You can configure the path of the dataset through parameters --root_path, --ass_path, and --pse_path
You can configure the path of the LLM through parameter--pretrained_llm_path, the LLM's hidden size through parameter --llm_hidden_size, and the LLM's architecture type through parameter --llm_architecture_type
python main.py \
--do_train \
--eval_batch_size 2 \
--batch_size 2 \
--gradient_accumulation_steps 8 \
--epoch 8
--pretrained_llm_path "the LLM's path you saved" \
--llm_architecture_type "the LLM's architecture type" \
python main.py \
--do_test \
--eval_batch_size 2 \
--pretrained_llm_path "the LLM's architecture type" \
--llm_architecture_type qwen3 \