1.Topic：Medical Text Classification

contest and data link：https://tianchi.aliyun.com/competition/entrance/532153?spm=a2c22.12281957.0.0.4c885d9bzbNNxD

2.Paper Title：《A Model Ensemble Approach With LLM for Chinese Text Classification》

Link：https://doi.org/10.1007/978-981-97-1717-0_20（Chinese text classification）
      https://doi.org/10.1007/978-981-97-1717-0_12

3.Architecture

4.Model Selection

Qwen-7b-Chat，ChatGLM2-6b，Macbert

5.Train methods

Qlora ，lora，FGM adversarial train

6.data format

{ "id": "id_0", "conversations": [ { "from": "user", "value": "糖尿病人可不可以服用伟哥? 请对上述的句子划分类别，从下面提供的类别中选出一个正确的选项：A.诊断 B.治疗 C.医疗常识 D.健康生活方式 E.流行病学 F.其他" }, { "from": "assistant", "value": "B.治疗" } ] }

7.label transfer

8.Run

1.run data_process.py

2.run sh lora.sh or train.py

3.run batch.py

4.run data_postprocess.py

9.rank（top2）

10.Citation

{email：[email protected]}