Video-LLaVA / scripts /v1_5 /eval /eval_image_mmvet.sh
LinB203
m
5244231
raw
history blame
682 Bytes
#!/bin/bash
CKPT_NAME="Video-LLaVA-7B"
CKPT="checkpoints/${CKPT_NAME}"
EVAL="eval"
python3 -m llava.eval.model_vqa \
--model-path ${CKPT} \
--question-file ${EVAL}/mm-vet/llava-mm-vet.jsonl \
--image-folder ${EVAL}/mm-vet/images \
--answers-file ${EVAL}/mm-vet/answers/${CKPT_NAME}.jsonl \
--temperature 0 \
--conv-mode vicuna_v1
mkdir -p ${EVAL}/mm-vet/results
python3 scripts/convert_mmvet_for_eval.py \
--src ${EVAL}/mm-vet/answers/${CKPT_NAME}.jsonl \
--dst ${EVAL}/mm-vet/results/${CKPT_NAME}.json
python3 scripts/eval_gpt_mmvet.py \
--mmvet_path ${EVAL}/mm-vet \
--ckpt_name ${CKPT_NAME} \
--result_path ${EVAL}/mm-vet/results