数据集文件结构如下,在huggingface上上传了两个dataset,分别是train和test的json文件 CMtMedQA |__ CMtMedQA.json |__ README.md CMtMedQA_test_v1 |__ CMtMedQA_test.json |__ README.md 作者及机构 Songhua Yang(郑州大学)Hanjie Zhao(郑州大学)Senbin Zhu (郑州大学)Guangyu Zhou (郑州大学)Hongfei ...
help='Type of model to use from HuggingFace') parser.add_argument('--use_kge', default=False, help='If KGEs should be place in input', type=str2bool) parser.add_argument('--use_dict', default=False, help='If KGEs should be place in input', type=str2bool) parser.add_argument(...
https://huggingface.co/datasets/shibing624/medical 使用 2. 微调 2.1 LoRA CUDA_VISIBLE_DEVICES=0pythonMedQA-ChatGLM/finetune.py\--do_train\--datasetmerged-cMedQA\--finetuning_typelora\--output_dir./med-lora\--per_device_train_batch_size32\--gradient_accumulation_steps256\--lr_scheduler_t...