What's Changed
- Add boilerplate code for Dataflux-Pytorch Lightning demo by @abhibyreddi in #57
- Refactor Dataflux simple demo loops and add retry flags by @MattIrv in #59
- Catch exception when loading arrays from raw bytes fails by @abhibyreddi in #38
- Implement data module for the pytorch lightning workload by @abhibyreddi in #60
- Update default retry config to match successful 1k-node benchmarks. by @MattIrv in #63
- Lightning text by @jdnurme in #64
- add limit_train_batches param by @jdnurme in #65
- Make it possible to deploy Pytorch Lightning image segmentation workload on a Ray cluster. by @abhibyreddi in #66
- Update demo loops to configure multiprocessing start method by @MattIrv in #68
- For mac and windows skip passing client storage to avoid pickling error in multiprocessing by @akansha1812 in #67
- Disable compose download when create and delete permissions are missing by @akansha1812 in #70
- Continuous test which installs gcs-torch-dataflux from PyPi and runs integration test. by @akansha1812 in #71
- Added lightning package to setup file and updated version for re… by @divrawal in #69
New Contributors
- @akansha1812 made their first contribution in #67
Full Changelog: v1.2.0...v1.3.0
Contributors
MattIrv, jdnurme, and 3 other contributors