GitLab

* fix URLs

* Update distributed.py

* remove unused tens tensor in example/imagenet/main_amp.py

* remove unused tens tensor in deprecated examples and tests/L1

* change .type().ScalarType() to .scalar_type() + at::ScalarType::X to at::kX

* revert scalar_type() to type() for AT_DISPATCH_FLOATING_TYPES_AND_HALF

* revert scalar_type() to type() in AT_DISPATCH_FLOATING_TYPES

* revert scalar_type() to type() for AT_DISPATCH_FLOATING_TYPES_AND_HALF in welford.cu

* revert scalar_type() to type() in layer_norm_cuda_kernel.cu

* revert at::kType  to at::ScalarType::Type

* use DISPATCH_FLOAT_AND_HALF to get rid of warnings

* add dispatch mechanisms for double+float and double+float+half

The main use of these functions (e.g.: `torch.{conv*, prelu}`) is via their `torch.nn`
wrapping layers.

The `torch.nn` layers are what contain the weights and call into these lower level
functions with the weights as a parameter in their `forward()` method.

The `torch.conv*` functions are already in the `FP16_CASTS` list due to amp's philosophy of
casting the parameters rather than the model/layer weights.

Conceptually `torch.prelu` is the same as the `torch.conv*` case, where its weight parameter
is passed in from its wrapper layer `torch.nn.PReLU`.

quick fix: make FusedLayerNorm compatible with cpu

Run interpolation in fp32 because it's faster

* Refactor to allow more flexible treatment of multiple optimizers/models/losses

* Adding _process_optimizers.py

* Created L0 tests (now passing).

* fix: minor print typo (#234)

* make L1 results easier to read

* L0 multiple model/optimizer/loss test fleshed out

* Adding test that master params remain synced across distributed processes

* Docstring updates

* Docstring updates

Fix a typo in optimized_sync_batchnorm_kernel.py

in line 54, running_var should be running_variance..

Conditionally run bmm functions in fp16 based on cuda version