History

aJupyter c696e163cd feat: Update Aiwei configuration.		2024-02-23 20:11:17 +08:00
..
images	refactor: modified xtuner_config and removed model_config	2024-01-26 21:37:15 +08:00
aiwei-internlm2_chat_7b_qlora.py	feat: Update Aiwei configuration.	2024-02-23 20:11:17 +08:00
baichuan2_13b_chat_qlora_alpaca_e3.py	feat：Add new finetune configurations and datasets	2024-02-23 11:36:58 +08:00
chatglm3_6b_lora_alpaca_e3.py	feat：Add new finetune configurations and datasets	2024-02-23 11:36:58 +08:00
ChatGLM3-6b-ft.md	docs：add finetune doc and update readme	2024-01-26 22:25:42 +08:00
deepseek_moe_16b_chat_qlora_oasst1_e3.py	feat：Add new finetune configurations and datasets	2024-02-23 11:36:58 +08:00
internlm2_1_8b_full_alpaca_e3.py	feat：Add new finetune configurations and datasets	2024-02-23 11:36:58 +08:00
internlm2_7b_chat_qlora_e3.py	feat: add web_internlm2 and upload s.t. scripts	2024-01-25 19:02:24 +08:00
mixtral_8x7b_instruct_qlora_oasst1_e3.py	feat：Add new finetune configurations and datasets	2024-02-23 11:36:58 +08:00
qwen1_5_0_5_B_full.py	feat：Add new finetune configurations and datasets	2024-02-23 11:36:58 +08:00
qwen_7b_chat_qlora_e3.py	feat: add web_internlm2 and upload s.t. scripts	2024-01-25 19:02:24 +08:00
README.md	docs：add finetune doc and update readme	2024-01-26 22:25:42 +08:00
requirements.txt	docs：add finetune doc and update readme	2024-01-26 22:25:42 +08:00

README.md

微调指南

本项目不仅在心理健康数据集上进行了微调，同时也对模型进行了自我认知微调，下面是微调的详细指南。

一、基于xtuner的微调🎉🎉🎉🎉🎉

环境准备

datasets==2.16.1
deepspeed==0.13.1
einops==0.7.0
flash_attn==2.5.0
mmengine==0.10.2
openxlab==0.0.34
peft==0.7.1
sentencepiece==0.1.99
torch==2.1.2
transformers==4.36.2
xtuner==0.1.11

也可以一键安装

cd xtuner_config/
pip3 install -r requirements.txt

微调

cd xtuner_config/
xtuner train internlm2_7b_chat_qlora_e3.py --deepspeed deepspeed_zero2

将得到的 PTH 模型转换为 HuggingFace 模型

即：生成 Adapter 文件夹

cd xtuner_config/
mkdir hf
export MKL_SERVICE_FORCE_INTEL=1

xtuner convert pth_to_hf internlm2_7b_chat_qlora_e3.py ./work_dirs/internlm_chat_7b_qlora_oasst1_e3_copy/epoch_3.pth ./hf

将 HuggingFace adapter 合并到大语言模型

xtuner convert merge ./internlm2-chat-7b ./hf ./merged --max-shard-size 2GB
# xtuner convert merge \
#     ${NAME_OR_PATH_TO_LLM} \
#     ${NAME_OR_PATH_TO_ADAPTER} \
#     ${SAVE_PATH} \
#     --max-shard-size 2GB

测试

cd demo/
python cli_internlm2.py

二、基于Transformers的微调🎉🎉🎉🎉🎉

请查看ChatGLM3-6b lora微调指南

其他

欢迎大家给xtuner和EmoLLM点点star~

🎉🎉🎉🎉🎉