strisunshinewentingwang/pytorch

Pull Requests Commits

Revert "Fixed einsum compatibility/performance issues (#46398)" (#47821)

gchanan committed 3 years ago

b6cb2caa

[vulkan] Enable prepacked addmm/mm for linear layers (#47815)

SS-JIA committed 3 years ago

cfe3defd

Port bmm and baddbmm from TH to ATen (#42553)

anjali411 committed 3 years ago

e1ee3bfc

[c10d] switch ProcessGroup to be managed by intrusive_ptr (#47343)

wanchaol committed 3 years ago

553ccccc

skip test_all_reduce_sum_cuda_async test case for ROCM (#47630)

KyleCZH committed 3 years ago

859e0543

[ROCm] add skipCUDAIfRocm to test_lingalg test_norm_fro_2_equivalence_old (#47809)

jeffdaily committed 3 years ago

2df56001

Spurious numpy writable warning (#47271)

ArtistBanda committed 3 years ago

2907447c

torch.dropout: fix non-contiguous layout input (#47552)

kshitij12345 committed 3 years ago

4b25d83e

[reland][c10d] switch ProcessGroupNCCL:Options to be managed by intrusive_ptr (#47807)

wanchaol committed 3 years ago

a02baa0c

[reland] [c10d] switch Store to be managed by intrusive_ptr (#47808)

wanchaol committed 3 years ago

665ac2f7

[reland][c10d] switch ProcessGroup::Work to be managed by intrusive_ptr (#47806)

wanchaol committed 3 years ago

70ae5685

[quant] Add support for 2D indices for quantized embedding operators (#47766)

supriyar committed 3 years ago

89b371bc

[quant][graphmode][fx][refactor] insert_observer (#47782)

jerryzh168 committed 3 years ago

47386722

[quant][refactor] factor out get_combined_dict function (#47781)

jerryzh168 committed 3 years ago

dd77d5a1

add cost_aware_partition (#47673)

scottxu0730 committed 3 years ago

b46787d6

Look in named-buffers of module for tensors (#47641)

mrzzd committed 3 years ago

c5834b6a

Refactor DDP uneven inputs control flags (#47394)

rohan-varma committed 3 years ago

c9f6e70c

Workaround PyTorch debug build crash using old GCC (#47805)

malfet committed 3 years ago

e8a73fbf

Added CUDA support for complex input for torch.triangular_solve (#46916)

IvanYashchuk committed 3 years ago

52ec8b93

Free original weight after prepacking in XNNPACK based op (#46541)

Akshit Khurana committed 3 years ago

a0c4aae3

Mark overriden Tensor method `override` (#47198)

Akshit Khurana committed 3 years ago

545f624a

Properly serialize types that only appear at function input (#47775)

suo committed 3 years ago

d4fa84bf

[Docs] Minor doc fixes for init_process_group (#47644)

osalpekar committed 3 years ago

32b4b512

[PyTorch] Avoid atomic refcounting in intrusive_ptr::make (#47100)

swolchok committed 3 years ago

0c54ea50

Automated submodule update: FBGEMM (#47605)

facebook-github-bot committed 3 years ago

f2b7c387

Add instruction on how to handle the potential linker error on Linux (#47593)

skyline75489 committed 3 years ago

fcd44ce6

Improve error messages for operator registration API (#47636)

smessmer committed 3 years ago

7864ae9f

Batching rule for torch.squeeze(tensor) (#47632)

zou3519 committed 3 years ago

05a76ed7

Fix transpose batching rule (#47628)

zou3519 committed 3 years ago

df887936

Make kwargs argument optional in _batched_grad_test (#47625)

zou3519 committed 3 years ago

f6ff6478

Newer Older