python main.py --model_name "deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B" --train_dataset_name "modelscope.gsm8k" --score_weights 1.0 1.0 1.0 --output_data_file1 "dara/origin.txt" --output_data_file2 "data/augmented.txt" --save_model_path1 "/models/fine-tuned-model1" --save_model_path2 "/models/fine-tuned-model2"  --gpu "0" --batch_size 16 --epochs 4 --learning_rate 1e-5 --use_lora 0 