The official repository of the paper "The Instinctive Bias: Spurious Images lead to Hallucination in MLLMs"

pip install -r requirements.txt

python eval_llava.py --model_path llava-hf/llava-1.5-7b-hf

We implement an toy example for LLaVA model. You can modify the corresponding part in the code for using different MLLMs.

python output.py

The average accuracy would be printed in the terminal.