|
export CUDA_VISIBLE_DEVICES=0 |
|
|
|
MODEL_PATHS=("/data4/sft_output/qwen2.5-7b-ins-1012/checkpoint-3200") |
|
|
|
for MODEL_PATH in "${MODEL_PATHS[@]}" |
|
do |
|
echo "Running scripts for model at ${MODEL_PATH}" |
|
|
|
|
|
python table_related_benchmarks/run_table_instruct_eval.py --model-path ${MODEL_PATH} |
|
wait |
|
|
|
|
|
python table_related_benchmarks/run_table_bench_eval.py --model_path ${MODEL_PATH} |
|
wait |
|
|
|
|
|
python table_related_benchmarks/run_text2sql_eval.py --model_path ${MODEL_PATH} --is_use_knowledge |
|
wait |
|
|
|
|
|
python general_benchmarks/MBPP/eval_instruct_vllm.py --model_path ${MODEL_PATH} |
|
wait |
|
|
|
|
|
python general_benchmarks/HumanEval/eval_instruct_vllm.py --model_path ${MODEL_PATH} |
|
wait |
|
|
|
|
|
python general_benchmarks/MMLU/evaluator.py --task cmmlu --lang zh --model_path ${MODEL_PATH} |
|
|
|
|
|
python general_benchmarks/MMLU/evaluator.py --task mmlu --lang en --model_path ${MODEL_PATH} |
|
done |
|
|
|
echo "All models processed." |
|
|