From 4167442627b1414ff8fdc86528812b46168c656b Mon Sep 17 00:00:00 2001 From: Piotr Kozakowski Date: Sun, 19 Nov 2017 20:23:26 +0100 Subject: Add weight normalization --- README.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'README.md') diff --git a/README.md b/README.md index 34e8530..4d39632 100644 --- a/README.md +++ b/README.md @@ -4,7 +4,7 @@ A PyTorch implementation of [SampleRNN: An Unconditional End-to-End Neural Audio ![A visual representation of the SampleRNN architecture](http://deepsound.io/images/samplernn.png) -It's based on the reference implementation in Theano: https://github.com/soroushmehr/sampleRNN_ICLR2017. Unlike the Theano version, our code allows training models with arbitrary number of tiers, whereas the original implementation allows maximum 3 tiers. However it doesn't have weight normalization and doesn't allow using LSTM units (only GRU). For more details and motivation behind rewriting this model to PyTorch, see our blog post: http://deepsound.io/samplernn_pytorch.html. +It's based on the reference implementation in Theano: https://github.com/soroushmehr/sampleRNN_ICLR2017. Unlike the Theano version, our code allows training models with arbitrary number of tiers, whereas the original implementation allows maximum 3 tiers. However it doesn't allow using LSTM units (only GRU). For more details and motivation behind rewriting this model to PyTorch, see our blog post: http://deepsound.io/samplernn_pytorch.html. ## Dependencies -- cgit v1.2.3-70-g09d2