evalscope/examples/tasks/eval_vlm_local.yaml

18 lines
564 B
YAML

eval_backend: VLMEvalKit
eval_config:
model:
- name: qwen_chat # vlmeval model name, refer to https://github.com/open-compass/VLMEvalKit/blob/main/vlmeval/config.py
model_path: ../models/Qwen-VL-Chat # /path/to/model_dir
data:
- SEEDBench_IMG
- ChartQA_TEST
mode: all
limit: 20
reuse: true
work_dir: outputs
nproc: 16
# judge model server config
OPENAI_API_KEY: EMPTY
OPENAI_API_BASE: http://localhost:8866/v1/chat/completions # judge model api
LOCAL_LLM: qwen2-7b-instruct # judge model id