strisunshinewentingwang/pytorch

Pull Requests Commits

Enable input pointer caching in XNNPACK integration. (#42840)

kimishpatel committed 4 years ago

b003f2cc

Change quantizer to account for input tensor's memory format. (#42178)

kimishpatel committed 4 years ago

b52e6d00

Reduce number of `prim::profile` (#43147)

Krovatkin committed 4 years ago

b1d31428

[ONNX] Export split_to_sequence as slice when output number is static (#42744)

BowenBao committed 4 years ago

8efa8983

[quant][graphmode][fx] Add support for general value ops (#43439)

jerryzh168 committed 4 years ago

ec9e6e07

Set CONSTEXPR_EXCEPT_WIN_CUDA as const while it is not constexpr (#43380)

mszhanyi committed 4 years ago

47e1b7a8

Revert D23223281: Add Enum TorchScript serialization and deserialization support

pbelevich committed 4 years ago

d94b10a8

centralize autograd dispatch key set (#43387)

Basil Hosmer committed 4 years ago

915fd1c8

[quant][graphmode][fx] Add support for general shape ops (#43438)

jerryzh168 committed 4 years ago

88b564ce

[quant][graphmode][fx] Add support for clamp (#43437)

jerryzh168 committed 4 years ago

192c4b00

Add prim::TypeCheck operation (#43026)

bzinodev committed 4 years ago

40c77f92

Fix bfloat16 erfinv get incorrect value problem for cpu path (#43399)

XiaobingSuper committed 4 years ago

98307a28

caffe2: expose CPUContext RandSeed for backwards compatibility with external RNG (#43239)

d4l3k committed 4 years ago

5e04bb2c

Call qnnpack's conv setup only if input pointer has changed. (#42008)

kimishpatel committed 4 years ago

fb12992b

Refactor qconv to reduce allocations. (#42007)

kimishpatel committed 4 years ago

04aa42a0

Simple caching allocator for CPU. (#42006)

kimishpatel committed 4 years ago

2a08566b

Allow Freezing of Module containing interface attribute (#41860)

bzinodev committed 4 years ago

abe878ce

[quant][graphmode][fx] Add support for instance_norm (#43377)

jerryzh168 committed 4 years ago

490d41aa

add support for optional int list with scalar fill (#43262)

Basil Hosmer committed 4 years ago

a5a6a3e6

Add Enum TorchScript serialization and deserialization support (#42963)

gmagogsfm committed 4 years ago

f269fb83

Workaround bugs in user side embedding meta info and better msgs (#43355)

Yinghai Lu committed 4 years ago

aa53b2d4

[quant][graphmode][fx] Add support for layer_norm (#43376)

jerryzh168 committed 4 years ago

aec917a4

[quant][graphmode][fx] Add support for elu (#43375)

jerryzh168 committed 4 years ago

089bb1a8

[quant][graphmode][fx] Add support for hardswish (#43374)

jerryzh168 committed 4 years ago

5a02c6b1

Mobile backward compatibility (#42413)

iseeyuan committed 4 years ago

93f1b5c8

[quant][graphmode][fx] Add support for mul and mul relu (#43373)

jerryzh168 committed 4 years ago

e96871ea

Revert D23252335: Refactor Vulkan context into its own files. Use RAII.

Mike Ruberry committed 4 years ago

6c772515

Fix enum constant printing and add FileCheck to all Enum tests (#42874)

gmagogsfm committed 4 years ago

8eb3de76

[quant][grapphmode][fx][test][refactor] Refactor quantized add test (#43372)

jerryzh168 committed 4 years ago

ff454cc4

[quant][graphmode][fx] Add support for batchnorm relu (#43335)

jerryzh168 committed 4 years ago

109ea59a

Newer Older