Nick Gibson 2f21dfb541 [TensorExpr] Eager reduction initialization & removal from ReduceOp (#38585) ... Summary: This PR removes the deferred initializer field from ReduceOp in favour of eagerly initializing buffers when they are created (either in the constructor of `LoopNest`, or in `rfactor()`). This allows a pretty good simplification of reduction logic, removing almost all of the reduction expander and the ReduceInitCleaner & unpopular NoOp node added in the last fix. Eager initialization is better for us anyway because it allows more opportunities to transform the initialization loop. Added a few more tests, testReduceOverSplitWithTail failed before this change due to a bug in splitWithTail which now can't happen. Pull Request resolved: https://github.com/pytorch/pytorch/pull/38585 Differential Revision: D21621551 Pulled By: nickgg fbshipit-source-id: 378137e5723b4a6d6e390239efb12adce22a8215		2020-05-18 15:56:43 -07:00
..
api	[aten] Pass std::function<> to thread_pool by value, instead of const ref. (#37681)	2020-05-05 08:41:38 -07:00
common
dist_autograd	Fix/relax CMake linter rules (#35574)	2020-03-27 16:52:33 -07:00
jit	[torchbind] Better error message when missing init. (#37474)	2020-05-13 17:38:31 -07:00
rpc	[TensorPipe] Use the new multi-payload message API (#37919)	2020-05-07 02:52:30 -07:00
tensorexpr	[TensorExpr] Eager reduction initialization & removal from ReduceOp (#38585)	2020-05-18 15:56:43 -07:00
__init__.py