Skip to content

[Fix] CEval ModelScope load and HF generate for causal LMs #2416

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
DeliWang wants to merge 1 commit into
base: main
Choose a base branch
from
Open

[Fix] CEval ModelScope load and HF generate for causal LMs #2416

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
6 changes: 4 additions & 2 deletions opencompass/datasets/ceval.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -20,7 +20,8 @@ def load(path: str, name: str, local_mode: bool = False):
dataset = {}
if environ.get('DATASET_SOURCE') == 'ModelScope':
from modelscope import MsDataset
dataset = MsDataset.load(dataset_name=path, subset_name=name)
dataset = MsDataset.load(
dataset_name=path, subset_name=name, trust_remote_code=True)
else:
for split in ['dev', 'val', 'test']:
filename = osp.join(path, split, f'{name}_{split}.csv')
Expand Down Expand Up @@ -72,7 +73,8 @@ def load(path: str, name: str):
dataset = {}
if environ.get('DATASET_SOURCE') == 'ModelScope':
from modelscope import MsDataset
dataset = MsDataset.load(dataset_name=path, subset_name=name)
dataset = MsDataset.load(
dataset_name=path, subset_name=name, trust_remote_code=True)
# 向该数据添加 'is_clean' 字段
annotations = CEvalDatasetClean.load_contamination_annotations(
path, 'val')
Expand Down
4 changes: 4 additions & 0 deletions opencompass/models/huggingface_above_v4_33.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -478,6 +478,8 @@ def generate(self,
self.logger.info(generation_kwargs)

# step-2: conduct model forward to generate output
# LLaMA and some causal LMs do not accept token_type_ids in generate().
tokens.pop('token_type_ids', None)
outputs = self.model.generate(**tokens, **generation_kwargs)
outputs = outputs[:, tokens['input_ids'].shape[1]:]

Expand Down Expand Up @@ -580,6 +582,8 @@ def generate(self,
generation_kwargs['pad_token_id'] = self.tokenizer.pad_token_id

# step-2: conduct model forward to generate output
# LLaMA and some causal LMs do not accept token_type_ids in generate().
tokens.pop('token_type_ids', None)
outputs = self.model.generate(**tokens, **generation_kwargs)
outputs = outputs[:, tokens['input_ids'].shape[1]:]

Expand Down