search

microsoft / tensorflow-directml-plugin

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

tensorflow-cpu 2.12.0 encounter issue. #367

Open yunot opened 1 year ago

yunot commented 1 year ago

The version of tensorflow-directml-plugin is: 0.4.0.dev230202 after I reinstalled tensorflow-cpu 2.12.0. I encountered the Error as below: tfdml_plugin.dll not found

leo-smi commented 1 year ago

same here

PhilippWillms commented 1 year ago

Also occurring for tensorflow-2.11.1 tensorflow-cpu-2.11.1

leo-smi commented 1 year ago

Try updating your GPU drivers, worked for me. For some reason intel driver assistent downgraded the driver, but I installed again.

leo-smi commented 1 year ago

tensorflow-directml-plugin is compatible with packages version < 2.10

PhilippWillms commented 1 year ago

tensorflow-directml-plugin is compatible with packages version < 2.10

You mean tensorflow versions below 2.10 ?

leo-smi commented 1 year ago

tensorflow-directml-plugin is compatible with packages version < 2.10

You mean tensorflow versions below 2.10 ?

my bad <= 2.10

I recommend deleting de env, create a new env with python 3.10 and just run pip install tensorflow-directml-plugin, this package will install the other dependencies by itself.

PatriceVignola commented 11 months ago

I apologize for the delay. We had to pause the development of this plugin until further notice. For the time being, all latest DirectML features and performance improvements are going into onnxruntime for inference scenarios. We'll update this issue if/when things change.

leo-smi commented 11 months ago

I apologize for the delay. We had to pause the development of this plugin until further notice. For the time being, all latest DirectML features and performance improvements are going into onnxruntime for inference scenarios. We'll update this issue if/when things change.

Is there a plan to make onnx works for training in windows soon?

PatriceVignola commented 11 months ago

ONNX Runtime has some support for training, but not for DirectML. It's not on our immediate roadmap as we're currently focused on inference performance for transformer models (e.g. Stable Diffusion, LLaMA 2).