GitLab

* 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

* 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

Allow multi-tensor unscale to handle FP16 output, so it can also be used for copy-scatter. Rename some options.

Added kernel to support sync BN for channel last tensor

adjusted kernel config for better perf.
removed divergence in welford warp reduction.

* test passes

* notes

* Using C++-side flatten and unflatten functions

* Adding csrc

* Persistent synchronization event so it doesn't need to be created and destroyed each time

* Interop with parameter flattening in SSD

* Added deterministic option to imagenet main.py

* Adding options to split gradient averaging and allreduce in pure fp32

* Fixing allreduce_maybe_retain call

* Fixing allreduce_fallback

* Also sync active_i_buckets from rank 0

* Making retain_allreduce_buffers compatible with/orthogonal to delay_allreduce=True|False

* Correcting syntax error, now all seems to work with SSD

* Optional cpp extension build

* Add mixed precision adam optimizer (#59)

* Add FusedAdam Optimizer to Apex that places all the math into a cuda kernel.

* Added fixes to fused_adam to get it to work with network.

* wip work on python interface for adam with options

* fix dispatch for halfs, add python options to handle optional half gradients and params

* cleanup, get rid of grid-stride loop

* [syncBN]
  added syncBN in native pure python apex
  added fused cuda kernels used for sync BN. Using welford for mean/var
    optional installation using 'python setup.py install --cuda_ext'
  added unit test with side to side comparison between apex sync BN with
    PyTorch BN. Notice that for pytorch BN implementation, because of
    numerical issue for mean/var, the output will be slightly off.

* [syncBN PR]
  added fp16 support
  addressing review comments on:
    1. updating last pow 2
    2. look for import error when importing syncBN kernel

* [syncBN PR]
  added convert function to insert SyncBatchNorm
  refactored some kernel code

* fixing type issue (fp16/fp32/fp64)
added Kahan summation
editing unit test to use pytorch primitive ops with double, passing reasonable tests now

* updating tensor creation calls

* fixing the all_reduce contiguous tensor

* transposed all reduce results

* [syncBN]
support fp16 input & fp32 layer for apex fp16
partially fixing launch configs
enabling imagenet example to run with --sync_bn

* [syncBN PR]
Documentation added

* adjusting README

* adjusting again

* added some doc to imagenet example

* [syncBN]
  warp-level reduction
  bug fix: warp reduction logic updated. check for dummy element to avoid nan.
  improved launch config for better reduction kernels. Further improvements
would be to increase grid size.

* [syncBN]
  fixing undefined behavior in __shfl_down_sync from divergent threads in warp
reduction.
  changing at::native::empty to at::empty (upstream comments)

Co-authored-by: Michael Carilli <mcarilli@gmail.com>