search

google / uis-rnn

This is the library for the Unbounded Interleaved-State Recurrent Neural Network (UIS-RNN) algorithm, corresponding to the paper Fully Supervised Speaker Diarization.
https://arxiv.org/abs/1810.04719
Apache License 2.0
1.56k stars 319 forks source link

Any plan on upgrading to pytorch 1.0+? #47

Closed tanvirfuad closed 4 years ago

tanvirfuad commented 5 years ago

Describe the question

Wanted to use in Tesla 4 on Google Cloud VM. Apparently, they do not support lower versions of CUDA other than CUDA 10 on T4. And, for pytorch 0.4.1, it requires a lower version of CUDA.

My background

I just finished my masters and working as a speech recognition engineer.

Have I read the README.md file?

Have I searched for similar questions from closed issues?

Have I tried to find the answers in the paper Fully Supervised Speaker Diarization?

Have I tried to find the answers in the reference Speaker Diarization with LSTM?

Have I tried to find the answers in the reference Generalized End-to-End Loss for Speaker Verification?

wq2012 commented 5 years ago

@tanvirfuad Thanks for the suggestion. It's a good point. I will take a look at that later.

Also, did you try to run uis-rnn (as it is) with PyTorch 1.0? And if yes, did you see any breakage?

I doubt it may just work correctly with PyTorch 1.0 without any changes.

tanvirfuad commented 5 years ago

Thank you for the quick reply. And yes, It does work on pytorch 1.0. Thank you so much for pointing the direction. Earlier, it was breaking for pytorch 1.1, but I did not try on 1.0. I actually forgot what error I was getting. If you want I can try that again and post here. Otherwise, please feel free to close the issue :)

wq2012 commented 5 years ago

No worries. I can give it a try with PyTorch 1.1 myself.