search

microsoft / tensorflow-directml-plugin

DirectML PluggableDevice plugin for TensorFlow 2
Apache License 2.0
185 stars 25 forks source link

tfdml_plugin.dll not found #331

Open halfman2016 opened 1 year ago

halfman2016 commented 1 year ago

tensorflow.python.framework.errors_impl.NotFoundError: C:\ProgramData\Miniconda3\envs\tfdml_plugin\lib\site-packages\tensorflow-plugins\tfdml_plugin.dll not found

when i install,run any code, like import tensorflow as tf,it is error ,not found tfdml_plugin.dll . but,the file is right in the path. how can i do ?

halfman2016 commented 1 year ago

N卡，A卡都一样，系统win11

phgilde commented 1 year ago

I have this issue too. Windows 10, Version 22H2 Build 19045.2251 Radeon RX570 GPU with newest drivers.

phgilde commented 1 year ago

I could fix the issue by installing tensorflow-cpu==2.10 with the --user flag.

Zhaopudark commented 1 year ago

The same bug as mine, I'll try to build TensorFlow-directml-plugin tomorrow, anyway today is too late.

PatriceVignola commented 1 year ago

Yes, we recently found that TF 2.11 is not backward compatible with plugins built for earlier versions. Since pluggable devices are still experimental, this is to be expected. In the future, we'll hard lock the version of TF until pluggable devices are in a more stable state.

We will soon be releasing a new version of tensorflow-directml-plugin that addresses the problem. In the meantime, please manually install tensorflow-cpu==2.10 to get rid of the problem.

KiTant commented 1 year ago

I still have this error. I installed tensorflow-cpu==2.10.0 with the flag --user and without this flag but I still have this error: tfdml_plugin.dll not found.

chanangaza commented 1 year ago

Try uninstalling all of Keras and Tensorflow libs (including Tensorboard). Then reinstall tensorflow-directml-plugin.

Worked for me.

KiTant commented 1 year ago

That worked thx

But i getting other error when training AI:

No OpKernel was registered to support Op 'CudnnRNN' used by {{node CudnnRNN}} with these attrs: [seed=0, dropout=0, T=DT_FLOAT, input_mode="linear_input", direction="unidirectional", rnn_mode="gru", seed2=0, is_training=true]
Registered devices: [CPU, GPU]
Registered kernels:
  <no registered kernels>

     [[CudnnRNN]]
     [[model/gru/PartitionedCall]] [Op:__inference_train_function_5775]
TeaTimeChimp commented 1 year ago

Re. CudnnRNN, sounds similar to this issue I commented on https://github.com/microsoft/tensorflow-directml-plugin/issues/364