From 8dce5f7e7c92d58926125ff7a54b7b3bb8a0e6db Mon Sep 17 00:00:00 2001 From: hzg0601 Date: Mon, 4 Sep 2023 19:26:35 +0800 Subject: [PATCH] =?UTF-8?q?update=20readme.md:=E5=A2=9E=E5=8A=A0=E5=A4=9A?= =?UTF-8?q?=E5=8D=A1=E5=90=AF=E5=8A=A8gpus=E5=8F=82=E6=95=B0=E4=B8=8D?= =?UTF-8?q?=E7=94=9F=E6=95=88=E7=9A=84=E8=AF=B4=E6=98=8E?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit --- README.md | 8 +++++++- 1 file changed, 7 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 9d48c99..577e7f7 100644 --- a/README.md +++ b/README.md @@ -273,7 +273,13 @@ max_gpu_memory="20GiB" `max_gpu_memory` 控制每个卡使用的显存容量。 -注：server_config.py的FSCHAT_MODEL_WORKERS字典中也 +注1：server_config.py的FSCHAT_MODEL_WORKERS字典中也增加了相关配置，如有需要也可通过修改FSCHAT_MODEL_WORKERS字典中对应参数实现多卡加载。 + +注2：少数情况下，gpus参数会不生效，此时需要通过设置环境变量CUDA_VISIBLE_DEVICES来指定torch可见的gpu,示例代码： + +```shell +CUDA_VISIBLE_DEVICES=0,1 python startup.py -a +``` #### 5.4 PEFT 加载(包括lora,p-tuning,prefix tuning, prompt tuning,ia3等)