Skip to content

Qwen2-VL-2B result for TextVQA differs from paper #720

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

Closed
soonchangAI opened this issue Jan 12, 2025 · 1 comment
Closed

Qwen2-VL-2B result for TextVQA differs from paper #720

soonchangAI opened this issue Jan 12, 2025 · 1 comment
Assignees

Comments

@soonchangAI
Copy link

Hi, The TextVQA accuracy recorded on paper is 79.7% for Qwen2-VL-2B ( Qwen/Qwen2-VL-2B-Instruct).
However, I obtained TextVQA result of 77.576% (-2.124%)

"Overall"
"77.576"

I have followed installation instruction from #492 (comment)

@PhoenixZ810
Copy link
Collaborator

PhoenixZ810 commented Jan 16, 2025

Hi,

Different environments, such as variations in the versions of CUDA, PyTorch, Flash-Attention, and Transformers, can indeed lead to fluctuations in the results.
Regarding the scores of 79.7 and 77.57, the difference between these two results is relatively minor and can be attributed to such environmental variations. These fluctuations are within an expected range and do not indicate a significant discrepancy.

We appreciate your understanding and are available for any further discussions or clarifications you might need.

@PhoenixZ810 PhoenixZ810 self-assigned this Jan 16, 2025
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

No branches or pull requests

2 participants