Download scripts/eval/sqa.sh from lazybug/VocoUPL: direct link, hf CLI and curl.
- Browser
- Download file 806 Bytes
-
https://huggingface.co/datasets/lazybug/VocoUPL/resolve/main/scripts/eval/sqa.sh
- Command line
-
hf download hf://datasets/lazybug/VocoUPL/scripts/eval/sqa.sh
-
curl -L -o sqa.sh https://huggingface.co/datasets/lazybug/VocoUPL/resolve/main/scripts/eval/sqa.sh
806 Bytes
| CUDA_VISIBLE_DEVICES=7 python -m llava.eval.model_vqa_science \ | |
| --model-path ./checkpoints/voco_llava_ckpt \ | |
| --question-file ./playground/data/eval/scienceqa/llava_test_CQM-A.json \ | |
| --image-folder ./playground/data/eval/scienceqa/images/test \ | |
| --answers-file ./playground/data/eval/scienceqa/answers/voco_llama.jsonl \ | |
| --single-pred-prompt \ | |
| --temperature 0 \ | |
| --voco_num 2 \ | |
| --conv-mode vicuna_v1 | |
| python llava/eval/eval_science_qa.py \ | |
| --base-dir ./playground/data/eval/scienceqa \ | |
| --result-file ./playground/data/eval/scienceqa/answers/voco_llama.jsonl \ | |
| --output-file ./playground/data/eval/scienceqa/answers/voco_llama_output.jsonl \ | |
| --output-result ./playground/data/eval/scienceqa/answers/voco_llama_result.json | |
| # bash scripts/eval/sqa.sh |