Comments (8)
- About the NoneType bug- are you sure you are using the correct version of TF? This repository may not work with other versions of TF.
- As with all neural networks, this model trains much faster on GPU than on CPU. If you don't have a GPU it will take a long time to train the network.
If you have further questions you can email me.
from rnn.wgan.
Dear Ofir,
My Tensorflow version is 1.1 as your code required. I also tried on other version of TF (with a little modification) but it still got the same error.
About the second problem, I will try training on GPU later. Thank you.
from rnn.wgan.
I'm not exactly sure why you are receiving this error but I think it could be because your checkpoint directory is incorrect.
Maybe it should be:
python generate.py --CKPT_PATH=./logs/Generator_RNN_CL_VL_TH-Discriminator_RNN-50-10-512-512-1509799522.05-/checkpoint --DISC_GRU_LAYERS=2 --GEN_GRU_LAYERS=2
(Notice the period I put at the beginning of the directory)
from rnn.wgan.
I tried that path but it still got the same error @@.
from rnn.wgan.
Ofir, in your code, file generate.py, lines 20:
_, inference_op = Generator(BATCH_SIZE, charmap_len, seq_len=SEQ_LEN)
You just initialize the Generator with 3 first parameter, so the rnn_cell parameter is still None, is that okay?
from rnn.wgan.
Yup the code is fine.
Make sure that this directory actually exists
./logs/Generator_RNN_CL_VL_TH-Discriminator_RNN-50-10-512-512-1509799522.05-/checkpoint
I think that the problem is that the directory you gave the program is wrong.
from rnn.wgan.
same issue in #10 . I'm also not familiar with the code, but after some code reading, I found maybe the problem is in the line L20 of generate.py, where you should pass the correct parameter rnn_cell
to the function call:
_, inference_op = Generator(BATCH_SIZE, charmap_len, seq_len=SEQ_LEN, rnn_cell=RNN_CELL) disc_fake = Discriminator(inference_op, charmap_len, SEQ_LEN, reuse=False, rnn_cell=RNN_CELL)
But this single fix won't let the code work. After some fail and trial, I found the line L238 in single_length_train.py has changed the ckp data path by arbitrarily adding a /ckp
suffix, which breaks the readme guide.
so, instead of using path like:
--CKPT_PATH=./logs/Generator_RNN_CL_VL_TH-Discriminator_RNN-50-10-512-512-1516543503.49-/checkpoint
you should type something like:
--CKPT_PATH=./logs/Generator_RNN_CL_VL_TH-Discriminator_RNN-50-10-512-512-1516543503.49-/checkpoint/seq-5/ckp
I'm so sorry to point out that maybe the merge from NickShahML breaks some code, which the authors may not intend to? :)
from rnn.wgan.
Thanks for reporting this. We've reverted the code back to what it was when we published the paper. We believe this issue should be solved, as that code was thoroughly tested by us.
If this issue persists, please post here.
from rnn.wgan.
Related Issues (11)
- Fisher GAN and wider GRU's Increase performance
- generate errors HOT 4
- How long does this take to train with the default hyperparameters? HOT 6
- Is there a way to generate variable length text? HOT 1
- Adapt code for Lyrics Generation HOT 2
- which file should download and how to preprocess? HOT 2
- Is there an option to turn off CL/VL/TH? HOT 1
- alpha_optimizer_op not actually run? HOT 2
- WGAN-GP vs. FisherGAN on langauge generation in practice HOT 1
- RNN with GANs vs Independent RNN Model HOT 2
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from rnn.wgan.