strisunshinewentingwang/pytorch

Pull Requests Commits

Revert D23503636: [pytorch][PR] [NNC] make inlining immediate (take 2) and fix bugs

Mikhail Zolotukhin committed 4 years ago

6474057c

[ONNX] Fix split export using slice (#43670)

neginraoof committed 4 years ago

539d029d

[FX] __str__ for GraphModule and Graph (#44166)

James Reed committed 4 years ago

af13faf1

[pytorch] remove code analyzer build folder between builds (#44148)

ann-ss committed 4 years ago

0e3cf6b8

Updates to SCCACHE for ROCm case (#44155)

pruthvistony committed 4 years ago

f38e7aee

replace the white list from default mappings (#41802)

VinodSKumar committed 4 years ago

2a1fc566

Control NCCL build parallelism via MAX_JOBS environment var (#44167)

malfet committed 4 years ago

4d431881

Limit MAX_JOBS to 18 for linux binary builds (#44168)

malfet committed 4 years ago

6aba58cf

Enable test_cublas_config_deterministic_error for windows (#42796)

guyang3532 committed 4 years ago

6cecf7ec

Register some backwards functions as operators (#44052)

zou3519 committed 4 years ago

9a5a7328

[BE] Use f-string in various Python functions (#44161)

malfet committed 4 years ago

0c01f136

[Codemod][FBSourceGoogleJavaFormatLinter] Daily `arc lint --take GOOGLEJAVAFORMAT`

generatedunixname89002005287564 committed 4 years ago

28b1360d

Optimize code path for adaptive_avg_pool2d when output size is (1, 1) (#43986)

xwang233 committed 4 years ago

f8f35fdd

[Codemod][FBSourceClangFormatLinter] Daily `arc lint --take CLANGFORMAT`

generatedunixname89002005287564 committed 4 years ago

ef28ee50

[te] Disable reductions by default (#44122)

bertmaher committed 4 years ago

98ad5ff4

[c2][cuda] small improvement to dedup adagrad by avoiding recompute of x_ij (#44173)

xianjiec committed 4 years ago

a37c199b

Add API for onnxifi with AOT Glow ONNX (#44021)

jfix71 committed 4 years ago

2f8a4334

[Message] Add what to do for missing operators.

iseeyuan committed 4 years ago

d2212568

Add tests against autograd precedence and multiple dispatch. (#44037)

Ailing Zhang committed 4 years ago

addfd7a9

Enable typechecks for torch.nn.quantized.modules.linear (#44154)

malfet committed 4 years ago

b60ffcdf

Enable CUDA 11 jobs for Windows nightly builds (#44086)

peterjc123 committed 4 years ago

538d3bd3

[quant] conv_transpose2d_prepack/conv_transpose2d_unpack (#40351)

z-a-f committed 4 years ago

69e38828

[android][jni] Support Tensor MemoryFormat in java wrappers (#40785)

IvanKobzarev committed 4 years ago

c40e3f9f

[NNC] make inlining immediate (take 2) and fix bugs (#43885)

nickgg committed 4 years ago

70aecd2a

[TVM] Support Fused8BitRowwiseQuantizedToFloat op (#44098)

Hao Lu committed 4 years ago

bc4a00c1

[TensorExpr] Fuser: rely on input types when checking whether a device is supported. (#44139)

Mikhail Zolotukhin committed 4 years ago

3105d8a9

fx qat: respect device affinity (#44115)

vkuzo committed 4 years ago

71510c60

[JIT] Add mypy type annotations for JIT (#43862)

Meghan Lele committed 4 years ago

7816d537

[jit] Better match behavior of loaded ScriptModules vs. freshly created ones (#43298)

suo committed 4 years ago

9dd8670d

[jit] fix segfault in attribute lookup on loaded ScriptModules (#43284)

suo committed 4 years ago

74f18476

Newer Older