search

1Konny / Beta-VAE

Pytorch implementation of β-VAE
MIT License
526 stars 125 forks source link

Need Help on Visdom #14

Open tsuttaket2 opened 5 years ago

tsuttaket2 commented 5 years ago

Hi, I rerun your implementation on my machine but it seems to not work for me. Can you please suggest how can I solve this

  1. I am going to begin with CelebA dataset
  2. I extracted the dataset to folder D:...\ Beta-VAE-master\data\CelebA\Training
  3. I deleted most of the photos to make the size manageable to see if it can run (left with around 40 phtos)
  4. install visdom, torch, torchvision
  5. initialize visdom (python -m visdom.server) Step 5
  6. Use my browser to access http://localhost:8097/ Step 6 Output
  7. python main.py --dataset celeba --seed 1 --lr 1e-4 --beta1 0.9 --beta2 0.999 --objective H --model H --batch_size 64 --z_dim 10 --max_iter 1.5e3 --beta 10 --viz_name celeba_H_beta10_z10 Step 7

Even if there is no error shown, I there is no output shown on either http://localhost:8097/ or in the folder D:...\Beta-VAE-master\outputs\celeba_H_beta10_z10 Step 6 Output

sharpe5 commented 4 years ago

All I can suggest is trying different combinations of packages and Python versions, perhaps there is a version conflict. Miniconda is your friend here. The screenshots show that you are using Windows, perhaps try it on Linux?