Skip to content

P
Paddle

afeixing77 / Paddle
与 Fork 源项目一致

Fork自 PaddlePaddle / Paddle

通知 1
Star 0
Fork 0
P
Paddle

体验新版 GitCode，发现更多精彩内容 >>

未验证 提交 a62bbd1d 编写于 作者: K Kexin Zhao 提交者： GitHub
浏览文件
操作

fix float16 demo location issue (#10948)

上级 fb43c6b4
隐藏空白更改
内联 并排
Showing with 2 addition and 6 deletion
paddle/contrib/float16/README.md
浏览文件 @ a62bbd1d
...@@ -89,7 +89,7 @@ cd Paddle ...@@ -89,7 +89,7 @@ cd Paddle
# to `FROM nvidia/cuda:9.0-cudnn7-devel-ubuntu16.04` and similarly for other configurations # to `FROM nvidia/cuda:9.0-cudnn7-devel-ubuntu16.04` and similarly for other configurations
nvidia-docker build -t paddle:float16 . nvidia-docker build -t paddle:float16 .
# After running this, different results will be written to different log files in Paddle/contrib/float16/ # After running this, different results will be written to different log files in Paddle/contrib/float16/
nvidia-docker run -it -v $PWD:/paddle paddle:float16 /paddle/contrib/float16/run_float16_demo.sh nvidia-docker run -it -v $PWD:/paddle paddle:float16 /paddle/paddle/contrib/float16/run_float16_demo.sh
``` ```
#### Accuracy #### Accuracy
......
paddle/contrib/float16/run_float16_demo.sh
浏览文件 @ a62bbd1d
...@@ -3,7 +3,7 @@ ...@@ -3,7 +3,7 @@
BUILD_PATH=/paddle/fp16_build BUILD_PATH=/paddle/fp16_build
WHEEL_PATH=$BUILD_PATH/python/dist WHEEL_PATH=$BUILD_PATH/python/dist
INFER_PATH=$BUILD_PATH/paddle/fluid/inference/tests/book INFER_PATH=$BUILD_PATH/paddle/fluid/inference/tests/book
DEMO_PATH=/paddle/contrib/float16 DEMO_PATH=/paddle/paddle/contrib/float16
# Use the single most powerful CUDA GPU on your machine # Use the single most powerful CUDA GPU on your machine
export CUDA_VISIBLE_DEVICES=0 export CUDA_VISIBLE_DEVICES=0
...@@ -50,7 +50,6 @@ do ...@@ -50,7 +50,6 @@ do
--repeat=1 \ --repeat=1 \
$INFER_PATH/test_inference_image_classification_vgg \ $INFER_PATH/test_inference_image_classification_vgg \
--data_set=imagenet \
--dirname=$DEMO_PATH/image_classification_imagenet_vgg.inference.model \ --dirname=$DEMO_PATH/image_classification_imagenet_vgg.inference.model \
--fp16_dirname=$DEMO_PATH/float16_image_classification_imagenet_vgg.inference.model \ --fp16_dirname=$DEMO_PATH/float16_image_classification_imagenet_vgg.inference.model \
--repeat=$REPEAT \ --repeat=$REPEAT \
...@@ -68,7 +67,6 @@ do ...@@ -68,7 +67,6 @@ do
--repeat=1 \ --repeat=1 \
$INFER_PATH/test_inference_image_classification_resnet \ $INFER_PATH/test_inference_image_classification_resnet \
--data_set=imagenet \
--dirname=$DEMO_PATH/image_classification_imagenet_resnet.inference.model \ --dirname=$DEMO_PATH/image_classification_imagenet_resnet.inference.model \
--fp16_dirname=$DEMO_PATH/float16_image_classification_imagenet_resnet.inference.model \ --fp16_dirname=$DEMO_PATH/float16_image_classification_imagenet_resnet.inference.model \
--repeat=$REPEAT \ --repeat=$REPEAT \
...@@ -86,7 +84,6 @@ do ...@@ -86,7 +84,6 @@ do
--repeat=1 \ --repeat=1 \
$INFER_PATH/test_inference_image_classification_vgg \ $INFER_PATH/test_inference_image_classification_vgg \
--data_set=cifar10 \
--dirname=$DEMO_PATH/image_classification_cifar10_vgg.inference.model \ --dirname=$DEMO_PATH/image_classification_cifar10_vgg.inference.model \
--fp16_dirname=$DEMO_PATH/float16_image_classification_cifar10_vgg.inference.model \ --fp16_dirname=$DEMO_PATH/float16_image_classification_cifar10_vgg.inference.model \
--repeat=$REPEAT \ --repeat=$REPEAT \
...@@ -104,7 +101,6 @@ do ...@@ -104,7 +101,6 @@ do
--repeat=1 \ --repeat=1 \
$INFER_PATH/test_inference_image_classification_vgg \ $INFER_PATH/test_inference_image_classification_vgg \
--data_set=cifar10 \
--dirname=$DEMO_PATH/image_classification_cifar10_resnet.inference.model \ --dirname=$DEMO_PATH/image_classification_cifar10_resnet.inference.model \
--fp16_dirname=$DEMO_PATH/float16_image_classification_cifar10_resnet.inference.model \ --fp16_dirname=$DEMO_PATH/float16_image_classification_cifar10_resnet.inference.model \
--repeat=$REPEAT \ --repeat=$REPEAT \
......
Markdown is supported
0% 或 .
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 或