This directory contains scripts for converting medical data to GLM-4 fine-tuning format.

• data_conversion_script.py - Main script to convert medical records to GLM-4 format
• prompt.py - Contains system and user prompts for medication prediction
• 候选药物列表.json - List of candidate drugs for medication prediction

cd chip/
python3 data_conversion_script.py

This will convert the medical data from:

• ../data/CDrugRed-A-v1/CDrugRed_train.jsonl → ../data/CDrugRed-A-v1/train.json
• ../data/CDrugRed-A-v1/CDrugRed_test-A.jsonl → ../data/CDrugRed-A-v1/test.json

cd ../finetune
python finetune.py ../data/CDrugRed-A-v1 THUDM/GLM-4-9B-0414 configs/medication_lora.yaml

You could modify the training parameters in configs/medication_lora.yaml