Fix issues with unwrapping modules with accelerate #963
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
There were a few issues at once that are now being fixed: - When unwrapping, pass keep_fp32_wrapper=True, or else the modules are not completely unwrapped and could not be pickled. - Only net.module_ was unwrapped, but there can be other modules and criteria. Those are now also unwrapped. - In some circumstances, unwrapping is undesired. For instance, when users want to continue training or benefit from AMP during inference. Therefore, an option was added to prevent automatic unwrapping: set unwrap_after_train=False.
Merged
BenjaminBossan
added a commit
that referenced
this pull request
May 17, 2023
Preparation for release of version 0.13.0 Release text: The new skorch release is here and it has some changes that will be exiting for some users. - First of all, you may have heard of the [PyTorch 2.0 release](https://pytorch.org/get-started/pytorch-2.0/), which includes the option to compile the PyTorch module for better runtime performance. This skorch release allows you to pass `compile=True` when initializing the net to enable compilation. - Support for training on multiple GPUs with the help of the [`accelerate`](https://huggingface.co/docs/accelerate/index) package has been improved by fixing some bugs and providing a dedicated [history class](https://skorch.readthedocs.io/en/latest/user/history.html#distributed-history). Our documentation contains more information on [what to consider when training on multiple GPUs](https://skorch.readthedocs.io/en/latest/user/huggingface.html#caution-when-using-a-multi-gpu-setup). - If you have ever been frustrated with your neural net not training properly, you know how hard it can be to discover the underlying issue. Using the new [`SkorchDoctor`](https://skorch.readthedocs.io/en/latest/helper.html#skorch.helper.SkorchDoctor) class will simplify the diagnosis of underlying issues. Take a look at the accompanying [notebook](https://nbviewer.org/github/skorch-dev/skorch/blob/master/notebooks/Skorch_Doctor.ipynb) Apart from that, a few bugs have been fixed and the included notebooks have been updated to properly install requirements on Google Colab. We are grateful for external contributors, many thanks to: - Kshiteej K (kshitij12345) - Muhammad Abdullah (abdulasiraj) - Royi (RoyiAvital) - Sawradip Saha (sawradip) - y10ab1 (y10ab1) Find below the list of all changes since v0.12.1 below: ### Added - Add support for compiled PyTorch modules using the `torch.compile` function, introduced in [PyTorch 2.0 release](https://pytorch.org/get-started/pytorch-2.0/), which can greatly improve performance on new GPU architectures; to use it, initialize your net with the `compile=True` argument, further compilation arguments can be specified using the dunder notation, e.g. `compile__dynamic=True` - Add a class [`DistributedHistory`](https://skorch.readthedocs.io/en/latest/history.html#skorch.history.DistributedHistory) which should be used when training in a multi GPU setting (#955) - `SkorchDoctor`: A helper class that assists in understanding and debugging the neural net training, see [this notebook](https://nbviewer.org/github/skorch-dev/skorch/blob/master/notebooks/Skorch_Doctor.ipynb) (#912) - When using `AccelerateMixin`, it is now possible to prevent unwrapping of the modules by setting `unwrap_after_train=True` (#963) ### Fixed - Fixed install command to work with recent changes in Google Colab (#928) - Fixed a couple of bugs related to using non-default modules and criteria (#927) - Fixed a bug when using `AccelerateMixin` in a multi-GPU setup (#947) - `_get_param_names` returns a list instead of a generator so that subsequent error messages return useful information instead of a generator `repr` string (#925) - Fixed a bug that caused modules to not be sufficiently unwrapped at the end of training when using `AccelerateMixin`, which could prevent them from being pickleable (#963)
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
There were a few issues at once that are now being fixed:
keep_fp32_wrapper=True, or else the modules are not completely unwrapped and could not be pickled. The test that assumed pickling would fail is now passing.net.module_was unwrapped, but there can be other modules and criteria. Those are now also unwrapped.unwrap_after_train=False.Note: This first fix could also work for bf16, but I cannot test it on my machine, so the test still assumes a
PicklingErrorto be raised.