📚 v1 - Docs Add the feature to resume training from a previous training

[shrinand1996]

What is the motivation for this task?

It will be great if we can resume training (for EfficientNet) since it takes time to train. Loading the pre trained weight will help get to desired accuracy quicker .

Describe the solution you'd like

An option in the config file that can load the pre trained weight and then resume training from there.

Additional context

No response

[blaz-r]

Hello. You can achieve this by adding resume_from_checkpoint somewhere in your config and providing a path to weights that you want to resume from.

This is not that well documented I thing. So I'll note it down as something that should be improved in v1 docs.

[samet-akcay]

Yes, let's keep this issue open as a doc improvement.

[Shakib-IO]

Hi @samet-akcay @blaz-r
I want to work on this.

[samet-akcay]

@Shakib-IO, sure go ahead. Thanks for your interest!

[Shakib-IO]

Hi @blaz-r
I'm unsure where I should insert resume_from_checkpoint in the documentation. Any insight you could provide would be greatly appreciated. I found a relevant source here.

Thanks!

[blaz-r]

Hi @Shakib-IO resume_from_checkpoint was part of config before the v1. With v1, this is a bit different. You can pass the checkpoint to fit directly:

anomalib/src/anomalib/engine/engine.py

Lines 462 to 497 in 34b3a90

	def fit(
	self,
	model: AnomalyModule,
	train_dataloaders: TRAIN_DATALOADERS | None = None,
	val_dataloaders: EVAL_DATALOADERS | None = None,
	datamodule: AnomalibDataModule | None = None,
	ckpt_path: str | Path | None = None,
	) -> None:
	"""Fit the model using the trainer.
	Args:
	model (AnomalyModule): Model to be trained.
	train_dataloaders (TRAIN_DATALOADERS | None, optional): Train dataloaders.
	Defaults to None.
	val_dataloaders (EVAL_DATALOADERS | None, optional): Validation dataloaders.
	Defaults to None.
	datamodule (AnomalibDataModule | None, optional): Lightning datamodule.
	If provided, dataloaders will be instantiated from this.
	Defaults to None.
	ckpt_path (str | None, optional): Checkpoint path. If provided, the model will be loaded from this path.
	Defaults to None.
	CLI Usage:
	1. you can pick a model, and you can run through the MVTec dataset.
	```python
	anomalib fit --model anomalib.models.Padim
	```
	2. Of course, you can override the various values with commands.
	```python
	anomalib fit --model anomalib.models.Padim --data <CONFIG | CLASS_PATH_OR_NAME> --trainer.max_epochs 3
	```
	4. If you have a ready configuration file, run it like this.
	```python
	anomalib fit --config <config_file_path>
	```
	"""

so passing --ckpt_path CKPT_PATH.
As for setting this via config, I am not sure how this exactly it's handled there.

[ashwinvaidya17]

@Shakib-IO are you still working on this?

[Shakib-IO]

No. @ashwinvaidya17