This package provides decorators for memory utilization maximization with PyTorch and CUDA by starting with a maximum parameter size and applying successive halving until no more out-of-memory exception occurs.
? Getting Started
Assume you have a function for batched computation of nearest neighbors using brute-force distance calculation.
import torch def knn(x, y, batch_size, k: int = 3): return torch.cat( [ torch.cdist(x[start : start + batch_size], y).topk(k=k, dim=1, largest=False).indices for start in range(0, x.shape, batch_size) ], dim=0, )
torch_max_mem you can decorate this function to reduce the batch size until no more out-of-memory error occurs.
import torch from torch_max_mem import maximize_memory_utilization @maximize_memory_utilization(parameter_name="batch_size") def knn(x, y, batch_size, k: int = 3): return torch.cat( [ torch.cdist(x[start : start + batch_size], y).topk(k=k, dim=0, largest=False).indices for start in range(0, x.shape, batch_size) ], dim=0, )
In the code, you can now always pass the largest sensible batch size, e.g.,
x = torch.rand(100, 100, device="cuda") y = torch.rand(200, 100, device="cuda") knn(x, y, batch_size=x.shape)
The most recent release can be installed from PyPI with:
$ pip install torch_max_mem
The most recent code and data can be installed directly from GitHub with:
$ pip install git+https://github.com/mberr/torch-max-mem.git
To install in development mode, use the following:
$ git clone git+https://github.com/mberr/torch-max-mem.git $ cd torch-max-mem $ pip install -e .
Contributions, whether filing an issue, making a pull request, or forking, are appreciated. See CONTRIBUTING.md for more information on getting involved.
The code in this package is licensed under the MIT License.
?️ For Developers
See developer instrutions
The final section of the README is for if you want to get involved by making a code contribution.
After cloning the repository and installing
pip install tox, the unit tests in the
tests/ folder can be
run reproducibly with:
Additionally, these tests are automatically re-run with each commit in a GitHub Action.
? Building the Documentation
$ tox -e docs
? Making a Release
After installing the package in development mode and installing
pip install tox, the commands for making a new release are contained within the
tox.ini. Run the following from the shell:
$ tox -e finish
This script does the following:
- Uses Bump2Version to switch the version number in the
src/torch_max_mem/version.pyto not have the
- Packages the code in both a tar archive and a wheel
- Uploads to PyPI using
twine. Be sure to have a
.pypircfile configured to avoid the need for manual input at this step
- Push to GitHub. You’ll need to make a release going with the commit where the version was bumped.
- Bump the version to the next patch. If you made big changes and want to bump the version by minor, you can
tox -e bumpversion minorafter.