strisunshinewentingwang/pytorch

Pull Requests Commits

[inductor][fx passes]batch linear in pre grad (#107759)

jackiexu1992 committed 1 year ago

9bda8f1e

Move `Constraint` class to torch.export() to avoid circular dependency in _dynamo package (#107750)

gmagogsfm committed 1 year ago

f8119f8b

[export] Serialize cond submodules (#107818)

angelayi committed 1 year ago

7bab98f1

[Inductor] Add new fused_attention pattern matcher (#107578)

Valentine233 committed 1 year ago

a5601355

Handle empty lists properly (#107803)

voznesenskym committed 1 year ago

9b2d43df

[DeviceMesh] init_device_mesh dosctring update to include one d mesh initialization (#107805)

wz337 committed 1 year ago

d707724a

Remove run torchbench. Torchbench runs are now part of the dynamo ci. (#107826)

xuzhao9 committed 1 year ago

26ae4883

Bump scipy from 1.9.0 to 1.10.1 in /.github/requirements (#104763)

dependabot[bot] committed 1 year ago

Verified 4fdfe33a

Support is_mtia() in TensorBase.h (#107723)

egienvalue committed 1 year ago

96c27c2d

Remove unnecessary import in python_variable.cpp (#107794)

janeyx99 committed 1 year ago

4fd42e62

Add tensor post accumulate grad hook API (#107063)

janeyx99 committed 1 year ago

6e71ad05

[TP][EZ] Update doc for TP parallel style (#107819)

fduwjj committed 1 year ago

3828cd4b

Revert "Add tensor post accumulate grad hook API (#107063)"

pytorchmergebot committed 1 year ago

432fce4e

Revert "Remove unnecessary import in python_variable.cpp (#107794)"

pytorchmergebot committed 1 year ago

bc079055

Add wait_tensor so print always has a correct result for AsyncCollectiveTensor (#107808)

ani300 committed 1 year ago

2c45a579

Move conda uploads into environment (#107807)

clee2000 committed 1 year ago

3d3f1826

Use docker-build env to access GHCR_PAT (#107655)

huydhn committed 1 year ago

9a365fe9

[Redo] Enhance fakepg: alltoall and alltoall_base (#107798)

xmfan committed 1 year ago

b74b8e33

Support integer implementations for padding(cpu and cuda) (#107755)

OrenLeung committed 1 year ago

726b7ff6

[profiler] move _enable_dynamo_cache_lookup_profiler (#107720)

davidberard98 committed 1 year ago

8c66f97c

[profiler] DISABLE_CUPTI_LAZY_REINIT for CUDA 12 as well (#107744)

davidberard98 committed 1 year ago

cb107c74

Update unary_ufuncs groupings to include primtorch types. (#107345)

ekamiti committed 1 year ago

4a022e21

[optim] Make casting to match params a hook (#106725)

janeyx99 committed 1 year ago

9f86d851

[export][reland] ExportedProgram.transform updates graph_signature automatically (#107792)

angelayi committed 1 year ago

92f6454f

[FSDP][Docs] Add note on `NCCL_CROSS_NIC=1` for HSDP (#107784)

awgu committed 1 year ago

2515ab93

Fix docs, missed a // in LaTeX for nadam (#107736)

janeyx99 committed 1 year ago

c0ba9a78

Port existing heuristics to TD framework (#107071)

ZainRizvi committed 1 year ago

36399d06

[mergebot] Flaky and broken trunk should take precedence over ic (#107761)

huydhn committed 1 year ago

d7f943ec

Decomp for aten.dropout (#106274)

SherlockNoMad committed 1 year ago

ee4b99cc

[core aten] Add ops to core aten set (#107766)

angelayi committed 1 year ago

50024d04

Newer Older