README_ZH.md

[**中文**](./README_ZH.md) | [**English**](./README.md)

<p align="center" width="100%">
<a href="https://github.com/ZJUFanLab/TCMChat" target="_blank"><img src="./logo.png" alt="TCMChat" style="width: 25%; min-width: 300px; display: block; margin: auto;"></a>
</p>

# TCMChat: Traditional Chinese Medicine Recommendation System based on Large Language Model

[![Code License](https://img.shields.io/badge/Code%20License-Apache_2.0-green.svg)](https://github.com/SCIR-HI/Huatuo-Llama-Med-Chinese/blob/main/LICENSE) [![Python 3.10.12](https://img.shields.io/badge/python-3.10.12-blue.svg)](https://www.python.org/downloads/release/python-390/)

## 新闻
[2024-11-1] 我们在Huggingface上完全开源了模型权重和训练数据集                 
[2024-5-17] huggingface 开源模型权重


## 应用

### 安装
```shell
git clone https://github.com/ZJUFanLab/TCMChat
cd TCMChat
```

创建conda 环境
```shell
conda create -n baichuan2 python=3.10 -y
```

首先安装依赖包，python环境建议3.10+
``` shell
pip install -r requirements.txt
```

### 权重下载
- [TCMChat](https://huggingface.co/daiyizheng/TCMChat): 基于baichuan2-7B-Chat的中药、方剂知识问答与推荐。

### 推理
#### 命令行测试

```shell
python cli_infer.py \
--model_name_or_path /your/model/path \
--model_type  chat
```

#### Web页面测试

```shell
python gradio_demo.py
```
我们提供了一个在线的体验工具：[https://xomics.com.cn/tcmchat](https://xomics.com.cn/tcmchat)


### 重新训练
#### 数据集下载

- [预训练数据](https://huggingface.co/datasets/ZJUFanLab/TCMChat-dataset-600k) 
- [微调数据](https://huggingface.co/datasets/ZJUFanLab/TCMChat-dataset-600k)
- [基准评测数据](https://github.com/ZJUFanLab/TCMChat/tree/master/evaluation/resources)


> 注意： 在执行预训练、微调和推理之前，请修改自己模型、数据等相关数据路径
#### 预训练

```shell
## slurm 集群
sbatch scripts/pretrain/baichuan2_7b_chat.slurm
##或者
bash scripts/pretrain/baichuan2_7b_chat.sh
```

#### 微调
```shell
## slurm 集群
sbatch scripts/sft/baichuan2_7b_chat.slurm
##或者
bash scripts/sft/baichuan2_7b_chat.sh
```
### 训练细节

请参考论文实验部分说明。

### 基准评估
#### 选择题
```shell
python evaluation/choices_evaluate/eval.py   --model_path_or_name /your/model/path --model_name  baichuan2-7b-chat --few_shot -sz herb --dev_file_path evaluation/resources/choice/single/tcm-herb_dev.csv --val_file_path evaluation/resources/choice/single/choice_herb_500.csv --log_dir logs/choices
```

#### 阅读理解
```shell
python infers/baichuan_infer.py \
--model_name_or_path /your/model/path / \
--model_type chat \
--save_path /your/save/data/path \
--data_path /your/data/path
##BertScore
python evaluation/question_rouge_bleu.py/question_bert_score.py
## BLEU METEOR
python evaluation/question_rouge_bleu.py/open_question_bleu.py
## ROUGE-x
python evaluation/question_rouge_bleu.py/open_question_rouge.py

```
#### 实体抽取
```shell
python infers/baichuan_infer.py \
--model_name_or_path /your/model/path / \
--model_type chat \
--save_path /your/save/data/path \
--data_path /your/data/path

python evaluation/ner_evaluate/tcm_entity_recognition.py

```
#### 医案诊断
```shell
python infers/baichuan_infer.py \
--model_name_or_path /your/model/path / \
--model_type chat \
--save_path /your/save/data/path \
--data_path /your/data/path

python evaluation/acc_evaluate/extract_syndrome.py

```
#### 中药或方剂推荐
```shell
python infers/baichuan_infer.py \
--model_name_or_path /your/model/path / \
--model_type chat \
--save_path /your/save/data/path \
--data_path /your/data/path

python evaluation/recommend_evaluate/mrr_ndcg_p_r.py

```
#### ADMET预测
##### 回归任务
```shell
python infers/baichuan_infer.py \
--model_name_or_path /your/model/path / \
--model_type chat \
--save_path /your/save/data/path \
--data_path /your/data/path

python evaluation/admet_evaluate/rmse_mae_mse.py

```
##### 分类任务
```shell
python infers/baichuan_infer.py \
--model_name_or_path /your/model/path / \
--model_type chat \
--save_path /your/save/data/path \
--data_path /your/data/path

python evaluation/admet_evaluate/acc_recall_f1.py

```