llama3.1-8b-instruct 中文DPO版

模型介绍

像原版instruct一样,喜欢用有趣中文和表情符号回答问题。
Github:https://github.com/CrazyBoyM/llama3-Chinese-chat

特点:偏好中文和emoji表情,且不损伤原instruct版模型能力。实测中文DPO版问答性能体验超过现在市面上任何llama3.1中文微调版 (微调会大面积破坏llama3.1原版能力,导致遗忘)
Alt text

训练细节

DPO(beta 0.5) + lora rank128, alpha256 + 打开"lm_head", "input_layernorm", "post_attention_layernorm", "norm"层训练.

模型部署

网页脚本文件:https://github.com/CrazyBoyM/llama3-Chinese-chat/blob/main/deploy/web_streamlit_for_instruct_v2.py
已经在模型文件夹里内置了一份web.py,可以直接使用。

pip install streamlit
pip install transformers==4.40.1
streamlit run web.py ./llama3.1-8b-instruct-dpo-zh

模型下载

SDK下载

#安装ModelScope
pip install modelscope
#SDK模型下载
from modelscope import snapshot_download
model_dir = snapshot_download('shareAI/llama3.1-8b-instruct-dpo-zh')

Git下载

#Git模型下载
git clone https://www.modelscope.cn/shareAI/llama3.1-8b-instruct-dpo-zh.git

目前已经上传 Q4 K_M 的GGUF量化模型,后续将进一步进行手机端或边缘算力端展示

Downloads last month
1,197
GGUF
Model size
8.03B params
Architecture
llama

4-bit

Inference Examples
Unable to determine this model's library. Check the docs .

Dataset used to train shareAI/llama3.1-8b-instruct-dpo-zh