Remove Framework Callbacks

nmichlo / disent

🧶 Modular VAE disentanglement framework for python built with PyTorch Lightning ▸ Including metrics and datasets ▸ With strongly supervised, weakly supervised and unsupervised methods ▸ Easily configured and run with Hydra config ▸ Inspired by disentanglement_lib

https://disent.michlo.dev

MIT License

122 stars 18 forks source link

Remove Framework Callbacks #9

Closed nmichlo closed 3 years ago

nmichlo commented 3 years ago

Pytorch-Lightning modules are supposed to be self-contained units that manage all hyper-parameters?

The frameworks currently take in two arguments that break this requirement.

make_optimiser_fn
make_model_fn

nmichlo commented 3 years ago

One possible solution might be to stop extending pytorch-lightning modulein base framework and rather leave this as an implementation detail for the researcher.

nmichlo commented 3 years ago

See: #19 for proposed changes

nmichlo commented 3 years ago

Fixed for optimiser in new development versions. See #19 for changes

Model params are still not tracked by default