GitLab

remove redundant linear layer class definition

add fuse_gradient_accumulation attribute to weights for simple targetting

reflect feedback and clean up the codes

arg change

Adds some backward compatibility code so old inference code still works.

- Accumulate encoder hidden state gradient to handle skip connection
- Correctly compute the number of layers in encoder / decoder for T5 model
- Ensure e weights are initialized the same way in embeddings
- Synchronize embedding gradients across encoder and decoder for T5 model
- Support for checkpoint loading and saving

Also includes following changes for inter-layer model-parallel implementation:
- Refactoring of model implementations
- Training loop changes to support inter-layer communication using `ring_exchange`
- New groups for inter-layer communication
- Checkpoint changes
- Command line arguments

Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com>

Signed-off-by: Boris Fomitchev <bfomitchev@nvidia.com>