[Feature] Add verifier for VQA/MCQ evaluation #424
Workflow file for this run
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
| name: pr_run_test | |
| on: | |
| pull_request: | |
| branches: | |
| - "main" | |
| paths-ignore: | |
| - "docs/**" | |
| - "**.md" | |
| concurrency: | |
| group: ${{ github.workflow }}-${{ github.event.pull_request.number || github.ref }} | |
| cancel-in-progress: true | |
| env: | |
| BASE_SCORE: '{"MMBench_V11_MINI":{"Qwen2-VL-7B-Instruct":0.8727272727272727,"InternVL2_5-8B":0.8727272727272727,"llava_onevision_qwen2_7b_si":0.8363636363636363},"MMStar_MINI":{"Qwen2-VL-7B-Instruct":0.6266666666666667,"InternVL2_5-8B":0.6333333333333333,"llava_onevision_qwen2_7b_si":0.49333333333333335},"AI2D_MINI":{"Qwen2-VL-7B-Instruct":0.7975708502024291,"InternVL2_5-8B":0.854251012145749,"llava_onevision_qwen2_7b_si":0.8178137651821862},"OCRBench_MINI":{"Qwen2-VL-7B-Instruct":16.6,"InternVL2_5-8B":16.4,"llava_onevision_qwen2_7b_si":12.9}}' | |
| jobs: | |
| vlm_test: | |
| if: ${{!cancelled()}} | |
| runs-on: [linux-a100] | |
| strategy: | |
| fail-fast: false | |
| matrix: | |
| model: [Qwen/Qwen2-VL-7B-Instruct,OpenGVLab/InternVL2_5-8B,lmms-lab/llava-onevision-qwen2-7b-si] | |
| dataset: ["MMBench_V11_MINI MMStar_MINI AI2D_MINI","OCRBench_MINI"] | |
| container: | |
| image: kkscilife/vlmevalkit_2:a100 | |
| options: "--gpus=all --ipc=host -e https_proxy=$https_proxy -e http_proxy=$http_proxy --pull never" | |
| volumes: | |
| - /mnt/187:/mnt/187 | |
| steps: | |
| - name: clone_repo | |
| uses: actions/checkout@v3 | |
| - name: evaluation_model | |
| run: | | |
| pip install -e . | |
| pre_model=$(echo ${{matrix.model}} | awk -F'/' '{print $1}') | |
| ln -s /mnt/187/$pre_model . | |
| if [ "${{matrix.model}}" = "lmms-lab/llava-onevision-qwen2-7b-si" ];then | |
| model_name="llava_onevision_qwen2_7b_si" | |
| else | |
| model_name=$(echo ${{matrix.model}} | awk -F'/' '{print $2}') | |
| fi | |
| nvidia-smi | |
| python run.py --data ${{matrix.dataset}} --model $model_name | |
| python .github/scripts/assert_score.py --dataset "${{matrix.dataset}}" --base_score $BASE_SCORE --model-name $model_name |