File tree Expand file tree Collapse file tree 2 files changed +15
-3
lines changed
Expand file tree Collapse file tree 2 files changed +15
-3
lines changed Original file line number Diff line number Diff line change @@ -193,7 +193,13 @@ accelerate launch --config_file accelerate_fsdp_config.yaml pefts/mft_accelerate
193193or
194194FSDP config in command line arguments
195195``` bash
196- sh ds_single_launch.sh
196+ sh fsdp_single_launch.sh
197+ ```
198+
199+ #### MultiNode Launch
200+ Refer to the deepspeed multi-node launch script below.
201+ ``` bash
202+ sh ds_multinode_launch.sh
197203```
198204
199205#### Traing Arguments
Original file line number Diff line number Diff line change @@ -146,7 +146,7 @@ QLoRA通过4-bit的nf4量化,且加入更多adapter,在大幅减少显存消
146146QLoRA论文指出,该方法可以在一张V100上对33B的模型进行微调,并且性能逼近全量参数微调。
147147
148148执行如下命令即可进行 Lora/QLora/全量 微调:
149- #### Launch via Deepspeed
149+ #### Deepspeed 单机启动
150150DeepSpeed配置在accelerate_ds_config.yaml中。
151151``` bash
152152accelerate launch --config_file accelerate_ds_config.yaml pefts/mft_accelerate.py --train_config configs/xxx_train_config.json --distributed_type " DeepSpeed"
@@ -163,7 +163,7 @@ DeepSpeed Zero3 配置在脚本中通过命令行输入
163163sh ds_zero3_single_launch.sh
164164```
165165
166- #### Launch via FSDP
166+ #### FSDP 单机启动
167167FSDP配置在accelerate_fsdp_config.yaml中。
168168``` bash
169169accelerate launch --config_file accelerate_fsdp_config.yaml pefts/mft_accelerate.py --train_config configs/xxx_train_config.json --distributed_type " FSDP"
@@ -175,6 +175,12 @@ FSDP配置在脚本中通过命令行输入。
175175sh fsdp_single_launch.sh
176176```
177177
178+ #### 多机启动
179+ 多机启动请参考如下deepspeed多机启动脚本
180+ ``` bash
181+ sh ds_multinode_launch.sh
182+ ```
183+
178184#### 训练参数
179185_ ** 训练需要的参数配置在``` configs/*_train_config ``` 中,主要参数说明如下:** _
180186
You can’t perform that action at this time.
0 commit comments