rishikksh20 / vae_tacotron2 Goto Github PK

I get the error "loss explode" in the training stage!
I'm not modifying the original hyperparameters, and I want to know how to solve the problem.

I run on gpu ，but the it shows that the process is killed

pip install -r -requirements.txt
Just install the libraries in the requirement? LWS?

when i ran python synthesize.py --model ='Tacotron'--mode ='eval'--reference_audio ='ref_1.wav' ,an issue occured!
ValueError: could not convert string to float: 'training_data/mels\speech-mel-00001.npy'

please help me solving this problem
thanks!!!

Hi, I run your code with tf1.6, cudnn10.1 and cudnn9.2. Both got

2019-08-13 23:13:50.199673: E tensorflow/stream_executor/cuda/cuda_dnn.cc:378] Loaded runtime CuDNN library: 7600 (compatibility version 7600) but source was compiled with 7102 (compatibility version 7100).  If using a binary install, upgrade your CuDNN library to match.  If building from sources, make sure the library loaded at runtime matches a compatible version specified during compile configuration.
2019-08-13 23:13:50.201349: F tensorflow/core/kernels/conv_ops.cc:717] Check failed: stream->parent()->GetConvolveAlgorithms( conv_parameters.ShouldIncludeWinogradNonfusedAlgo<T>(), &algorithms) 

I google this error, and most said it is related to the cudnn version. May I ask what is cudnn version you used? Or could you give some advice? Thanks a lot.

what does this mean? I'm an amatuer at this sort of thing

@rishikksh20 Hi, I use this repo with Blizzard2013 dataset instead of ljspeech dataset with default settings. I want to know whether I am training this vae_model right. or not ? What a successful error curve should look like with this repo? The followings are my curves. I wonder whether the difference between kl_loss and reconstruction_loss is too wide. Thanks for any help?

G:\vae_tacotron2-master>python train.py --model='Tacotron'
Traceback (most recent call last):
File "train.py", line 33, in
main()
File "train.py", line 24, in main
raise ValueError('please enter a valid model to train: {}'.format(accepted_models))
ValueError: please enter a valid model to train: ['Tacotron', 'Wavenet']