diff --git a/opencompass/configs/datasets/aime2024/aime2024_cascade_eval_gen_5e9f4f.py b/opencompass/configs/datasets/aime2024/aime2024_cascade_eval_gen_5e9f4f.py index 4542722c..78c81c16 100644 --- a/opencompass/configs/datasets/aime2024/aime2024_cascade_eval_gen_5e9f4f.py +++ b/opencompass/configs/datasets/aime2024/aime2024_cascade_eval_gen_5e9f4f.py @@ -13,6 +13,7 @@ from opencompass.datasets.arc_prize_public_evaluation import pad_array_with_valu from opencompass.openicl.icl_prompt_template import PromptTemplate from opencompass.openicl.icl_retriever import ZeroRetriever from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.datasets import generic_llmjudge_postprocess from opencompass.datasets import Aime2024Dataset from opencompass.evaluator import ( CascadeEvaluator, @@ -20,7 +21,6 @@ from opencompass.evaluator import ( MATHVerifyEvaluator ) -from opencompass.datasets import generic_llmjudge_postprocess aime2024_reader_cfg = dict(input_columns=['question'], output_column='answer') @@ -94,13 +94,13 @@ cascade_evaluator = dict( type=Aime2024Dataset, path='opencompass/aime2024', reader_cfg=aime2024_reader_cfg, - n=2, + n=32, ), judge_cfg=dict(), dict_postprocessor=dict(type=generic_llmjudge_postprocess), ) ), - # parallel=False, + parallel=False, ) @@ -116,6 +116,6 @@ aime2024_datasets = [ reader_cfg=aime2024_reader_cfg, infer_cfg=aime2024_infer_cfg, eval_cfg=aime2024_eval_cfg, - n=2,# Evaluate the dataset with 2 times + n=32,# Evaluate the dataset with 2 times ) ] diff --git a/opencompass/configs/datasets/math/math_500_cascade_eval_gen_6ff468.py b/opencompass/configs/datasets/math/math_500_cascade_eval_gen_6ff468.py new file mode 100644 index 00000000..e2fcf167 --- /dev/null +++ b/opencompass/configs/datasets/math/math_500_cascade_eval_gen_6ff468.py @@ -0,0 +1,117 @@ +""" +Summary: A config for AIME-2024 Evaluation. +Setting: + Shot: 0-shot + Evaluator: + - CascadeEvaluator + - MATHVerifyEvaluator + - GenericLLMEvaluator +Avaliable Models: + - Instruct/Chat Models +""" + +from opencompass.openicl.icl_prompt_template import PromptTemplate +from opencompass.openicl.icl_retriever import ZeroRetriever +from opencompass.openicl.icl_inferencer import GenInferencer +from opencompass.datasets import generic_llmjudge_postprocess +from opencompass.datasets import MATHDataset +from opencompass.evaluator import ( + CascadeEvaluator, + GenericLLMEvaluator, + MATHVerifyEvaluator +) + +# ----------------------------- Detailed Config ----------------------------- + +math_reader_cfg = dict(input_columns=['problem'], output_column='solution') +math_infer_cfg = dict( + prompt_template=dict( + type=PromptTemplate, + template=dict( + round=[ + dict(role='HUMAN', prompt='{problem}\nRemember to put your final answer within \\boxed{}.'), + ] + ), + ), + retriever=dict(type=ZeroRetriever), + inferencer=dict(type=GenInferencer), +) + + +GRADER_TEMPLATE = """ + Please as a grading expert, judge whether the final answers given by the candidates below are consistent with the standard answers, that is, whether the candidates answered correctly. + + Here are some evaluation criteria: + 1. Please refer to the given standard answer. You don't need to re-generate the answer to the question because the standard answer has been given. You only need to judge whether the candidate's answer is consistent with the standard answer according to the form of the question. Don't try to answer the original question. You can assume that the standard answer is definitely correct. + 2. Because the candidate's answer may be different from the standard answer in the form of expression, before making a judgment, please understand the question and the standard answer first, and then judge whether the candidate's answer is correct, but be careful not to try to answer the original question. + 3. Some answers may contain multiple items, such as multiple-choice questions, multiple-select questions, fill-in-the-blank questions, etc. As long as the answer is the same as the standard answer, it is enough. For multiple-select questions and multiple-blank fill-in-the-blank questions, the candidate needs to answer all the corresponding options or blanks correctly to be considered correct. + 4. Some answers may be expressed in different ways, such as some answers may be a mathematical expression, some answers may be a textual description, as long as the meaning expressed is the same. And some formulas are expressed in different ways, but they are equivalent and correct. + 5. If the prediction is given with \\boxed{}, please ignore the \\boxed{} and only judge whether the candidate's answer is consistent with the standard answer. + + Please judge whether the following answers are consistent with the standard answer based on the above criteria. Grade the predicted answer of this new question as one of: + A: CORRECT + B: INCORRECT + Just return the letters "A" or "B", with no text around it. + + Here is your task. Simply reply with either CORRECT, INCORRECT. Don't apologize or correct yourself if there was a mistake; we are just trying to grade the answer. + + + : \n{problem}\n\n\n + : \n{solution}\n\n\n + : \n{prediction}\n\n\n + + Judging the correctness of candidates' answers: +""".strip() + + +cascade_evaluator = dict( + type=CascadeEvaluator, + rule_evaluator=dict( + type=MATHVerifyEvaluator, + ), + llm_evaluator= dict( + dict( + type=GenericLLMEvaluator, + prompt_template=dict( + type=PromptTemplate, + template=dict( + begin=[ + dict( + role='SYSTEM', + fallback_role='HUMAN', + prompt="You are a helpful assistant who evaluates the correctness and quality of models' outputs.", + ) + ], + round=[ + dict(role='HUMAN', prompt=GRADER_TEMPLATE), + ], + ), + ), + dataset_cfg=dict( + type=MATHDataset, + path='opencompass/math', + file_name = 'test_prm800k_500.json', + reader_cfg=math_reader_cfg, + n=4, + ), + judge_cfg=dict(), + dict_postprocessor=dict(type=generic_llmjudge_postprocess), + ) + ), + parallel=False, +) + +math_datasets = [ + dict( + type=MATHDataset, + abbr=f'math_prm800k_500', + path='opencompass/math', + file_name = 'test_prm800k_500.json', + reader_cfg=math_reader_cfg, + infer_cfg=math_infer_cfg, + eval_cfg=dict( + evaluator=cascade_evaluator, + ), + n=4, + ) +] diff --git a/opencompass/evaluator/cascade_evaluator.py b/opencompass/evaluator/cascade_evaluator.py index 9227c505..fc68e24f 100644 --- a/opencompass/evaluator/cascade_evaluator.py +++ b/opencompass/evaluator/cascade_evaluator.py @@ -182,7 +182,7 @@ class CascadeEvaluator(BaseEvaluator): self.llm_evaluator._out_dir = f'{self._out_dir}_llm_judge' # Generate random hash suffix - llm_results_path = f'{self.llm_evaluator._out_dir}_{self.dataset_replica_idx}' # noqa + llm_results_path = f'{self.llm_evaluator._out_dir}_replica{self.dataset_replica_idx}' # noqa # Check if results already exist to avoid re-evaluation if os.path.exists(llm_results_path): diff --git a/opencompass/evaluator/generic_llm_evaluator.py b/opencompass/evaluator/generic_llm_evaluator.py index 7a0fd538..65db2061 100644 --- a/opencompass/evaluator/generic_llm_evaluator.py +++ b/opencompass/evaluator/generic_llm_evaluator.py @@ -58,8 +58,6 @@ class GenericLLMEvaluator(BaseEvaluator): def build_inferencer(self): """Build LLM Inference.""" if not self.output_path: - # output_path = self._out_dir - # self.output_path = f'{output_path}.json' self.output_path = self._out_dir out_dir, out_name = osp.split(self.output_path)