jinhan / tacotron2-gst Goto Github PK
View Code? Open in Web Editor NEWTacotron2 with Global Style Tokens
License: BSD 3-Clause "New" or "Revised" License
Tacotron2 with Global Style Tokens
License: BSD 3-Clause "New" or "Revised" License
Hello @jinhan ,
Thanks for sharing such great repo. _ was trynign to train in fp16. But it gives me following error. Please can you help me fix it.. Thanks.. I ahve enablep16 training.
Command: python -m multiproc train.py --output_directory=outdir --log_directory=logdir
Traceback (most recent call last):
File "train.py", line 285, in <module>
args.warm_start, args.n_gpus, args.rank, args.group_name, hparams)
File "train.py", line 222, in train
optimizer.backward(loss)
File "/projects/playground/tacotron2-gst/fp16_optimizer.py", line 362, in backward
self.loss_scaler.backward(loss.float())
File "/projects/playground/tacotron2-gst/loss_scaler.py", line 80, in backward
scaled_loss.backward()
File "/pyenvs/taco/lib/python3.6/site-packages/torch/tensor.py", line 118, in backward
torch.autograd.backward(self, gradient, retain_graph, create_graph)
File "/pyenvs/taco/lib/python3.6/site-packages/torch/autograd/__init__.py", line 93, in backward
allow_unreachable=True) # allow_unreachable flag
File "/projects/playground/tacotron2-gst/distributed.py", line 157, in allreduce_params
coalesced = _flatten_dense_tensors(grads)
File "/projects/playground/tacotron2-gst/distributed.py", line 19, in _flatten_dense_tensors
flat = torch.cat([t.contiguous().view(-1) for t in tensors], dim=0)
RuntimeError: Expected object of scalar type Half but got scalar type Float for sequence element 3 in sequence argument at position #1 'tensors'
```
Hi,
Im trying to use GST in my tacotron2 project, but problem that no matter how i try, gst tokens not working s intended.
They doesn't add any style to the audio. Maybe there are some tricks not mentioned in the paper?
Can u share your experience and pitfalls during the training?
Many thanks)
My code is based on the current repository Nvidia tacotron2 and I used your GST code.
Using ref audio work more or less, but I can't get to work style tokens.
It looks like this
https://i.imgur.com/wK7vZ73.png
Hi, recently I tried run codes on chinese data set , I just edit text_to_sequence, but grad_norm is always 'nan', please help me~~!
thanks!
Getting following error :
/pytorch/aten/src/THC/THCTensorIndex.cu:362: void indexSelectLargeIndex(TensorInfo<T, IndexType>, TensorInfo<T, IndexType>, TensorInfo<long, IndexType>, int, int, IndexType, IndexType, long) [with T = float, IndexType = unsigned int, DstDim = 2, SrcDim = 2, IdxDim = -2, IndexIsMajor = true]: block: [119,0,0], thread: [95,0,0] Assertion `srcIndex < srcSelectDimSize` failed.
Traceback (most recent call last):
File "train.py", line 285, in <module>
args.warm_start, args.n_gpus, args.rank, args.group_name, hparams)
File "train.py", line 213, in train
y_pred = model(x)
File "/home/rishikesh/.local/lib/python3.6/site-packages/torch/nn/modules/module.py", line 489, in __call__
result = self.forward(*input, **kwargs)
File "/home/rishikesh/Dev/tacotron2-gst/model.py", line 516, in forward
transcript_outputs = self.encoder(embedded_inputs, input_lengths)
File "/home/rishikesh/.local/lib/python3.6/site-packages/torch/nn/modules/module.py", line 489, in __call__
result = self.forward(*input, **kwargs)
File "/home/rishikesh/Dev/tacotron2-gst/model.py", line 177, in forward
x = F.dropout(F.relu(conv(x)), drop_rate, self.training)
File "/home/rishikesh/.local/lib/python3.6/site-packages/torch/nn/functional.py", line 749, in dropout
else _VF.dropout(input, p, training))
RuntimeError: Creating MTGP constants failed. at /pytorch/aten/src/THC/THCTensorRandom.cu:35
Environement:
python : 3.6
OS: Ubuntu 18.04
Pytorch : 1.0
tensorflow : 1.6
A declarative, efficient, and flexible JavaScript library for building user interfaces.
๐ Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. ๐๐๐
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google โค๏ธ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.