Nvidia-physicsnemo

Fixed

- Fix bug for `ConvFullyConnectedArch`.
- Updating `Activation.SILU` test to conform with updated nvFuser kernel generation.

Added

- Added instructions for docker build on ARM architecture.
- Added domain decomposition examples using X-PINN and FB-PINN style.

Changed

- Integrated the network architecture layers into PhysicsNeMo-Core.

Fixed

- Fixed Gradient Aggregation bug.

Security

- Upgrade Pillow and Sympy to their latest versions.
- Upgrade Scikit-Learn version.

Dependencies

- Updated base PyTorch container to 23.10 and Optix version to 7.3.0

Added

- Example for reservoir modeling using PINO and FNO

Added

- Added a CHANGELOG.md

Removed

- Accompanying licenses (will provide in the PhysicsNeMo docker image).

Fixed

- Arch `from_config` bug for literal params.
- Fixed fused SiLU activation test.
- Update `np.bool` to `np.bool_`.
- Added a workaround fix for the CUDA graphs error in multi-node runs

Dependencies

- Updated the base container to latest PyTorch base container which is based on torch 2.0
- Container now supports CUDA 12, Python 3.10
- Updated symengine to >=0.10.0 and vtk to >=9.2.6

Added

- Added ReGen score-based data assimilation example

Changed

Deprecated

Removed

Fixed

Security

Dependencies

Added

- Added version checks to ensure compatibility with older PyTorch for distributed
utilities and ShardTensor

Fixed

- `EntryPoint` error that occured during physicsnemo checkpoint loading