strisunshinewentingwang/pytorch

Pull Requests Commits

Fixes cat backward formula to return correct gradient values for R -> C case (#51681)

anjali411 committed 3 years ago

bd3ae117

[quant] Support 2 dim input in quantized batchnorm 1d (#51597)

jerryzh168 committed 3 years ago

d8742eee

Fix caffee2 for LLVM trunk

htyu committed 3 years ago

5d123ecf

mem-efficient learnable fake quantization (#49315)

haichuan-fb committed 3 years ago

0c60922f

[FX] Move examples to pytorch/examples (#51686)

James Reed committed 3 years ago

7918f37e

Extend subgraph_rewriter logic (#51532)

Ansley Ussery committed 3 years ago

f2c4deab

Type-annotate tools/generate_torch_version (#51637)

malfet committed 3 years ago

627ec8ba

[optim] make functional api be private (#51316) (#51665)

vincentqb committed 3 years ago

50d903f1

Beef up {jacobian, hessian} vectorize docs; eliminate a warning (#51638)

zou3519 committed 3 years ago

45e5562f

Revert D25074763: [WIP] Update foreach APIs to use scalar lists

ngimel committed 3 years ago

443a431a

Revert D25502940: Refactor ForeachUnaryOps.cu

ngimel committed 3 years ago

d1bc1ab8

Updates linalg documentation per feature review process (#51620)

Mike Ruberry committed 3 years ago

16cfe970

move ZipDataset to Zip DataPipe (#51599)

lixinyu committed 3 years ago

1ee0c42d

Autograd doc note fix (#51661)

anjali411 committed 3 years ago

34d4d799

[Static Runtime] Native stack for contiguous inputs (#50863)

Marat Subkhankulov committed 3 years ago

0d9ca21d

Replace AT_ASSERTM in ATen/core (#51579)

xuhdev committed 3 years ago

fe67438f

Replace all AT_ASSERTM in ATen/native (#51147)

xuhdev committed 3 years ago

c60dacd4

[quant][graphmode][fx] Enable inception_v3 and googlenet static quant test (#51402)

jerryzh168 committed 3 years ago

f38e1d2d

Use new TensorPipe functions to create channels (#51550)

lw committed 3 years ago

8e53bf01

Use new TensorPipe functions to create transports (#51549)

lw committed 3 years ago

56ef24bc

Removed typographical error from tech docs (#51286)

reisner-helene committed 3 years ago

47557b95

Add support for generating faithful at::cpu signatures (#51499)

ezyang committed 3 years ago

333a0c8b

Add api.structured; switch structured kernels to use const Tensor& everywhere (#51490)

ezyang committed 3 years ago

81c7c3ba

Relax type signature for tools.codegen.api.translate (#51477)

ezyang committed 3 years ago

648cdb7d

[Gradient Compression] Replace torch.sqrt(torch.sum(col ** 2)) by torch.norm() (#51629)

Yi Wang committed 3 years ago

43df03de

replace silufp16 with cubic interpolation (#51645)

Hector Yuen committed 3 years ago

00675292

Enable FLOPS Computation for Experimental Kineto Profiler (#51503)

xuzhao9 committed 3 years ago

cae43798

[Gloo] Use TORCH_CHECK for ensuring tag is nonnegative (#51370)

osalpekar committed 3 years ago

3361d365

Prevent CUDAFuture from using uninitialized device index (#51505)

mrshenli committed 3 years ago

a3f2fe0d

fix misspelling in swa_utils.pyi (#51608)

Jasha10 committed 3 years ago

a651696a

Newer Older