OpenCompass/configs/datasets/compassbench_v1_3/compassbench_v1_3_knowledge.py

95 lines
4.3 KiB
Python
Raw Normal View History

from opencompass.openicl.icl_prompt_template import PromptTemplate
from opencompass.openicl.icl_retriever import ZeroRetriever
from opencompass.openicl.icl_inferencer import GenInferencer
from opencompass.openicl.icl_evaluator import CircularEvaluator, AccEvaluator
from opencompass.datasets.compassbench_obj import (
CompassBenchObjectiveV1_3,
compassbench_objective_v1_3_postprocess,
)
from opencompass.utils.text_postprocessors import first_option_postprocess
prompt_cn = {
'single_choice_cn': '以下是一道单项选择题请你根据你了解的知识给出正确的答案选项。请你一步步推理并在最后用“答案选项为X”来回答其中X是ABCD中你认为正确的选项序号\n下面是你要回答的题目:\n{question}\n让我们一步步解决这个问题:',
'cloze_cn': '以下是一道填空题,请你根据你了解的知识一步步思考后把你的最终答案放到\\boxed{}中。\n下面是你要回答的题目:\n{question}\n让我们一步步解决这个问题:',
}
prompt_en = {
'single_choice_en': "Here is a single-choice question. Please give the correct answer based on your knowledge. Please reason step by step and answer with 'The answer is X' at the end, where X is the option letter you think is correct.\nHere is the question you need to answer:\n{question}\nLet's solve this problem step by step:",
'cloze_en': "Here is a fill-in-the-blank question. Please think step by step based on your knowledge and put your final answer in \\boxed{}. Here is the question you need to answer:\n{question}\nLet's solve this problem step by step:",
}
douknow_sets = {
'wiki_en_sub_500_人文科学':['single_choice_en'],
'wiki_en_sub_500_社会科学':['single_choice_en'],
'wiki_en_sub_500_生活常识':['single_choice_en'],
'wiki_en_sub_500_自然科学-工科':['single_choice_en'],
'wiki_en_sub_500_自然科学-理科':['single_choice_en'],
'wiki_zh_sub_500_人文科学': ['single_choice_cn'],
'wiki_zh_sub_500_社会科学': ['single_choice_cn'],
'wiki_zh_sub_500_生活常识': ['single_choice_cn'],
'wiki_zh_sub_500_自然科学-工科':['single_choice_cn'],
'wiki_zh_sub_500_自然科学-理科':['single_choice_cn'],
}
data_path = './data/compassbench_v1_3/knowledge'
# Set up the prompts
CircularEval = True
compassbench_knowledge_datasets = []
for _split in list(douknow_sets.keys()):
for _name in douknow_sets[_split]:
if 'cn' in _name:
single_choice_prompts = prompt_cn
cloze_prompts = prompt_cn
else:
single_choice_prompts = prompt_en
cloze_prompts = prompt_en
if 'single_choice' in _name:
template_round = [dict(role='HUMAN', prompt=single_choice_prompts[_name])]
pred_postprocessor = dict(type=first_option_postprocess, options='ABCD')
evaluator = dict(type=CircularEvaluator if CircularEval else AccEvaluator)
dataset_name = _name + '_circular' if CircularEval else _name
dataset_abbr = (
'compassbench-' + _split + '_circular'
if CircularEval
else 'compassbench-' + _split
)
else:
template_round = [dict(role='HUMAN', prompt=cloze_prompts[_name])]
pred_postprocessor = dict(
type=compassbench_objective_v1_3_postprocess, name=_name
)
evaluator = dict(type=AccEvaluator)
dataset_name = _name
dataset_abbr = 'compassbench-' + _split
douknow_infer_cfg = dict(
prompt_template=dict(
type=PromptTemplate, template=dict(round=template_round)
),
retriever=dict(type=ZeroRetriever),
inferencer=dict(type=GenInferencer, max_out_len=2048),
)
douknow_eval_cfg = dict(
evaluator=evaluator,
pred_postprocessor=pred_postprocessor,
)
compassbench_knowledge_datasets.append(
dict(
type=CompassBenchObjectiveV1_3,
path=f'{data_path}/{_split}.jsonl',
name=dataset_name,
abbr=dataset_abbr,
reader_cfg=dict(input_columns=['question'], output_column='answer'),
infer_cfg=douknow_infer_cfg,
eval_cfg=douknow_eval_cfg,
)
)
del _split, _name