#!/bin/bash CUDA_VISIBLE_DEVICES=3 python -m llava.eval.model_vqa_loader \ --model-path voco_llama_ckpt \ --question-file ./playground/data/eval/pope/llava_pope_test.jsonl \ --image-folder /group/40033/public_datasets/coco/val2014 \ --answers-file ./playground/data/eval/pope/answers/voco_llama.jsonl \ --voco_num 2 \ --temperature 0 \ --conv-mode vicuna_v1 python llava/eval/eval_pope.py \ --annotation-dir ./playground/data/eval/pope/coco \ --question-file ./playground/data/eval/pope/llava_pope_test.jsonl \ --result-file ./playground/data/eval/pope/answers/voco_llama.jsonl