Skip to content
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.

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

VHELM update #2592

Merged
merged 133 commits into from
May 3, 2024
Merged

VHELM update #2592

merged 133 commits into from
May 3, 2024

Conversation

teetone
Copy link
Member

@teetone teetone commented Apr 26, 2024

Changes:

  • New scenarios
  • IDEFICS 2 support

@teetone teetone requested a review from yifanmai April 26, 2024 17:54
setup.cfg Outdated Show resolved Hide resolved
src/helm/benchmark/metrics/common_metric_specs.py Outdated Show resolved Hide resolved
src/helm/benchmark/metrics/evaluate_reference_metrics.py Outdated Show resolved Hide resolved
src/helm/benchmark/run_specs/vlm_run_specs.py Outdated Show resolved Hide resolved
src/helm/clients/openai_client.py Outdated Show resolved Hide resolved
src/helm/clients/vertexai_client.py Outdated Show resolved Hide resolved
src/helm/clients/vertexai_client.py Show resolved Hide resolved
src/helm/clients/openai_client.py Outdated Show resolved Hide resolved
@teetone teetone requested a review from yifanmai April 28, 2024 18:58

- name: huggingface/llava-v1.6-vicuna-7b-hf
model_name: uw-madison/llava-v1.6-vicuna-7b-hf
tokenizer_name: hf-internal-testing/llama-tokenizer
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Use the respective tokenizers (vicuna etc)?

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think vicuna uses the llama tokenizer if I'm not mistaken.


return {"predictions": [{"text": raw_response.candidates[0].text}]}
if not candidates:
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We should make this condition tighter... I'll send you a follow-up pull request to fix it.


_models_lock: Lock = Lock()
_models: Dict[str, Optional[Vision2SeqModelProcessor]] = {
"HuggingFaceM4/idefics2-8b": None,
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Any reason we need to explicitly declare models here? Most clients do not validate which model names are supported (because the list can change frequently).

Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

It was for name validation. Do you think I can just remove this line then?

@teetone teetone merged commit 6d3ad7f into main May 3, 2024
6 checks passed
@teetone teetone deleted the vh branch May 3, 2024 08:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants