lucidrains/muse-maskgit-pytorch

AssertionError while trying to train with base_maskgit_trainer() after finished training with vae_trainer()

easonoob opened this issue · 0 comments

AssertionError while trying to train with base_maskgit_trainer() after finished training with vae_trainer(). I finished training the vae trainer with about 123k iterations, then while I moved on to the base maskgit trainer, the following error occured:

123000: vae loss: 1.0881760120391846 - discr loss: 1.993680715560913
training complete
Resuming VAE from:  other/vae.23000.base.pt
Traceback (most recent call last):
  File "muse_train.py", line 352, in <module>
    base_maskgit_trainer()
  File "muse_train.py", line 156, in base_maskgit_trainer
    vae.load(args.resume_from.replace('.pt' , '.base.pt'))    # you will want to load the exponentially moving averaged VAE
  File "/home/eason/PyTorch/muse-pytorch/muse_maskgit_pytorch/vqgan_vae.py", line 408, in load
    assert path.exists()
AssertionError