Spaces:
Running
Running
| # Copyright 2024 Bytedance Ltd. and/or its affiliates | |
| # | |
| # Licensed under the Apache License, Version 2.0 (the "License"); | |
| # you may not use this file except in compliance with the License. | |
| # You may obtain a copy of the License at | |
| # | |
| # http://www.apache.org/licenses/LICENSE-2.0 | |
| # | |
| # Unless required by applicable law or agreed to in writing, software | |
| # distributed under the License is distributed on an "AS IS" BASIS, | |
| # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. | |
| # See the License for the specific language governing permissions and | |
| # limitations under the License. | |
| import re | |
| from typing import Dict, List | |
| from mathruler.grader import extract_boxed_content, grade_answer | |
| def format_reward(predict: str) -> float: | |
| pattern = re.compile(r"<think>.*</think>.*\\boxed\{.*\}.*", re.DOTALL) | |
| format_match = re.fullmatch(pattern, predict) | |
| return 1.0 if format_match else 0.0 | |
| def accuracy_reward(predict: str, ground_truth: str) -> float: | |
| answer = extract_boxed_content(predict) | |
| try: | |
| return 1.0 if grade_answer(answer, ground_truth) else 0.0 | |
| except: | |
| return 0.0 | |
| def compute_score(predicts: List[str], ground_truths: List[str], format_weight: float = 0.1) -> List[Dict[str, float]]: | |
| scores = [] | |
| for predict, ground_truth in zip(predicts, ground_truths): | |
| predict = re.sub(r"\s*(<|>|/)\s*", r"\1", predict) # handle qwen2.5vl-32b format | |
| format_score = format_reward(predict) | |
| accuracy_score = accuracy_reward(predict, ground_truth) | |
| scores.append( | |
| { | |
| "overall": (1 - format_weight) * accuracy_score + format_weight * format_score, | |
| "format": format_score, | |
| "accuracy": accuracy_score, | |
| } | |
| ) | |
| return scores | |