提交 a1fc2c30 编写于 作者: littletomatodonkey's avatar littletomatodonkey 提交者: Tingquan Gao

add cpu note, test=document_fix

上级 3388c47e
......@@ -63,6 +63,9 @@ export CUDA_VISIBLE_DEVICES=0
set CUDA_VISIBLE_DEVICES=0
```
* If you want to train on cpu device, you can modify the field `use_gpu: True` in the config file to `use_gpu: False`, or you can append `-o use_gpu=False` in the training command, which means override the value of `use_gpu` as False.
### Train from scratch
* Train ResNet50_vd
......@@ -71,6 +74,14 @@ set CUDA_VISIBLE_DEVICES=0
python3 tools/train.py -c ./configs/quick_start/ResNet50_vd.yaml
```
If you want to train on cpu device, the command is as follows.
```shell
python3 tools/train.py -c ./configs/quick_start/ResNet50_vd.yaml -o use_gpu=False
```
Similarly, for the following commands, if you want to train on cpu device, you can append `-o use_gpu=False` in the command.
The validation `Top1 Acc` curve is shown below.
![](../../images/quick_start/r50_vd_acc.png)
......
......@@ -46,8 +46,7 @@ cd ../
### 2.2 环境说明
* 下面所有的训练过程均在`单卡V100`机器上运行。
* 首先需要设置可用的显卡设备id
* 下面所有的训练过程均在`单卡V100`机器上运行。首先需要设置可用的显卡设备id。
如果使用mac或者linux,可以使用下面的命令进行设置。
......@@ -61,6 +60,7 @@ export CUDA_VISIBLE_DEVICES=0
set CUDA_VISIBLE_DEVICES=0
```
* 如果希望在cpu上训练,可以将配置文件中的`use_gpu: True`修改为`use_gpu: False`,或者在训练脚本后面添加`-o use_gpu=False`,表示传入参数,覆盖默认的`use_gpu`值。
## 三、模型训练
......@@ -72,6 +72,15 @@ set CUDA_VISIBLE_DEVICES=0
python3 tools/train.py -c ./configs/quick_start/ResNet50_vd.yaml
```
如果希望在cpu上训练,训练脚本如下所示。
```shell
python3 tools/train.py -c ./configs/quick_start/ResNet50_vd.yaml -o use_gpu=False
```
下面的训练任务中,如果希望使用cpu训练,也可以在训练脚本中添加`-o use_gpu=False`
验证集的`Top1 Acc`曲线如下所示,最高准确率为0.2735。
![](../../images/quick_start/r50_vd_acc.png)
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册