v1.2.0

Version 1.2.0 - Online

All Changes: v1.1.0...v1.2.0 and in CHANGELOG

What's Changed

• Allow for custom external gradients in torch_tensor_backward by @joewallwork in #528
• Update intel versions in CI by @jatkinson1000 in #567
• Add optimisers by @jatkinson1000 in #320
• Test out differentiating through a model by @joewallwork in #486
• Overhaul pt2ts by @joewallwork in #555
• Unittest ftorch_utils Python module by @joewallwork in #563
• Add optimizer deletion to the torch_delete interface by @jatkinson1000 in #572
• Update intel action in workflow and pin version. by @jatkinson1000 in #573
• Fix a number of broken links in the docs. by @jatkinson1000 in #574
• Fixes for Zizmor security alerts by @joewallwork in #575
• Update Windows install docs by @joewallwork in #578
• Allow pre-trained models with pt2ts by @joewallwork in #565
• CMake: decouple unit/integration test selection by @jatkinson1000 in #584
• Fortran training example by @joewallwork in #581
• CI fixes after merging online training by @joewallwork in #589
• Add Flang test workflow by @joewallwork in #595
• Ensure model output file is not None by @joewallwork in #608
• Fix CI workflow issues for GNU, macOS, and Flang by @joewallwork in #603
• fix(ci): switch Windows CPU CI to Ninja and fix Intel compiler environment by @jatkinson1000 with @Copilot in #611
• Implement MSE loss function by @joewallwork in #591
• Implement torch_model_save by @joewallwork in #587
• Implement cross entropy loss by @joewallwork in #593
• Add some clarification docs to Cross Entropy Loss. by @jatkinson1000 in #612
• Check for consistent device types when applying overloaded operators by @joewallwork in #614
• Add installation for ftorch_loss.mod by @joewallwork in #616
• Release v1.2.0 by @jatkinson1000 in #615

Full Changelog: v1.1.0...v1.2.0