strisunshinewentingwang/pytorch

Pull Requests Commits

zheng-xq committed 4 years ago

afcf991d

Support fp16 output in OnnxifiOp (#38846)

Yinghai Lu committed 4 years ago

e9902358

Perf optimization for conv and gemm kernels. (#37626)

kimishpatel committed 4 years ago

65e8fe18

convbn fusion: add backwards compatibility support (#38820)

vkuzo committed 4 years ago

0b2a8615

qat syncbn: add test coverage (#38738)

vkuzo committed 4 years ago

4d5d9c04

[quant][graphmode] Different rule for add/add_/mul/mul_ (#38667)

jerryzh168 committed 4 years ago

a8d8fc55

Use union to cast between incompatible function pointers (#38842)

malfet committed 4 years ago

57d6e19d

Updating submodules

svcscm committed 4 years ago

6cf5c71b

Revert "Revert D21593870: Kill AT_DISPATCH_ALL_TYPES_AND_C10_COMPLEX_AND2" (#38814)

zasdfgbnm committed 4 years ago

48116ac8

port `scatter_add` to ATen (CUDA) (#38262)

nikitaved committed 4 years ago

f80df4ca

Add HIP to the memory profiler device list (#38795)

Ilia Cherniavskii committed 4 years ago

83fa3f1c

Replace import cpp_benchmark with `torch.utils.cpp_benchmark` (#38832)

malfet committed 4 years ago

c02e7c46

[bug] Binomial distribution BTRS algorithm has small chance of returning -1 (#38456)

jamestwebber committed 4 years ago

9c88b23f

Fix broken windows build due per channel quantization stack land. (#38828)

kimishpatel committed 4 years ago

267a8da1

Fast path for contiguous tensor (#38732)

Ailing Zhang committed 4 years ago

a7a69ad1

fix the device inconsistency for import convert_sync_batchnorm (#38729)

jjsjann123 committed 4 years ago

5b8a79ab

Back out "[RPC] [Minor] RPC entry point cleanup"

mingzhe09088 committed 4 years ago

6736a76c

[CircleCI] Add Python3.8-gcc9 config (#38747)

malfet committed 4 years ago

604533dd

Added per channel separate test cases for fc and deconv tests. (#37624)

kimishpatel committed 4 years ago

5b1814e4

Changes to enable per channel support on dynamic linear. (#37623)

kimishpatel committed 4 years ago

0a554aee

Enabled per channel quantized static linear/conv (#37622)

kimishpatel committed 4 years ago

b8eae1e3

Added per channel kernels for depthwise conv. (#37621)

kimishpatel committed 4 years ago

1c9a110b

Changes to enable per channel requant. (#37620)

kimishpatel committed 4 years ago

1f16d4ce

Enable per channel zero point. (#37619)

kimishpatel committed 4 years ago

622f5b68

Interface changes to enable per channel quant. (#37618)

kimishpatel committed 4 years ago

f1991ca8

Revert D21593870: Kill AT_DISPATCH_ALL_TYPES_AND_C10_COMPLEX_AND2

malfet committed 4 years ago

96d7defb

support complex types for tanh_cuda and tanh_backward_cuda (#38786)

kshitij12345 committed 4 years ago

3b254acd

Remove deprecated cuDNN API from caffe2 (#38680)

zasdfgbnm committed 4 years ago

51b25218

Enforce tensorboard minimum version as 1.15 (#35952)

peterbell10 committed 4 years ago

90400f48

Kill AT_DISPATCH_ALL_TYPES_AND_C10_COMPLEX_AND2 (#38459)

zasdfgbnm committed 4 years ago

4b248393

Older