I am really interested in Tensorflow MLIR, and hope to create custom dialects from Tensorflow’s existing dialects.
I found that recently there is a TensorFlow Graph IR , it seems quite exciting. May I ask what is its def status by now?
Does it support basic runtime? I mean, could I use it for actually run by now? I am going to use it for optimizing ResNet training, is it ok now?
Besides, Is there any dev discussion group for now? Like PyTorch’s developer group in slack.