langml does not match tensorflow_gpu

[buthi]

No description provided.

[buthi]

I replayed your paper and found that only CPU was used for training. Please ask if you can provide a way to use GPU

[SeanLee97]

Sorry for the delayed reply.

Firstly, please ensure that the GPU is available in your environment, you can test the availability as follows:

$ python -c "import tensorflow as tf; print('CUDA Avaliable:', tf.test.is_gpu_available())"

If the GPU is not available, please ensure that the installed TensorFlow is the GPU version.

If the GPU is available, you should check the TF and Keras versions.

$ pip list | egrep "tensorflow|keras"

or

$ conda list | egrep "tensorflow|keras"

for conda.

tensorflow==1.15.0 and Keras==2.3.1 are recommended.

[buthi]

抱歉延迟回复。

首先，请确保GPU在您的环境中可用，您可以通过以下方式测试可用性：

$ python -c "import tensorflow as tf; print('CUDA Avaliable:', tf.test.is_gpu_available())"

如果 GPU 不可用，请确保安装的 TensorFlow 是 GPU 版本。

如果 GPU 可用，则应检查 TF 和 Keras 版本。

$ pip list | egrep "tensorflow|keras"

或者

$ conda list | egrep "tensorflow|keras"

对于康达。

tensorflow==1.15.0并被Keras==2.3.1推荐。

Thank you for your guidance. I checked it again as you instructed.


There should be no problems, but the GPU utilization is 0 during training, so only CPU should be used

[BABBU18]

请问GPU的问题解决了吗