870 | 2032, |
| 1871 | 10598, |
| 1872 | 19141, |
| 1873 | 2032, |
| 1874 | 10598, |
| 1875 | 1891, |
| 1876 | 2032, |
| 1877 | 1113, |
| 1878 | 2195, |
| 1879 | 1316, |
| 1880 | 1113, |
| 1881 | 7286, |
| 1882 | 2032, |
| 1883 | 1113, |
| 1884 | 1782, |
| 1885 | 3758, |
| 1886 | 1909, |
| 1887 | 29507, |
| 1888 | 11549, |
| 1889 | 1113, |
| 1890 | 11661, |
| 1891 | 2032, |
| 1892 | 8135, |
| 1893 | 19141, |
| 1894 | 3010, |
| 1895 | 1743, |
| 1896 | 10925, |
| 1897 | 7, |
| 1898 | 3, |
| 1899 | 1083, |
| 1900 | 1605, |
| 1901 | 1164, |
| 1902 | 16875, |
| 1903 | 781, |
| 1904 | 781, |
| 1905 | 16998, |
| 1906 | 2294, |
| 1907 | 1686, |
| 1908 | 4, |
| 1909 | 5, |
| 1910 | 1501, |
| 1911 | 7567, |
| 1912 | 1629, |
| 1913 | 2032, |
| 1914 | 1113, |
| 1915 | 1295, |
| 1916 | 29498, |
| 1917 | 1537, |
| 1918 | 1991, |
| 1919 | 1316, |
| 1920 | 1113, |
| 1921 | 17452, |
| 1922 | 2032, |
| 1923 | 10598, |
| 1924 | 19141, |
| 1925 | 2032, |
| 1926 | 1113, |
| 1927 | 4684, |
| 1928 | 1046, |
| 1929 | 8474, |
| 1930 | 1113, |
| 1931 | 1081, |
| 1932 | 2032, |
| 1933 | 1113, |
| 1934 | 29508, |
| 1935 | 29518, |
| 1936 | 29538, |
| 1937 | 29549, |
| 1938 | 29550, |
| 1939 | 29552, |
| 1940 | 29555, |
| 1941 | 29551, |
| 1942 | 29542, |
| 1943 | 29507, |
| 1944 | 10925, |
| 1945 | 2, |
| 1946 | 8, |
| 1947 | 10598, |
| 1948 | 4557, |
| 1949 | 2032, |
| 1950 | 10598, |
| 1951 | 29475, |
| 1952 | 17329, |
| 1953 | 2032, |
| 1954 | 29473, |
| 1955 | 29518, |
| 1956 | 29502, |
| 1957 | 29493, |
| 1958 | 1113, |
| 1959 | 6074, |
| 1960 | 2032, |
| 1961 | 1113, |
| 1962 | 29485, |
| 1963 | 1958, |
| 1964 | 3938, |
| 1965 | 8474, |
| 1966 | 1113, |
| 1967 | 3613, |
| 1968 | 29498, |
| 1969 | 1081, |
| 1970 | 2032, |
| 1971 | 1113, |
| 1972 | 29508, |
| 1973 | 29518, |
| 1974 | 29538, |
| 1975 | 29549, |
| 1976 | 29550, |
| 1977 | 29552, |
| 1978 | 29555, |
| 1979 | 29551, |
| 1980 | 29542, |
| 1981 | 18163, |
| 1982 | 9, |
| 1983 | ], |
| 1984 | [ |
| 1985 | 1, |
| 1986 | 17, |
| 1987 | 1073, |
| 1988 | 1855, |
| 1989 | 1420, |
| 1990 | 26554, |
| 1991 | 18, |
| 1992 | 5, |
| 1993 | 1091, |
| 1994 | 19227, |
| 1995 | 4994, |
| 1996 | 2811, |
| 1997 | 1429, |
| 1998 | 5165, |
| 1999 | 1897, |
| 2000 | 1429, |
| 2001 | 5165, |
| 2002 | 2811, |
| 2003 | 16753, |
| 2004 | 2391, |
| 2005 | 2811, |
| 2006 | 1429, |
| 2007 | 1689, |
| 2008 | 1095, |
| 2009 | 45629, |
| 2010 | 1897, |
| 2011 | 1429, |
| 2012 | 14653, |
| 2013 | 2811, |
| 2014 | 1429, |
| 2015 | 1071, |
| 2016 | 3083, |
| 2017 | 1278, |
| 2018 | 3519, |
| 2019 | 17253, |
| 2020 | 1294, |
| 2021 | 1261, |
| 2022 | 5970, |
| 2023 | 39249, |
| 2024 | 1429, |
| 2025 | 26204, |
| 2026 | 2811, |
| 2027 | 16753, |
| 2028 | 4994, |
| 2029 | 2811, |
| 2030 | 1429, |
| 2031 | 6371, |
| 2032 | 1897, |
| 2033 | 1429, |
| 2034 | 48649, |
| 2035 | 2811, |
| 2036 | 16753, |
| 2037 | 29363, |
| 2038 | 2811, |
| 2039 | 16753, |
| 2040 | 4994, |
| 2041 | 2811, |
| 2042 | 1429, |
| 2043 | 3607, |
| 2044 | 1897, |
| 2045 | 1429, |
| 2046 | 14653, |
| 2047 | 2811, |
| 2048 | 1429, |
| 2049 | 1784, |
| 2050 | 5970, |
| 2051 | 2564, |
| 2052 | 1034, |
| 2053 | 47579, |
| 2054 | 1429, |
| 2055 | 15760, |
| 2056 | 2811, |
| 2057 | 12161, |
| 2058 | 29363, |
| 2059 | 4964, |
| 2060 | 2821, |
| 2061 | 27028, |
| 2062 | 6, |
| 2063 | 3, |
| 2064 | 22177, |
| 2065 | 4304, |
| 2066 | 2662, |
| 2067 | 4, |
| 2068 | 9, |
| 2069 | 1689, |
| 2070 | 1095, |
| 2071 | 45629, |
| 2072 | 32, |
| 2073 | 19227, |
| 2074 | 29363, |
| 2075 | 2811, |
| 2076 | 1429, |
| 2077 | 42572, |
| 2078 | 46005, |
| 2079 | 2, |
| 2080 | 7, |
| 2081 | 19227, |
| 2082 | 113824, |
| 2083 | 2811, |
| 2084 | 1032, |
| 2085 | 1050, |
| 2086 | 1048, |
| 2087 | 1044, |
| 2088 | 1429, |
| 2089 | 8979, |
| 2090 | 2811, |
| 2091 | 1429, |
| 2092 | 1099, |
| 2093 | 79092, |
| 2094 | 46005, |
| 2095 | 8, |
| 2096 | ], |
| 2097 | ) |
| 2098 | |
| 2099 | ids = tuple_ids[mistral_tokenizer.is_tekken] |
| 2100 | expected_tokens = tuple_expected_tokens[mistral_tokenizer.is_tekken] |
| 2101 | actual_tokens = mistral_tokenizer.convert_ids_to_tokens( |
| 2102 | ids, skip_special_tokens=skip_special_tokens |
| 2103 | ) |
| 2104 | assert actual_tokens == expected_tokens |
| 2105 | |
| 2106 | assert mistral_tokenizer.convert_ids_to_tokens([]) == [] |
| 2107 | |
| 2108 | def test_grammar_factory(self, mistral_tokenizer: MistralTokenizer) -> None: |
| 2109 | # works in this case cause Mistral 7B is < v11 and SPM |
| 2110 | if not mistral_tokenizer.is_tekken: |
| 2111 | with pytest.raises(AttributeError): |
| 2112 | mistral_tokenizer.grammar_factory # noqa: B018 |
| 2113 | return |
| 2114 | factory = mistral_tokenizer.grammar_factory |
| 2115 | assert isinstance(factory, GrammarFactory) |
| 2116 | |
| 2117 | # Test caching |
| 2118 | factory_2 = mistral_tokenizer.grammar_factory |
| 2119 | assert factory is factory_2 |
| 2120 | |
| 2121 | def test_llg_tokenizer(self, mistral_tokenizer: MistralTokenizer) -> None: |
| 2122 | if not mistral_tokenizer.is_tekken: |
| 2123 | with pytest.raises(ValueError): |
| 2124 | mistral_tokenizer.llg_tokenizer # noqa: B018 |
| 2125 | return |
| 2126 | |
| 2127 | llg_tokenizer = mistral_tokenizer.llg_tokenizer |
| 2128 | assert isinstance(llg_tokenizer, llguidance.LLTokenizer) |
| 2129 | |
| 2130 | # Test caching |
| 2131 | llg_tokenizer_2 = mistral_tokenizer.llg_tokenizer |
| 2132 | assert llg_tokenizer is llg_tokenizer_2 |
| 2133 | |
| 2134 | @pytest.mark.parametrize( |
| 2135 | "messages,tools,tekken_expected_substrings,spm_expected_substrings", |
| 2136 | [ |
| 2137 | pytest.param( |
| 2138 | [{"role": "user", "content": "Hello"}], |
| 2139 | [{"type": "function", "function": {"name": "do_nothing"}}], |
| 2140 | ["do_nothing", '"description": ""', '"parameters": {}'], |
| 2141 | ["do_nothing", '"description":▁""', '"parameters":▁{}'], |
| 2142 | id="tool_without_description_and_parameters", |
| 2143 | ), |
| 2144 | pytest.param( |
| 2145 | [ |
| 2146 | {"role": "user", "content": "Do nothing"}, |
| 2147 | { |
| 2148 | "role": "assistant", |
| 2149 | "content": "", |
| 2150 | "tool_calls": [ |
| 2151 | { |
| 2152 | "id": "123456789", |
| 2153 | "type": "function", |
| 2154 | "function": { |
| 2155 | "name": "do_nothing", |
| 2156 | "arguments": None, |
| 2157 | }, |
| 2158 | } |
| 2159 | ], |
| 2160 | }, |
| 2161 | { |
| 2162 | "role": "tool", |
| 2163 | "tool_call_id": "123456789", |
| 2164 | "content": "done", |
| 2165 | }, |
| 2166 | ], |
| 2167 | [{"type": "function", "function": {"name": "do_nothing"}}], |
| 2168 | ["do_nothing"], |
| 2169 | ["do_nothing"], |
| 2170 | id="tool_call_with_none_arguments", |
| 2171 | ), |
| 2172 | ], |
| 2173 | ) |
| 2174 | def test_apply_chat_template_tool_optional_fields( |
| 2175 | self, |
| 2176 | mistral_tokenizer: MistralTokenizer, |
| 2177 | messages: list[ChatCompletionMessageParam], |
| 2178 | tools: list[dict[str, Any]], |
| 2179 | tekken_expected_substrings: list[str], |
| 2180 | spm_expected_substrings: list[str], |
| 2181 | ) -> None: |
| 2182 | output = mistral_tokenizer.apply_chat_template( |
| 2183 | messages, |
| 2184 | tools=tools, |
| 2185 | add_generation_prompt=True, |
| 2186 | ) |
| 2187 | decoded = mistral_tokenizer.tokenizer.decode(output, SpecialTokenPolicy.KEEP) |
| 2188 | |
| 2189 | expected = ( |
| 2190 | tekken_expected_substrings |
| 2191 | if mistral_tokenizer.is_tekken |
| 2192 | else spm_expected_substrings |
| 2193 | ) |
| 2194 | for substring in expected: |
| 2195 | assert substring in decoded |
| 2196 | |
| 2197 | def test_apply_chat_template_tools_not_mutated( |
| 2198 | self, mistral_tokenizer: MistralTokenizer |
| 2199 | ) -> None: |
| 2200 | messages: list[ChatCompletionMessageParam] = [ |
| 2201 | {"role": "user", "content": "Hello"}, |
| 2202 | ] |
| 2203 | tools: list[dict[str, Any]] = [ |
| 2204 | { |
| 2205 | "type": "function", |
| 2206 | "function": { |
| 2207 | "name": "get_weather", |
| 2208 | "description": "Gets weather.", |
| 2209 | "parameters": { |
| 2210 | "type": "object", |
| 2211 | "properties": { |
| 2212 | "city": {"type": "string"}, |
| 2213 | }, |
| 2214 | }, |
| 2215 | }, |
| 2216 | }, |
| 2217 | ] |
| 2218 | original_tools = copy.deepcopy(tools) |
| 2219 | |
| 2220 | mistral_tokenizer.apply_chat_template( |
| 2221 | messages, |
| 2222 | tools=tools, |
| 2223 | add_generation_prompt=True, |
| 2224 | ) |
| 2225 | |
| 2226 | assert tools == original_tools |
| 2227 | |
| 2228 | @pytest.mark.parametrize( |
| 2229 | "reasoning_key", |
| 2230 | ["reasoning", "reasoning_content"], |
| 2231 | ) |
| 2232 | def test_apply_chat_template_reasoning_assistant( |
| 2233 | self, mistral_tokenizer: MistralTokenizer, reasoning_key: str |
| 2234 | ) -> None: |
| 2235 | if not mistral_tokenizer.is_tekken: |
| 2236 | pytest.skip("Reasoning tokens only supported on tekken tokenizers") |
| 2237 | |
| 2238 | messages: list[dict[str, Any]] = [ |
| 2239 | {"role": "user", "content": "What is 2+2?"}, |
| 2240 | { |
| 2241 | "role": "assistant", |
| 2242 | "content": "4", |
| 2243 | reasoning_key: "2+2 equals 4", |
| 2244 | }, |
| 2245 | {"role": "user", "content": "Are you sure?"}, |
| 2246 | ] |
| 2247 | |
| 2248 | output = mistral_tokenizer.apply_chat_template( |
| 2249 | messages, |
| 2250 | add_generation_prompt=True, |
| 2251 | ) |
| 2252 | decoded = mistral_tokenizer.tokenizer.decode(output, SpecialTokenPolicy.KEEP) |
| 2253 | |
| 2254 | assert "[THINK]2+2 equals 4[/THINK]" in decoded |
| 2255 | |
| 2256 | |
| 2257 | def test_convert_ids_to_tokens_pre_args_tekken(): |
| 2258 | """convert_ids_to_tokens must not raise on pre-[ARGS] Tekken tokenizers. |
| 2259 | |
| 2260 | Tekken tokenizers before v11 (e.g. Ministral-8B-Instruct-2410) have no |
| 2261 | [ARGS] special token. This guards the is_special([ARGS]) gate in |
| 2262 | MistralTokenizer.convert_ids_to_tokens, which must skip [ARGS] rather than |
| 2263 | call get_special_token(args) unconditionally (which raises on Tekken). |
| 2264 | """ |
| 2265 | tokenizer = MistralTokenizer.from_pretrained("mistralai/Ministral-8B-Instruct-2410") |
| 2266 | assert tokenizer.is_tekken |
| 2267 | assert not tokenizer.tokenizer.is_special(SpecialTokens.args) |
| 2268 | |
| 2269 | ids = tokenizer.encode("Hello world !", add_special_tokens=False) |
| 2270 | tokens = tokenizer.convert_ids_to_tokens(ids, skip_special_tokens=True) |
| 2271 | assert len(tokens) > 0 |
| 2272 | |
| 2273 | |
| 2274 | # --------------------------------------------------------------------------- |
| 2275 | # validate_request_params – reasoning_effort validation (no tokenizer needed) |
| 2276 | # --------------------------------------------------------------------------- |
| 2277 | |
| 2278 | |
| 2279 | @pytest.mark.parametrize("reasoning_effort", [None, "none", "high"]) |
| 2280 | def test_validate_request_params_valid_effort(reasoning_effort: str | None) -> None: |
| 2281 | request = ChatCompletionRequest( |
| 2282 | model="test-model", |
| 2283 | messages=[], |
| 2284 | reasoning_effort=reasoning_effort, |
| 2285 | ) |
| 2286 | assert validate_request_params(request) is None |
| 2287 | |
| 2288 | |
| 2289 | def test_validate_request_params_rejects_unsupported_effort() -> None: |
| 2290 | request = ChatCompletionRequest( |
| 2291 | model="test-model", |
| 2292 | messages=[], |
| 2293 | reasoning_effort="medium", |
| 2294 | ) |
| 2295 | with pytest.raises(VLLMValidationError, match="reasoning_effort") as exc_info: |
| 2296 | validate_request_params(request) |
| 2297 | assert exc_info.value.parameter == "reasoning_effort" |
| 2298 | |
| 2299 | |
| 2300 | def test_validate_request_params_rejects_chat_template() -> None: |
| 2301 | request = ChatCompletionRequest( |
| 2302 | model="test-model", |
| 2303 | messages=[], |
| 2304 | chat_template="{{ messages }}", |
| 2305 | ) |
| 2306 | with pytest.raises(VLLMValidationError, match="chat_template") as exc_info: |
| 2307 | validate_request_params(request) |
| 2308 | assert exc_info.value.parameter == "chat_template" |
| 2309 | |
| 2310 | |
| 2311 | # --------------------------------------------------------------------------- |
| 2312 | # apply_chat_template reasoning_effort passthrough (v15 tokenizer) |
| 2313 | # --------------------------------------------------------------------------- |
| 2314 | |
| 2315 | _PASSTHROUGH_MESSAGES = [{"role": "user", "content": "Hello"}] |
| 2316 | |
| 2317 | |
| 2318 | @pytest.fixture(scope="module") |
| 2319 | def v15_mistral_tokenizer() -> MistralTokenizer: |
| 2320 | """Load the v15 Mistral tokenizer; skip if unavailable.""" |
| 2321 | try: |
| 2322 | return MistralTokenizer.from_pretrained("mistralai/Mistral-Small-4-119B-2603") |
| 2323 | except Exception: |
| 2324 | pytest.skip("v15 tokenizer unavailable") |
| 2325 | raise AssertionError("unreachable") |
| 2326 | |
| 2327 | |
| 2328 | @pytest.fixture(scope="module") |
| 2329 | def v13_mistral_tokenizer() -> MistralTokenizer: |
| 2330 | """Load the v13 Mistral tokenizer; skip if unavailable.""" |
| 2331 | try: |
| 2332 | return MistralTokenizer.from_pretrained("mistralai/Magistral-Small-2509") |
| 2333 | except Exception: |
| 2334 | pytest.skip("v13 tokenizer unavailable") |
| 2335 | raise AssertionError("unreachable") |
| 2336 | |
| 2337 | |
| 2338 | def test_v15_apply_chat_template_passes_reasoning_effort_high( |
| 2339 | monkeypatch: pytest.MonkeyPatch, |
| 2340 | v15_mistral_tokenizer: MistralTokenizer, |
| 2341 | ) -> None: |
| 2342 | captured_kwargs: list[dict] = [] |
| 2343 | |
| 2344 | def fake_apply_chat_template(**kwargs): |
| 2345 | captured_kwargs.append(kwargs) |
| 2346 | return [0] |
| 2347 | |
| 2348 | monkeypatch.setattr( |
| 2349 | v15_mistral_tokenizer.transformers_tokenizer, |
| 2350 | "apply_chat_template", |
| 2351 | fake_apply_chat_template, |
| 2352 | ) |
| 2353 | |
| 2354 | v15_mistral_tokenizer.apply_chat_template( |
| 2355 | messages=_PASSTHROUGH_MESSAGES, |
| 2356 | reasoning_effort="high", |
| 2357 | ) |
| 2358 | |
| 2359 | assert captured_kwargs[-1]["reasoning_effort"] == "high" |
| 2360 | |
| 2361 | |
| 2362 | def test_v15_apply_chat_template_passes_reasoning_effort_none_by_default( |
| 2363 | monkeypatch: pytest.MonkeyPatch, |
| 2364 | v15_mistral_tokenizer: MistralTokenizer, |
| 2365 | ) -> None: |
| 2366 | captured_kwargs: list[dict] = [] |
| 2367 | |
| 2368 | def fake_apply_chat_template(**kwargs): |
| 2369 | captured_kwargs.append(kwargs) |
| 2370 | return [0] |
| 2371 | |
| 2372 | monkeypatch.setattr( |
| 2373 | v15_mistral_tokenizer.transformers_tokenizer, |
| 2374 | "apply_chat_template", |
| 2375 | fake_apply_chat_template, |
| 2376 | ) |
| 2377 | |
| 2378 | v15_mistral_tokenizer.apply_chat_template(messages=_PASSTHROUGH_MESSAGES) |
| 2379 | |
| 2380 | assert "reasoning_effort" in captured_kwargs[-1] |
| 2381 | assert captured_kwargs[-1]["reasoning_effort"] is None |
| 2382 | |
| 2383 | |
| 2384 | def test_pre_v15_apply_chat_template_omits_reasoning_effort( |
| 2385 | monkeypatch: pytest.MonkeyPatch, |
| 2386 | v13_mistral_tokenizer: MistralTokenizer, |
| 2387 | ) -> None: |
| 2388 | captured_kwargs: list[dict] = [] |
| 2389 | |
| 2390 | def fake_apply_chat_template(**kwargs): |
| 2391 | captured_kwargs.append(kwargs) |
| 2392 | return [0] |
| 2393 | |
| 2394 | monkeypatch.setattr( |
| 2395 | v13_mistral_tokenizer.transformers_tokenizer, |
| 2396 | "apply_chat_template", |
| 2397 | fake_apply_chat_template, |
| 2398 | ) |
| 2399 | |
| 2400 | v13_mistral_tokenizer.apply_chat_template( |
| 2401 | messages=_PASSTHROUGH_MESSAGES, |
| 2402 | reasoning_effort="high", |
| 2403 | ) |
| 2404 | |
| 2405 | assert "reasoning_effort" not in captured_kwargs[-1] |