strisunshinewentingwang/pytorch

Pull Requests Commits

[nnc] Tweak log_nnc_sleef so vectorization kicks in (#51491)

bertmaher committed 3 years ago

a23e82df

Out variant for embedding_bag_4bit_rowwise_offsets (#51324)

Marat Subkhankulov committed 3 years ago

5b0a6482

port `index_fill_` from TH to ATen. (#50578)

nikitaved committed 3 years ago

b198cf4f

Hashing logic for c10::complex (#51441)

anjali411 committed 3 years ago

09bc5879

[doc] Deprecate torch.cholesky in favor of torch.linalg.cholesky (#51460)

heitorschueroff committed 3 years ago

8fa328f8

[doc] Fix linalg.cholesky doc consistency issues (#51459)

heitorschueroff committed 3 years ago

8583f7cb

[Gradient Compression] Allow BatchedPowerSGD to run vanilla allreduce for the first K iterations (#51270)

Yi Wang committed 3 years ago

c0807803

add git submodule troubleshoot to CONTRIBUTING.md (#51458)

Rong Rong (AI Infra) committed 3 years ago

718e4b11

[NNC] Generate C++ code for Allocate and Free (#51070)

Cheng Chang committed 3 years ago

109bc104

Add sgn to torch.rst so that it appears in the built docs (#51479)

anjali411 committed 3 years ago

642afcb1

[PyTorch] Outline OperatorEntry::assertSignatureIsCorrect fail path (#51269)

swolchok committed 3 years ago

d1ddc5d6

[PyTorch] check isValidUnboxed() in the dispatcher (#51247)

swolchok committed 3 years ago

9877777f

[PyTorch] Pass TensorOptions by value (#51165)

swolchok committed 3 years ago

4495b49f

[PyTorch] Add C10_ALWAYS_INLINE to critical dispatcher paths (#51245)

swolchok committed 3 years ago

341c76dc

[PyTorch] Refactor Dispatcher to inline less code in fast path (#51163)

swolchok committed 3 years ago

673687e7

[Pytorch Mobile] Expose _export_operator_list to python (#51312)

JacobSzwejbka committed 3 years ago

ec611aca

[WIP][FX] Add Interpreter and Transformer (#50420)

James Reed committed 3 years ago

609f76f2

[Resubmission][Gradient Compression] Refactor default_hooks.py and powerSGD_hook.py by creating a util function that make a vanilla allreduce future (#51400)

Yi Wang committed 3 years ago

0831984e

[PyTorch] Devirtualize TensorImpl::has_storage (#51049)

swolchok committed 3 years ago

6c242967

[PyTorch] Devirtualize TensorImpl::storage_offset (#51048)

swolchok committed 3 years ago

765062c0

[testing] Add OpInfo for ceil and floor (#51198)

kshitij12345 committed 3 years ago

50fa415a

[SobolEngine] Update direction numbers to 21201 dims (#49710)

Balandat committed 3 years ago

449098c2

Fix pickling for Tensor subclasses (redo) (#47732)

hameerabbasi committed 3 years ago

b1907f5e

Support complex number list in JIT (#51145)

anjali411 committed 3 years ago

508bab43

Fixes docs (#51439)

Mike Ruberry committed 3 years ago

40c0fffb

[fbgemm_gpu] Use the latest philox_cuda_state API for stochastic rounding (#51004)

jianyuh committed 3 years ago

d1dcd5f2

fixing index clamping for upsample nearest kernel backward (#51240)

jjsjann123 committed 3 years ago

0e1c5cb3

[1.8] Add additional tests for object-based APIs (#51341)

rohan-varma committed 3 years ago

9cf62a4b

[Collective APIs] Make python object collective API args consistent (#50625)

rohan-varma committed 3 years ago

c2556281

Create op benchmark for stack (#51263)

Marat Subkhankulov committed 3 years ago

721ba97e

Newer Older