Skip to content

[Feature] expand mcq eval judge model support and fix single-image PNG saving #1431

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
PeterWangyi wants to merge 2 commits into
base: main
Choose a base branch
from
+20 −3
Open

[Feature] expand mcq eval judge model support and fix single-image PNG saving #1431

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
296c28e
[Feature] expand mcq eval judge model support, tag acc filenames, and…
PeterWangyi Dec 22, 2025
b9790eb
[Fix] update HF cache env list
PeterWangyi Feb 4, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion vlmeval/dataset/image_base.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -141,7 +141,7 @@ def dump_image(self, line):
decode_base64_to_image_file(line['image'], tgt_path)
tgt_path = [tgt_path]
else:
tgt_path = osp.join(self.img_root, f"{line['index']}.jpg")
tgt_path = osp.join(self.img_root, f"{line['index']}.png")
if not read_ok(tgt_path):
decode_base64_to_image_file(line['image'], tgt_path)
tgt_path = [tgt_path]
Expand Down
19 changes: 18 additions & 1 deletion vlmeval/dataset/image_mcq.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -261,7 +261,24 @@ def evaluate_heuristic(self, eval_file, **judge_kwargs):
circular = True

model = judge_kwargs.get('model', 'exact_matching')
name_str_map = {'chatgpt-0125': 'openai', 'gpt-4-0125': 'gpt4'}
# For the full list of supported judge backends / model aliases, see:
# vlmeval/dataset/utils/judge_util.py (build_judge: model_map)
name_str_map = {
'gpt-4-turbo': 'gpt-4-turbo',
'gpt-4-0613': 'gpt-4-0613',
'gpt-4-0125': 'gpt-4-0125',
'gpt-4-0409': 'gpt-4-0409',
'chatgpt-1106': 'chatgpt-1106',
'chatgpt-0125': 'chatgpt-0125',
'gpt-4o': 'gpt-4o',
'gpt-4o-0806': 'gpt-4o-0806',
'gpt-4o-1120': 'gpt-4o-1120',
'gpt-4o-mini': 'gpt-4o-mini',
}
if model != 'exact_matching' and model not in name_str_map:
raise ValueError(
f'Unsupported judge model: {model}. Allowed: {list(name_str_map)} + exact_matching'
)
name_str = name_str_map[model] if model in name_str_map else model

if model == 'exact_matching':
Expand Down
2 changes: 1 addition & 1 deletion vlmeval/smp/file.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -76,7 +76,7 @@ def LMUDataRoot():


def HFCacheRoot():
cache_list = ['HUGGINGFACE_HUB_CACHE', 'HF_HOME']
cache_list = ['HF_HUB_CACHE', 'HUGGINGFACE_HUB_CACHE', 'HF_HOME']
for cache_name in cache_list:
if cache_name in os.environ and osp.exists(os.environ[cache_name]):
if os.environ[cache_name].split('/')[-1] == 'hub':
Expand Down