strisunshinewentingwang/pytorch

Pull Requests Commits

Resolve rendezvous race condition

mingzhe09088 committed 4 years ago

cccf5e70

[jit] fix the NoneType param/buffer hack (#32745)

suo committed 4 years ago

3552be10

enable empty batch for all flavor of convolutions (#32709)

ngimel committed 4 years ago

2e359ef8

[pytorch][embeddingbag_8bit] Add include_last_offset option to Fused 8bit EmbeddingBag and parallelize the op (#32683)

jianyuh committed 4 years ago

a840afbe

Logspace fixes (#32744)

Enealor committed 4 years ago

b565d9b3

[quantization] Remove incorrect fp16 dynamic linear/relu op

James Reed committed 4 years ago

fc2ff791

Remove -Werror from test/cpp_extensions/setup.py (#32704)

pbelevich committed 4 years ago

9357b911

Fix ivalue_inl.h:353:29: warning: comparison of unsigned expression >= 0 is always true (#32778)

zasdfgbnm committed 4 years ago

8b187e8f

Revert D19597036: More code fakefp16 mapping unification

ezyang committed 4 years ago

c47c78d0

Refreshing numel on a stride update is pointless. (#32116)

ezyang committed 4 years ago

3ee6673e

Delete resize_dim() (#32114)

ezyang committed 4 years ago

8c6f52ac

Expunge last two sites of resize_dim (#32112)

ezyang committed 4 years ago

b371eab8

Delete copy/move constructors on these RAII guards. (#32727)

ezyang committed 4 years ago

c7df28a2

Add missing C10_API to dispatch key TLS setter/getters (#32557)

ezyang committed 4 years ago

5ffa1efa

Improve documentation in dispatcher; remove unnecessary optional (#32533)

ezyang committed 4 years ago

3b479228

Port BCELoss to ATen to increase accuracy (#31365)

kurtamohler committed 4 years ago

8cb05e72

Make VC++ version a parametrizable option for Windows CI. (#32043)

ezyang committed 4 years ago

50d82f51

Fix TensorProtosDBInput AttributeError (#32274)

deepali-c committed 4 years ago

e84f9d9d

Randomize xla port (#32718)

Ailing Zhang committed 4 years ago

8693164a

clean up GIL usuage (#32748)

zhaojuanmao committed 4 years ago

b5d8982a

[android] fbjni DoNotStrip annotation for oss native methods (#32567)

IvanKobzarev committed 4 years ago

eab99ab0

Improved speed of frobenous norm for non-complex dtype (#30871)

dylanbespalko committed 4 years ago

2471ddc9

Custom RNG DispatchKey (#32325)

pbelevich committed 4 years ago

b1c85dd9

More code fakefp16 mapping unification

Yinghai Lu committed 4 years ago

642c9ef9

Deduplication of type casting codes (#32730)

zasdfgbnm committed 4 years ago

d119de8a

apply linter to rpc test files (#32659)

rohan-varma committed 4 years ago

cbb744f0

Fix crash of SobolEngine if default tensor type is cuda (#32496)

Balandat committed 4 years ago

8bc889e5

added exception args to the returned error message (#32693)

Coderx7 committed 4 years ago

c7bf4d22

Get rid of some unused THGenerate*Type defines. (#32657)

gchanan committed 4 years ago

c35ca84e

Make sure temporary vectors are properly initialized in avx2 code (#32722)

albanD committed 4 years ago

594cadeb

Newer Older