strisunshinewentingwang/pytorch

Pull Requests Commits

Specify if mismatch is input or output in export (#107145)

ezyang committed 1 year ago

10ce16be

Use expect_true to make split with unbacked sizes work. (#106788)

ezyang committed 1 year ago

5673c087

Add expect_true for irrefutable guards (#106720)

ezyang committed 1 year ago

e1ee10e6

[ptd] make multithreaded pg wait for readiness before the 1st collective (#106954)

xunnanxu committed 1 year ago

388ba7e5

[ONNX] Fix diagnostic log and add unittest (#107158)

BowenBao committed 1 year ago

e9cb7179

[ONNX] Public diagnostic options for 'dynamo_export' (#106741)

BowenBao committed 1 year ago

19a76290

[Reland] Add OnCompletion Hook to ProcessGroup (#106988) (#107233)

mrshenli committed 1 year ago

45128ab6

[BE] Add missing override to remove build warning spam (#107191)

soulitzer committed 1 year ago

d9dc4b2b

Add some support for detecting false aliasing in AOTAutograd (#106461)

bdhirsh committed 1 year ago

8c44cfef

AOTAutograd: allow input mutations on inputs that are non-contiguous (#106460)

bdhirsh committed 1 year ago

517ba2ad

Use cpuinfo to determine c10::ThreadPool thread number (#107010)

cyyever committed 1 year ago

ad047654

[ROCm] Enable hipsolver unit tests for batched linalg drivers (#106620)

alugorey committed 1 year ago

7fb543e3

[gtest-static-listing] Make decision for caffe2 targets (#107129)

fgasperij committed 1 year ago

ed078212

Revert "Add OnCompletion Hook to ProcessGroup (#106988)"

pytorchmergebot committed 1 year ago

fd214aa8

Enable torchgen for MTIA dispatch key (#107046)

egienvalue committed 1 year ago

2abcfc40

adding fused uint4x2_mixed_mm to inductor (#106516)

HDCharles committed 1 year ago

935f2dd0

[CUDNN][CUDNN V8 API] LRU Cache for cuDNN frontend `ExecutionPlan` (#104369)

eqy committed 1 year ago

4add06eb

[submodule] Pin fmtlib/fmt to 10.1.0 (#106672)

cyyever committed 1 year ago

b7431d81

[export] Refactor `constrain_as_value` and `constrain_as_size` (#106591)

tugsbayasgalan committed 1 year ago

20c5add1

[TP][DTensor Perf]Fix DTensor Spec hash (#107181)

fduwjj committed 1 year ago

d6c120d7

Remove type promotion workaround (#107202)

mlazos committed 1 year ago

2d841bcb

grad_mode decorators without paren (#107086)

andreasfloros committed 1 year ago

c9c90765

Add OnCompletion Hook to ProcessGroup (#106988)

mrshenli committed 1 year ago

ba1da47e

Support third-party devices to use the init_process_group method with… (#107113)

QSHLGZ committed 1 year ago

2624da63

CI upgradeapalooza `bionic`->`focal`, `gcc7`->`gcc9`, `clang7`->`clang10` (#105260)

malfet committed 1 year ago

574442ba

Introduce CUDA-only `_scaled_mm` op (#106844)

malfet committed 1 year ago

9440a8cb

Add `generate_opcheck_tests`, a PT2 crossref testing mechanism (#106903)

zou3519 committed 1 year ago

e4e9aa28

[PT-D][FSDP] Handle corner case of load with multi-backend PG (#107172)

rohan-varma committed 1 year ago

ddf36c82

Add distributed/c10d *.hpp files to lintrunner (#107160)

mrshenli committed 1 year ago

064d813f

[Easy] Make Work::retrieveOpType a const function (#107141)

mrshenli committed 1 year ago

facadc6c

Newer Older