File size: 12,837 Bytes
da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb 18bf871 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb da35c69 44529bb |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 |
from dataclasses import dataclass
from typing import Any, Final
from llmdataparser.base_parser import HuggingFaceDatasetParser, HuggingFaceParseEntry
from llmdataparser.prompts import MMLU_PRO_SYSTEM_PROMPT, MMLU_SYSTEM_PROMPT
MMLU_VALID_ANSWERS: Final[set[str]] = {"A", "B", "C", "D"}
MMLU_PRO_VALID_ANSWERS: Final[set[str]] = {
"A",
"B",
"C",
"D",
"E",
"F",
"G",
"H",
"I",
"J",
}
MMLU_VALID_ANSWER_STR: Final[str] = ", ".join(sorted(MMLU_VALID_ANSWERS))
MMLU_PRO_VALID_ANSWER_STR: Final[str] = ", ".join(sorted(MMLU_PRO_VALID_ANSWERS))
@dataclass(frozen=True, kw_only=True, slots=True)
class MMLUParseEntry(HuggingFaceParseEntry):
"""Custom entry class for MMLU, with fields specific to this dataset parser."""
raw_choices: list[str]
task_name: str
@classmethod
def create(
cls,
prompt: str,
answer: str,
raw_question: str,
raw_choices: list[str],
raw_answer: str,
task_name: str,
) -> "MMLUParseEntry":
if answer not in MMLU_VALID_ANSWERS:
raise ValueError(
f"Invalid answer_letter '{answer}'; must be one of {MMLU_VALID_ANSWER_STR}"
)
if not task_name:
raise ValueError("Task name cannot be empty")
return cls(
prompt=prompt,
answer=answer,
raw_question=raw_question,
raw_answer=raw_answer,
raw_choices=raw_choices,
task_name=task_name,
)
@dataclass(frozen=True, kw_only=True, slots=True)
class MMLUProParseEntry(HuggingFaceParseEntry):
"""Custom entry class for MMLU, with fields specific to this dataset parser."""
raw_choices: list[str]
task_name: str
@classmethod
def create(
cls,
prompt: str,
answer: str,
raw_question: str,
raw_choices: list[str],
raw_answer: str,
task_name: str,
) -> "MMLUProParseEntry":
if answer not in MMLU_PRO_VALID_ANSWERS:
raise ValueError(
f"Invalid answer_letter '{answer}'; must be one of {MMLU_PRO_VALID_ANSWER_STR}"
)
if not task_name:
raise ValueError("Task name cannot be empty")
return cls(
prompt=prompt,
answer=answer,
raw_question=raw_question,
raw_choices=raw_choices,
raw_answer=raw_answer,
task_name=task_name,
)
class MMLUDatasetParser(HuggingFaceDatasetParser[MMLUParseEntry]):
"""Base class for MMLU dataset parsers with common functionality."""
_default_system_prompt = MMLU_SYSTEM_PROMPT
def _get_task_from_entry(self, data_entry: dict[str, Any]) -> str:
"""Get the task name from the data entry or default task name."""
task_name = data_entry.get("subject")
return task_name if task_name else (self._current_task or self._default_task)
def process_entry(
self, row: dict[str, Any], task_name: str | None = None, **kwargs: Any
) -> MMLUParseEntry:
"""
Generate a prompt and expected answer from the given row.
Args:
row: A data point to be formatted.
task_name: Optional task name for the entry.
**kwargs: Additional keyword arguments.
Returns:
MMLUParseEntry: The formatted entry object.
"""
task = task_name or self._get_current_task(row)
# Ensure task is not None
final_task = task or self._default_task
choices = "\n".join(
f"{chr(65 + i)}. {choice}" for i, choice in enumerate(row["choices"])
)
raw_question = row["question"]
raw_choices = row["choices"]
raw_answer = str(row["answer"]) # Ensure raw_answer is a string
prompt = f"{self._system_prompt}\nQuestion: {raw_question}\n{choices}\nAnswer:"
answer_letter = chr(65 + int(raw_answer)) # Convert index to 'A', 'B', 'C', 'D'
return MMLUParseEntry.create(
prompt=prompt,
answer=answer_letter,
raw_question=raw_question,
raw_choices=raw_choices,
raw_answer=raw_answer,
task_name=final_task,
)
class BaseMMLUDatasetParser(MMLUDatasetParser):
"""Parser for the original MMLU dataset."""
_data_source = "cais/mmlu"
_default_task = "all"
_task_names = [
"abstract_algebra",
"anatomy",
"astronomy",
"business_ethics",
"clinical_knowledge",
"college_biology",
"college_chemistry",
"college_computer_science",
"college_mathematics",
"college_medicine",
"college_physics",
"computer_security",
"conceptual_physics",
"econometrics",
"electrical_engineering",
"elementary_mathematics",
"formal_logic",
"global_facts",
"high_school_biology",
"high_school_chemistry",
"high_school_computer_science",
"high_school_european_history",
"high_school_geography",
"high_school_government_and_politics",
"high_school_macroeconomics",
"high_school_mathematics",
"high_school_microeconomics",
"high_school_physics",
"high_school_psychology",
"high_school_statistics",
"high_school_us_history",
"high_school_world_history",
"human_aging",
"human_sexuality",
"international_law",
"jurisprudence",
"logical_fallacies",
"machine_learning",
"management",
"marketing",
"medical_genetics",
"miscellaneous",
"moral_disputes",
"moral_scenarios",
"nutrition",
"philosophy",
"prehistory",
"professional_accounting",
"professional_law",
"professional_medicine",
"professional_psychology",
"public_relations",
"security_studies",
"sociology",
"us_foreign_policy",
"virology",
"world_religions",
]
class MMLUReduxDatasetParser(MMLUDatasetParser):
"""Parser for the MMLU Redux dataset."""
_data_source = "edinburgh-dawg/mmlu-redux"
_default_task = "anatomy"
_task_names = [
"anatomy",
"astronomy",
"business_ethics",
"clinical_knowledge",
"college_chemistry",
"college_computer_science",
"college_mathematics",
"college_medicine",
"college_physics",
"conceptual_physics",
"econometrics",
"electrical_engineering",
"formal_logic",
"global_facts",
"high_school_chemistry",
"high_school_geography",
"high_school_macroeconomics",
"high_school_mathematics",
"high_school_physics",
"high_school_statistics",
"high_school_us_history",
"human_aging",
"logical_fallacies",
"machine_learning",
"miscellaneous",
"philosophy",
"professional_accounting",
"professional_law",
"public_relations",
"virology",
]
class TMMLUPlusDatasetParser(MMLUDatasetParser):
"""Parser for the TMMLU+ dataset."""
_data_source = "ikala/tmmluplus"
_default_task = "taiwanese_hokkien"
_task_names = [
"engineering_math",
"dentistry",
"traditional_chinese_medicine_clinical_medicine",
"clinical_psychology",
"technical",
"culinary_skills",
"mechanical",
"logic_reasoning",
"real_estate",
"general_principles_of_law",
"finance_banking",
"anti_money_laundering",
"ttqav2",
"marketing_management",
"business_management",
"organic_chemistry",
"advance_chemistry",
"physics",
"secondary_physics",
"human_behavior",
"national_protection",
"jce_humanities",
"politic_science",
"agriculture",
"official_document_management",
"financial_analysis",
"pharmacy",
"educational_psychology",
"statistics_and_machine_learning",
"management_accounting",
"introduction_to_law",
"computer_science",
"veterinary_pathology",
"accounting",
"fire_science",
"optometry",
"insurance_studies",
"pharmacology",
"taxation",
"trust_practice",
"geography_of_taiwan",
"physical_education",
"auditing",
"administrative_law",
"education_(profession_level)",
"economics",
"veterinary_pharmacology",
"nautical_science",
"occupational_therapy_for_psychological_disorders",
"basic_medical_science",
"macroeconomics",
"trade",
"chinese_language_and_literature",
"tve_design",
"junior_science_exam",
"junior_math_exam",
"junior_chinese_exam",
"junior_social_studies",
"tve_mathematics",
"tve_chinese_language",
"tve_natural_sciences",
"junior_chemistry",
"music",
"education",
"three_principles_of_people",
"taiwanese_hokkien",
]
def process_entry(
self, row: dict[str, Any], task_name: str | None = None, **kwargs: Any
) -> MMLUParseEntry:
"""Process a single TMMLU+ entry."""
# Extract choices in order
raw_choices = [row["A"], row["B"], row["C"], row["D"]]
choices = "\n".join(
f"{chr(65 + i)}. {choice}" for i, choice in enumerate(raw_choices)
)
raw_question = row["question"]
raw_answer = row["answer"]
prompt = f"{self._system_prompt}\nQuestion: {raw_question}\n{choices}\nAnswer:"
task = task_name or self._get_current_task(row)
return MMLUParseEntry.create(
prompt, raw_answer, raw_question, raw_choices, raw_answer, task
)
class MMLUProDatasetParser(HuggingFaceDatasetParser[MMLUProParseEntry]):
"""Parser for the MMLU Pro dataset."""
_data_source = "TIGER-Lab/MMLU-Pro"
_default_task = "default"
_task_names = ["default"]
_hidden_task_names = [
"math",
"physics",
"chemistry",
"law",
"engineering",
"other",
"economics",
"health",
"psychology",
"business",
"biology",
"philosophy",
"computer_science",
"history",
]
_default_system_prompt = MMLU_PRO_SYSTEM_PROMPT
def _get_task_from_entry(self, data_entry: dict[str, Any]) -> str:
"""Get the task name from the data entry or default task name."""
if data_entry is not None:
task_name = data_entry.get("category")
if task_name:
return task_name
return self._current_task or self._default_task
def process_entry(
self, row: dict[str, Any], task_name: str | None = None, **kwargs: Any
) -> MMLUProParseEntry:
"""
Generate a prompt and expected answer from the given row.
Args:
row (dict[str, Any]): A data point to be formatted with MMLU Pro specific structure
containing 'question', 'options', 'answer', and 'answer_index' keys.
Returns:
MMLUParseEntry: The formatted entry object.
"""
task = task_name or self._get_current_task(row)
# Ensure task is not None
final_task = task or self._default_task
# Extract choices in order
raw_choices = row["options"]
choices = "\n".join(
f"{chr(65 + i)}. {choice}" for i, choice in enumerate(raw_choices)
)
raw_question = row["question"]
raw_answer = row["answer"]
answer_index = row["answer_index"]
prompt = f"{self._system_prompt}\nQuestion: {raw_question}\n{choices}\nAnswer:"
answer_letter = chr(
65 + answer_index
) # Convert index to 'A', 'B', 'C', 'D', etc.
return MMLUProParseEntry.create(
prompt, answer_letter, raw_question, raw_choices, raw_answer, final_task
)
if __name__ == "__main__":
# Example usage of MMLU Pro parser
parser = MMLUProDatasetParser()
parser.load()
parser.parse()
# Get parsed data with correct type
parsed_data = parser.get_parsed_data
# Print example entry
if parsed_data:
example = parsed_data[0]
print("\nExample parsed entry:")
print(f"Task: {example.task_name}")
print(f"Question: {example.raw_question}")
print("Choices:")
for i, choice in enumerate(example.raw_choices):
print(f"{chr(65 + i)}. {choice}")
print(f"Correct Answer: {example.answer}")
|