peterjc123

> aten::len ops is not supported when I try to convert my resnet model to tflite Could you please tell me what is the corresponding op in PyTorch (e.g. `torch.mean`)?...

> Any plan to support quantized::instance_norm? Would you please give me an example with a TFLite model quantized InstanceNorm layer? Once it is supported in TFLite, we will look into...

> > > Any plan to support quantized::instance_norm? > > > > > > Would you please give me an example with a TFLite model quantized InstanceNorm layer? Once it...

> Is there any chance to implement torch `aten::scaled_dot_product_attention`? https://pytorch.org/docs/master/generated/torch.nn.functional.scaled_dot_product_attention.html says it could be done as > > ``` > # Efficient implementation equivalent to the following: > scale_factor =...

Can [tflite2tf](https://github.com/alibaba/TinyNeuralNetwork/tree/main/contrib/tflite2tf) fulfill your needs? Why do you want keras models? I don't think we will support something like training models in Tensorflow.

Before it is implemented, I guess it is easier if you replace F.scaled_dot_product_attention with the given implementation.

In order to get it supported, we will need to support the new TorchScript schema first. Also, we should probably figure out better way to reuse the conversation logic of...

@BmanClark Could you please tell me which version of Tensorflow (Lite) are you testing against?

It is indeed a conversion error. Could you please upload a TorchScript model and the shape of the inputs so that we can reproduce this issue.

> TFLite is 2.8.0. The line number is out a few from latest, but the code seems pretty much the same in transpose_conv.cc I'll work on getting a torchscript version...