open-compass / open-compass/VLMEvalKit
LLaVA v1.5 7b very low performance on TextVQA
Nobody has claimed this yet.
- Dominant language
- Python
- Stars
- 4.4k
- Forks
- 768
- Avg merge
- 1d 10h
- Merged PRs (30d)
- 17
Description
Dear authors,
Thanks for sharing this great codebase.
I tried to test LLaVA-v1.5-7b (llava_v1.5_7b) model on TextVQA_VAL and only get about 21.88 accuracy, which is much lower than the 58.2 performance reported in the paper. I understand that the codebase is not aimed to reproduce the exact results but it seems that the gap is too big.
Is there any reason why this happens and any quick fix?
Thanks very much!
Contributor guide
No contributing guide indexed for this repository
First steps
- Read the whole issue, then the project's contributing guide.
- Comment on the issue to say you are picking it up — it saves two people doing the same work.
- Fork the repository and make your change on a branch.
- Open a pull request that references the issue number.
Research direction
Start by reproducing the reported 21.88 result with the llava_v1.5_7b model on TextVQA_VAL, then compare the evaluation setup with the 58.2 result cited in the paper. Done means identifying and documenting the cause of the discrepancy, or providing a verified fix and matching evaluation result.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, pytorch
- Domain
- computer-vision, machine-learning, testing-qa
- Issue type
- Bug
- Difficulty
- 4/5
- Estimated time
- 3-5 days
- Activity status
- Stale
- Clarity
- Needs clarification
- Newbie friendliness
- 25/100