jiangzhs/OpenCompass
1
0
Fork 0
mirror of https://github.com/open-compass/opencompass.git synced 2025-05-30 16:03:24 +08:00
Code Issues Actions 2 Packages Projects Releases Wiki Activity

update

Browse Source
This commit is contained in:
zhulinJulia24 2025-04-09 10:42:47 +08:00
parent b3efc3f2df
commit 019c61e9f5
1 changed files with 2 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
opencompass/datasets/subjective/commonbench.py
View File

@ -8,7 +8,7 @@ from opencompass.registry import DICT_POSTPROCESSORS
from .utils import get_judgeanswer_and_reference from .utils import get_judgeanswer_and_reference
def post_process_mtbench(judgement: str): def post_process(judgement: str):
"""Input a string like below: """Input a string like below:
xxx[[5]]xxx, and extract the score xxx[[5]]xxx, and extract the score
@ -38,8 +38,6 @@ def get_capability_results(judged_answers, references):
s = total_score / capability_counts[capability] s = total_score / capability_counts[capability]
s = round(s, 2) s = round(s, 2)
capability_avg_ratings[capability] = s capability_avg_ratings[capability] = s
columns = list(capability_avg_ratings.keys())
columns.insert(0, columns.pop(columns.index('total')))
return capability_avg_ratings return capability_avg_ratings
@ -48,7 +46,7 @@ def get_capability_results(judged_answers, references):
def commonbench_postprocess( def commonbench_postprocess(
output: dict, output: dict,
output_path: str, output_path: str,
post_process: Optional[callable] = post_process_mtbench, post_process: Optional[callable] = post_process,
) -> dict: ) -> dict:
judged_answers, references = get_judgeanswer_and_reference( judged_answers, references = get_judgeanswer_and_reference(
output, output_path, post_process) output, output_path, post_process)