Andrzej Kotlowski 0885c58296 Add Bfloat16 scalar support to gloo backend (#113557) ... There was missing support for bfloat scalars. When I use gloo backend `torch.distributed.init_process_group(backend='gloo')` and run `torch.nn.parallel.DistributedDataParallel(model)` and _model_ has Bfloat16 features I receive following error: `RuntimeError: Invalid scalar type` This change fix this issue. c10::BFloat16 defines conversions from/to float, so calculations are made on float for bfloat. Pull Request resolved: https://github.com/pytorch/pytorch/pull/113557 Approved by: https://github.com/XilunWu, https://github.com/jgong5		2023-11-17 21:16:54 +00:00
..
aot_inductor	[AOTInductor] Rename model_runner to model_container_runner (#111324)	2023-11-16 19:14:22 +00:00
api	[Cmake] Check that gcc-9.4 or newer is used (#112858)	2023-11-06 17:19:53 +00:00
c10d	Add Bfloat16 scalar support to gloo backend (#113557)	2023-11-17 21:16:54 +00:00
common
dist_autograd	Remove unneeded linking in CMake targets (#109192)	2023-09-15 19:43:25 +00:00
jit	[c10] Move profiler clock to libc10 for timestamps (#111972)	2023-10-27 16:18:40 +00:00
lazy	test_lazy: skip HashTest.Scalar (#112747)	2023-11-16 01:22:58 +00:00
lite_interpreter_runtime	Fix typo under test directory (#111304)	2023-10-16 23:06:06 +00:00
monitor
profiler	[c10] Move profiler clock to libc10 for timestamps (#111972)	2023-10-27 16:18:40 +00:00
rpc	Fix typo under test directory (#111304)	2023-10-16 23:06:06 +00:00
tensorexpr	[ROCm] remove HCC references (#111975)	2023-10-26 02:39:10 +00:00
__init__.py