strisunshinewentingwang/pytorch

Pull Requests Commits

Add assertion to make sure init op is always fp16 compatible in fp16 training

chocjy committed 5 years ago

deadf3ba

Generate only one Type class per backend (#19295)

Roy Li committed 5 years ago

689dd800

Make complex its own backend (#19275)

Roy Li committed 5 years ago

189f3060

Add ScalarType argument to Type::options() (#19270)

Roy Li committed 5 years ago

ab78449e

Generate cases for all ScalarTypes in Type functions that call to TH (#19230)

Roy Li committed 5 years ago

a044ba1a

Fix clang-format. (#19550)

Mikhail Zolotukhin committed 5 years ago

868933a4

Fix some typos in jit README

mrshenli committed 5 years ago

1bd5f2c1

Match JIT signature with triu_indices / tril_indices. (#19484)

gchanan committed 5 years ago

5afc2747

Make one_hot non-differentiable. (#19524)

gchanan committed 5 years ago

9eb48e1b

Remove 'BoolTensor', 'IndexTensor' from frontend specifications. (#19523)

gchanan committed 5 years ago

67330374

Have _embedding_bag_dense_backward match JIT signature. (#19522)

gchanan committed 5 years ago

39446015

Have embedding_dense_backward match JIT signature. (#19521)

gchanan committed 5 years ago

e3523979

Update mkldnn-bridge to fix crash issue in DNNLOWP dequantize op (#19159)

gujinghui committed 5 years ago

638ffac3

Hook up non_differentiability in derivatives.yaml when no autograd function is generated. (#19520)

gchanan committed 5 years ago

83373e77

Move non_differentiable_arg_names from autograd functions to differentiability_info. (#19519)

gchanan committed 5 years ago

8868a4f2

Move cuFFT plan cache note outside Best Practices (#19538)

ssnl committed 5 years ago

6d307db5

Revert D14689639: [pytorch] Allow passing lists as trace inputs.

suo committed 5 years ago

26f1c6d4

Improve optimizations for DNNLOWP support on MKL-DNN (#18843)

gujinghui committed 5 years ago

c96c91da

Make Observer class as template Quant class for QuantConfig (#19418)

nishantpdce committed 5 years ago

fe87327c

Support compilation on gcc-7.4.0 (#19470)

Sam Leeman-Munk committed 5 years ago

9f4f7e16

Improve embedding_bag add kernel (#19329)

James Reed committed 5 years ago

d17c22d0

Make finding unused model parameters optional (#19515)

pietern committed 5 years ago

6325b6e4

Disallow std::vector arguments (#19511)

smessmer committed 5 years ago

63e2833c

Drop instead of pop (#19503)

smessmer committed 5 years ago

1ac14b03

Add minimalistic implementation of subgraph matcher. (#19322)

Mikhail Zolotukhin committed 5 years ago

9818c7cb

Fix op benchmarks error in OSS environment (#19518)

mingzhe09088 committed 5 years ago

26f12af5

fix AI-PEP path error (#19514)

mingzhe09088 committed 5 years ago

5da7b74d

First step at container aliasing (#18710)

eellison committed 5 years ago

a421f882

Fix relu bug for empty tensor (#19451)

xiaomengy committed 5 years ago

f5fe7aa0

Allow passing lists as trace inputs.

efaust committed 5 years ago

2d4875b8

Newer Older