From 3a6428340ffea25bae9c276af160547ca14af300 Mon Sep 17 00:00:00 2001 From: HongCheng Date: Fri, 10 May 2024 04:04:25 +0900 Subject: [PATCH] Update installation and dpo in README.md --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 9ec39ce..1cf4588 100644 --- a/README.md +++ b/README.md @@ -211,7 +211,7 @@ - 使用QLoRA训练Baichuan2时，需要安装torch==2.0，并且卸载xformers和apex。 - 使用QLoRA训练Qwen时，需将flash-attn卸载，否则会报错。 ```bash -pip install requirements.txt +pip install -r requirements.txt ``` 如果需要开启Unsloth，建议安装或者更新以下Python包： @@ -303,7 +303,7 @@ torchrun --nproc_per_node={num_gpus} train.py --train_args_file train_args/sft/q 💻 单卡QLoRA进行DPO训练： ```bash -python train.py --train_args_file train_args/sft/qlora/minicpm-2b-dpo-qlora.json +python train.py --train_args_file train_args/dpo/qlora/minicpm-2b-dpo-qlora.json ``` ## 模型使用