FTorch: a library for coupling PyTorch models to Fortran

Version 1.2.0 - Online

All Changes: https://github.com/Cambridge-ICCS/FTorch/compare/v1.1.0...v1.2.0 and in CHANGELOG

What's Changed

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

Full Changelog: https://github.com/Cambridge-ICCS/FTorch/compare/v1.1.0...v1.2.0