Open
Description
Dear Author,
First of all, thank you for your work and for proposing an interesting approach to address the model hallucination issue.
However, after running the code in practice, I encountered some problems. The model's performance does not align with the results presented in the paper, and it even performs worse than the standard approach without using the strategy (the experiment used the LLaVA 1.5-7B model, with an F1 score of 81.5 for SID, while the standard LLaVA achieved an F1 score of 81.6).
I noticed that others have similar issues, so I would like to ask what might be the cause of this and whether there are any precautions to take when running the code you provided.
Looking forward to your reply.
Metadata
Metadata
Assignees
Labels
No labels