qianxiao1111's picture
upgrade: add benchmarks eval
2a26d3b
raw
history blame
902 Bytes
export CUDA_VISIBLE_DEVICES=0
python -m gen_baseline_tableinstruct \
--model-path /data1/workspace/lly/liliyao/models/Qwen/Qwen2-7B-Instruct \
--json-path /data4/workspace/lly/table_instruct_eval/data/TableInstruct/eval_data \
--output-path /data4/workspace/lly/table_instruct_eval/eval/all/1-vLLM \
--num-gpus-total 1 \
--num-gpus-per-model 1 \
--dataset-part all_test \
--inference-type vLLM \
--inference-config vLLM_config.json
wait
python -m gen_baseline_tableinstruct \
--model-path /data1/workspace/lly/liliyao/models/Qwen/Qwen2-7B-Instruct \
--json-path /data4/workspace/lly/table_instruct_eval/data/TableInstruct/eval_data \
--output-path /data4/workspace/lly/table_instruct_eval/eval/all/1-TGI \
--num-gpus-total 1 \
--num-gpus-per-model 1 \
--dataset-part all_test \
--inference-type TGI \
--inference-config TGI_config.json