George S

2 Test Cases [ITensor + constant input types] - Indices as input [Differing Semantics between PyTorch and TensorRT --> PyTorch gets Int64, TRT gets Int32] - Known operator to assemble...

I'm not sure what the `empty_like` lowers to, but potentially you could add another operation in the `nn.Module` so that the graph is non-empty. It is likely the case that...

Thanks for the analysis @apbose - this is very helpful. It looks like the `constant_folding` lowering pass is freezing the memory for the `empty_like` operator and storing it as an...

See [this decomposition](https://github.com/pytorch/pytorch/blob/e5e0685f614171ab3e3a22c85dd4ee8cd0cf09ae/aten/src/ATen/functorch/BatchRulesScatterOps.cpp#L881-L890) for an alternative approach.

So, in this case would the implementation not be functional without the `slice_scatter` decomposition? Additionally, if the `slice_scatter` decomposition changes the behavior of `torch.slice_scatter`, in the sense that the example...

@bowang007 - yes, there is support for that operator, as here: https://github.com/pytorch/TensorRT/blob/cf3a6887626c648e5747fdbfa5bc62b361a82b02/py/torch_tensorrt/dynamo/conversion/aten_ops_converters.py#L123-L152

I am able to reproduce this error in the TorchScript path on the latest `main` and with NeMo toolkit `1.20.0`. It seems to stem from tensor addition operators which are...

When tracing this model with the `torch_compile` IR option on `main`, we encounter the errors from #2183 and #2227, for which a fix is in-progress. I will post an update...

The issue does not seem to be with the Conformer architecture itself, since inference in plain PyTorch is working, and it is scripting to TorchScript successfully. There is a possibility...

Regarding the TorchScript path, the bug occurs on [this line](https://github.com/NVIDIA/NeMo/blob/ef730aae74fd7e5a91efa0a9f37e0746388c12db/nemo/collections/asr/parts/submodules/multi_head_attention.py#L243), where the shape of `matrix_ac` and `matrix_bd` disagree. Specifically, the issue is that this line attempts to drop extra elements...