Download Latest Version lightning-2.4.0.tar.gz (620.6 kB) Get Updates
Home / 2.4.0
Name Modified Size InfoDownloads / Week
Parent folder
lightning-2.4.0.tar.gz 2024-08-07 620.6 kB
0
pytorch-lightning-2.4.0.tar.gz 2024-08-07 625.3 kB
0
pytorch_lightning-2.4.0-py3-none-any.whl 2024-08-07 815.2 kB
0
lightning-2.4.0-py3-none-any.whl 2024-08-07 811.0 kB
0
lightning_fabric-2.4.0-py3-none-any.whl 2024-08-07 250.0 kB
0
lightning-fabric-2.4.0.tar.gz 2024-08-07 194.2 kB
0
Lightning v2.4 source code.tar.gz 2024-08-07 16.3 MB
0
Lightning v2.4 source code.zip 2024-08-07 17.1 MB
0
README.md 2024-08-07 6.8 kB
0
Totals: 9 Items   36.7 MB 0

Lightning AI :zap: is excited to announce the release of Lightning 2.4. This is mainly a compatibility upgrade for PyTorch 2.4 and Python 3.12, with a sprinkle of a few features and bug fixes.

Did you know? The Lightning philosophy extends beyond a boilerplate-free deep learning framework: We've been hard at work bringing you Lightning Studio. Code together, prototype, train, deploy, host AI web apps. All from your browser, with zero setup.

Changes

PyTorch Lightning

Added - Made saving non-distributed checkpoints fully atomic ([#20011](https://github.com/Lightning-AI/pytorch-lightning/pull/20011)) - Added `dump_stats` flag to `AdvancedProfiler` ([#19703](https://github.com/Lightning-AI/pytorch-lightning/issues/19703)) - Added a flag `verbose` to the `seed_everything()` function ([#20108](https://github.com/Lightning-AI/pytorch-lightning/pull/20108)) - Added support for PyTorch 2.4 ([#20010](https://github.com/Lightning-AI/pytorch-lightning/pull/20010)) - Added support for Python 3.12 ([20078](https://github.com/Lightning-AI/pytorch-lightning/pull/20078)) - The `TQDMProgressBar` now provides an option to retain prior training epoch bars ([#19578](https://github.com/Lightning-AI/pytorch-lightning/pull/19578)) - Added the count of modules in train and eval mode to the printed `ModelSummary` table ([#20159](https://github.com/Lightning-AI/pytorch-lightning/pull/20159))
Changed - Triggering KeyboardInterrupt (Ctrl+C) during `.fit()`, `.evaluate()`, `.test()` or `.predict()` now terminates all processes launched by the Trainer and exits the program ([#19976](https://github.com/Lightning-AI/pytorch-lightning/pull/19976)) - Changed the implementation of how seeds are chosen for dataloader workers when using `seed_everything(..., workers=True)` ([#20055](https://github.com/Lightning-AI/pytorch-lightning/pull/20055)) - NumPy is no longer a required dependency ([#20090](https://github.com/Lightning-AI/pytorch-lightning/issues/20090))
Removed - Removed support for PyTorch 2.1 ([#20009](https://github.com/Lightning-AI/lightning/pull/20009)) - Removed support for Python 3.8 ([#20071](https://github.com/Lightning-AI/lightning/pull/20071))
Fixed - Avoid LightningCLI saving hyperparameters with `class_path` and `init_args` since this would be a breaking change ([#20068](https://github.com/Lightning-AI/pytorch-lightning/pull/20068)) - Fixed an issue that would cause too many printouts of the seed info when using `seed_everything()` ([#20108](https://github.com/Lightning-AI/pytorch-lightning/pull/20108)) - Fixed `_LoggerConnector`'s `_ResultMetric` to move all registered keys to the device of the logged value if needed ([#19814](https://github.com/Lightning-AI/pytorch-lightning/issues/19814)) - Fixed `_optimizer_to_device` logic for special 'step' key in optimizer state causing performance regression ([#20019](https://github.com/Lightning-AI/lightning/pull/20019)) - Fixed parameter counts in `ModelSummary` when model has distributed parameters (DTensor) ([#20163](https://github.com/Lightning-AI/pytorch-lightning/pull/20163))

Lightning Fabric

Added - Made saving non-distributed checkpoints fully atomic ([#20011](https://github.com/Lightning-AI/pytorch-lightning/pull/20011)) - Added a flag `verbose` to the `seed_everything()` function ([#20108](https://github.com/Lightning-AI/pytorch-lightning/pull/20108)) - Added support for PyTorch 2.4 ([#20028](https://github.com/Lightning-AI/pytorch-lightning/pull/20028)) - Added support for Python 3.12 ([20078](https://github.com/Lightning-AI/pytorch-lightning/pull/20078))
Changed - Changed the implementation of how seeds are chosen for dataloader workers when using `seed_everything(..., workers=True)` ([#20055](https://github.com/Lightning-AI/pytorch-lightning/pull/20055)) - NumPy is no longer a required dependency ([#20090](https://github.com/Lightning-AI/pytorch-lightning/issues/20090))
Removed - Removed support for PyTorch 2.1 ([#20009](https://github.com/Lightning-AI/lightning/pull/20009)) - Removed support for Python 3.8 ([#20071](https://github.com/Lightning-AI/lightning/pull/20071))
Fixed - Fixed an attribute error when loading a checkpoint into a quantized model using the `_lazy_load()` function ([#20121](https://github.com/Lightning-AI/lightning/pull/20121)) - Fixed `_optimizer_to_device` logic for special 'step' key in optimizer state causing performance regression ([#20019](https://github.com/Lightning-AI/lightning/pull/20019))


Full commit list: 2.3.0 -> 2.4.0

Contributors

We thank all our contributors who submitted pull requests for features, bug fixes and documentation updates.

New Contributors

Did you know?

Chuck Norris can solve NP-hard problems in polynomial time. In fact, any problem is easy when Chuck Norris solves it.

Source: README.md, updated 2024-08-07