Calling trainer.fit fails with: AttributeError: 'dict' object has no attribute 'pretty'

[nithinivi]

🐛 Bug

dataset = MNIST(os.getcwd(), download=True, transform=transforms.ToTensor())
train, val = random_split(dataset, [55000, 5000])

autoencoder = LitAutoEncoder()
trainer = pl.Trainer()
trainer.fit(autoencoder, DataLoader(train))

Calling trainer.fit fails with AttributeError: 'dict' object has no attribute 'pretty'

To Reproduce

GPU available: True, used: False
TPU available: None, using: 0 TPU cores
/home/nithin/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/utilities/distributed.py:50: UserWarning: GPU available but not used. Set the --gpus flag when calling the script.
  warnings.warn(*args, **kwargs)
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
<ipython-input-9-ec8282b1f4ec> in <module>
      4 autoencoder = LitAutoEncoder()
      5 trainer = pl.Trainer()
----> 6 trainer.fit(autoencoder, DataLoader(train))

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/trainer/trainer.py in fit(self, model, train_dataloader, val_dataloaders, datamodule)
    508         self.call_hook('on_fit_start')
    509 
--> 510         results = self.accelerator_backend.train()
    511         self.accelerator_backend.teardown()
    512 

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/accelerators/accelerator.py in train(self)
     54 
     55     def train(self):
---> 56         self.trainer.setup_trainer(self.trainer.model)
     57         return self.train_or_test()
     58 

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/trainer/trainer.py in setup_trainer(self, model)
    442             self.logger.log_hyperparams(ref_model.hparams_initial)
    443             self.logger.log_graph(ref_model)
--> 444             self.logger.save()
    445 
    446         # wait for all to join if on distributed

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/utilities/distributed.py in wrapped_fn(*args, **kwargs)
     38     def wrapped_fn(*args, **kwargs):
     39         if rank_zero_only.rank == 0:
---> 40             return fn(*args, **kwargs)
     41 
     42     return wrapped_fn

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/loggers/tensorboard.py in save(self)
    234         # save the metatags file if it doesn't exist
    235         if not self._fs.isfile(hparams_file):
--> 236             save_hparams_to_yaml(hparams_file, self.hparams)
    237 
    238     @rank_zero_only

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/pytorch_lightning/core/saving.py in save_hparams_to_yaml(config_yaml, hparams)
    386         with fs.open(config_yaml, "w", encoding="utf-8") as fp:
    387             try:
--> 388                 OmegaConf.save(hparams, fp)
    389                 return
    390             except (UnsupportedValueType, ValidationError):

~/.pyenv/versions/3.7.3/lib/python3.7/site-packages/omegaconf/omegaconf.py in save(config, f, resolve)
    268         :param resolve: True to save a resolved config (defaults to False)
    269         """
--> 270         data = config.pretty(resolve=resolve)
    271         if isinstance(f, (str, pathlib.Path)):
    272             with io.open(os.path.abspath(f), "w", encoding="utf-8") as file:

AttributeError: 'dict' object has no attribute 'pretty'

Expected behavior

Calling trainer.fit(autoencoder, DataLoader(train)) should train MNIST classifier without errors.

Environment

• CUDA:
  • GPU:
    • GeForce GTX 1050 Ti
  • available: True
  • version: 10.2
• Packages:
  • numpy: 1.19.5
  • pyTorch_debug: False
  • pyTorch_version: 1.7.1
  • pytorch-lightning: 1.1.6
  • tqdm: 4.50.0
• System:
  • OS: Linux
  • architecture:
    • 64bit
  • processor: x86_64
  • python: 3.7.3
  • version: change Checkpoint callback's save_best_only to save_top_k #70~18.04.1-Ubuntu SMP Tue Jan 12 17:18:00 UTC 2021

Additional context

The program was running on jupyter notebook.

[omry]

Try to upgrade to OmegaConf 2.0, you have an incompatible version installed.

[Borda]

@nithinivi what OmegaConf are you using?

[nithinivi]

@Borda @omry The version number of OmegaConf is 2.0.0

[omry]

@nithinivi,

config.pretty() is deprecated in OmegaConf 2.0 and the source code no longer contain calls to it.
The stack trace is definitely not from OmegaConf 2.0. If you think this is wrong please provide repro instructions from scratch (including the creation of a virtualenv or a conda environment).

By the way, please upgrade to latest OmegaConf 2.0 release (currently 2.0.6).

[nithinivi]

I created a new conda env and installed the torch and pytorch-ligthing.
Tried out the same code and the issue was not reproduced. So I do believe this was a issues with the environment setup and it's not using OmegaConf 2.0.

I have installed python using pyenv and executing the code in a jupyter notebook.

[roytseng-tw]

I had the same issue with omegaconf 2.0.1rc11.
After update to the latest version 2.0.6, the issue is gone.

[rohitgr7]

I believe issue is resolved now. Please feel free to reopen if required :)