Tamas Bela Feher

Results 82 comments of Tamas Bela Feher

I believe the issues have been addressed by @Nyrio. @bixia1 could you have a look?

Tagging @bixia1 and @DEKHTIARJonathan for review.

I have resolved the conflicts and added more tests. Sorry for the delay. @bixia1 this is ready for review

Thanks @meena-at-work for this PR! The converter looks good, but we would still need to test the converter with the `batch_dims` attribute. And since we are changing this converter, it...

There are two more places where we need change: - https://github.com/tensorflow/tensorflow/blob/036f00f02ce50e9f34619dab48ac8ed7956f684c/tensorflow/python/compiler/tensorrt/trt_convert.py#L1467 needs to change to ```python if first_input is None: ``` - https://github.com/tensorflow/tensorflow/blob/036f00f02ce50e9f34619dab48ac8ed7956f684c/tensorflow/python/compiler/tensorrt/trt_convert.py#L1488-L1492 needs to be replaced by ```python args,...

Hi @viclafargue, one thing to keep in mind: protocol buffers have a size limitation of 2 GiB ([cheet-sheet](https://developers.google.com/protocol-buffers/docs/encoding#cheat-sheet)). While it is not an issue here, we should still think about...

Yes, I agree. We can return to this question after https://github.com/rapidsai/raft/pull/652

@tarang-jain feel free to add any other issues to the list above that we want to track (related to #3044)