Nauman Mustafa

I also got the same error, looking forward to solution!

It is problem in windows, works pretty well in linux!

https://colab.research.google.com/github/openai/DALL-E/blob/master/notebooks/usage.ipynb

While standard APIs provide all the necessary information to build solid agents, most of the reinforcement learning literature relies on visual information. Secondly, in order to build agents which learn...

The last time I checked, the data stored in `calculated.gg` and decoding it with carball is not very accurate in terms of time intervals. I think it also happens in...

It should be possible to train smaller models quickly. follow #10 I am training on Colab Pro

Even smaller model works pretty neat: ```python vae = DiscreteVAE( num_layers = 3, num_tokens = 4096, codebook_dim = 512, hidden_dim = 256, ) ``` Here are the samples:  ...

Yes these results are for VAE - took only ~30 min to an 1hr on colab pro (V100) - I am in process of training DALLE - results should be...

Sure! [Here is the notebook](https://colab.research.google.com/drive/1YResxh6eNioat4h-oFdr7APhLrQAjBtq?usp=sharing) so far. Also there is an update after discussion on #12 and applying the fix here are the results I got which is not so...

Here are results after a few hours of training of DALL-E:    Loss is still very high right now but its going down slowly