Merge pull request #764 from mlcommons/scap3yvt-patch-1

Added documentation for extending optimizer submodule

GANDLF/optimizers/README.md

@@ -1,9 +1,12 @@
-# GANDLF Preprocessing
+# GANDLF Optimizers
 
 ## Adding a new algorithm
 
-- Define a new submodule under `GANDLF.optimizers`.
-- Ensure that the new algorithm is wrapped in a function which returns a scheduler, by following one of the examples in `GANDLF.optimizers.sgd`.
-    - If the new function is from a pre-defined package, put it under `GANDLF.optimizers.wrap_${package_name}.py`.
-- Add the algorithm's identifier to `GANDLF.optimizers.__init__.global_optimizer_dict` as appropriate.
-- Call the new algorithm from the config using the `optimizer` key.
+- For an optimizer defined in PyTorch [[ref](https://pytorch.org/docs/stable/optim.html#algorithms)], update the `GANDLF.optimizers.wrap_torch.py` submodule.
+- For a custom optimizer, create a new submodule called `GANDLF.optimizers.${awesome_optimizer}.py`. Ensure that it inherits from PyTorch's base optimizer class [[ref](https://pytorch.org/docs/stable/optim.html#base-class)]
+- If a new dependency needs to be used, update GaNDLF's [`setup.py`](https://github.com/mlcommons/GaNDLF/blob/master/setup.py) with the new requirement.
+  - Define a new submodule under `GANDLF.optimizers` as `GANDLF.optimizers.wrap_${package_name}.py`.
+  - Ensure that the new algorithm is wrapped in a function which returns an object with the PyTorch optimizer type. Use any of the optimizers in `GANDLF.optimizers.wrap_torch.py` as an example.
+- Add the algorithm's identifier to `GANDLF.optimizers.__init__.global_optimizer_dict` with an appropriate key.
+- Call the new algorithm from the config using the `optimizer` key.
+- [Update the tests!](https://mlcommons.github.io/GaNDLF/extending/#update-tests)https://mlcommons.github.io/GaNDLF/extending/#update-tests

GANDLF/schedulers/README.md

@@ -2,8 +2,11 @@
 
 ## Adding a new algorithm
 
-- Define a new submodule under `GANDLF.schedulers`.
-- Ensure that the new algorithm is wrapped in a function which returns a schedulers, by following one of the examples in `GANDLF.schedulers.triangle`.
-    - If the new function is from a pre-defined package, put it under `GANDLF.schedulers.wrap_${package_name}.py`.
-- Add the algorithm's identifier to `GANDLF.schedulers.__init__.global_schedulerss_dict` as appropriate.
-- Call the new algorithm from the config using the `scheduler` key.
+- For a scheduler defined in PyTorch [[ref](https://pytorch.org/docs/stable/optim.html#how-to-adjust-learning-rate)], update the `GANDLF.schedulers.wrap_torch.py` submodule.
+- For a custom scheduler, create a new submodule called `GANDLF.schedulers.${awesome_optimizer}.py`. Ensure that it inherits from PyTorch's base optimizer class [[ref](https://pytorch.org/docs/stable/optim.html#base-class)]. Follow the example of [`torch.optim.lr_scheduler.LinearLR`](https://pytorch.org/docs/stable/_modules/torch/optim/lr_scheduler.html#LinearLR) as an example.
+- If a new dependency needs to be used, update GaNDLF's [`setup.py`](https://github.com/mlcommons/GaNDLF/blob/master/setup.py) with the new requirement.
+  - Define a new submodule under `GANDLF.schedulers` as `GANDLF.schedulers.wrap_${package_name}.py`.
+  - Ensure that the new algorithm is wrapped in a function which returns an object with the PyTorch optimizer type. Use any of the optimizers in `GANDLF.schedulers.wrap_torch.py` as an example.
+- Add the algorithm's identifier to `GANDLF.schedulers.__init__.global_optimizer_dict` with an appropriate key.
+- Call the new algorithm from the config using the `scheduler` key.
+- [Update the tests!](https://mlcommons.github.io/GaNDLF/extending/#update-tests)https://mlcommons.github.io/GaNDLF/extending/#update-tests