Error running basic use with trainer.py

[bgheneti]

Hi, thanks for this repo and your amazing work!

I am trying to get the Basic Use example for validation working, after installing the repo on an instance, and am running into some issues.

python trainer.py fit --model ClayMAEModule --data ClayDataModule --config configs/config.yaml --trainer.fast_dev_run=True seems to produce the following error:

error: Parser key "model":
Unable to load config 'ClayMAEModule'
  - Parser key "model": Unable to load config "ClayMAEModule"

---

So instead I run python trainer.py fit --config configs/config.yaml --trainer.fast_dev_run=True but get the following error:

[rank0]:     n_train, n_test = _validate_shuffle_split(
[rank0]:                       ^^^^^^^^^^^^^^^^^^^^^^^^
[rank0]:   File "/home/banti_understory_ai/miniforge3/envs/claymodel/lib/python3.11/site-packages/sklearn/model_selection/_split.py", line 2481, in _validate_shuffle_split
[rank0]:     raise ValueError(
[rank0]: ValueError: With n_samples=0, test_size=0.19999999999999996 and train_size=None, the resulting train set will be empty. Adjust any of the aforementioned parameters.

As a note I also changed devices to 1 and num_nodes to 1 in config.yaml to accommodate my machine

Looking at the config.yaml it seems like I need data in the correct location (/fsx). Is this correct? If so which data do I need to obtain (just one or all of the platforms?) and how do I need to structure it there? Any help getting acclimated is appreciated 🙂

[olivaboom]

Hi @bgheneti ! Did you ever figure out how to solve this issue? I am getting the same error trying to use the basic use example

[shashankvasisht]

Its a data not found error. Have you updated the config file with the correct data path ? make sure that the model path is correct and compatible with the data loader