strisunshinewentingwang/pytorch

Pull Requests Commits

shard mack to 3

clee2000 committed 2 years ago

d1625348

[fx] fix type promotion in `binary_magic_impl` (#91376)

nkaretnikov committed 2 years ago

88b3810c

[fx] rewrite `FloorDiv` to match Python better (#90906)

nkaretnikov committed 2 years ago

d13207c7

Move XLA test job to 4xlarge (#92269)

huydhn committed 2 years ago

5e0d3458

Fix AOTAutograd 2.0 perf regression involving as_strided (#92255)

ezyang committed 2 years ago

ded2b47b

Clean up more clang-tidy supression (#92203)

cyyever committed 2 years ago

9b716a06

Refactor inductor to use standard BACKENDS dict (#92187)

jansel committed 2 years ago

bbce4184

Add sym_size/stride/numel/storage_offset to native_function.yaml (#91919)

SherlockNoMad committed 2 years ago

0388400f

[dtensor] enable op db tests by using multithreaded test case (#92198)

wanchaol committed 2 years ago

801d831d

[dtensor] switch pointwise op tests to use DTensorOpsTestBase (#92197)

wanchaol committed 2 years ago

2ce63ef2

[threaded_pg] full rewrite of MultiThreadedTestCase to enable device_type tests (#91650)

wanchaol committed 2 years ago

e16979c9

[threaded_pg] enable subpg creation and concurrent collective (#91649)

wanchaol committed 2 years ago

9942ddd5

Fix oneDNN double checkout issue and Upgrade oneDNN to v2.7.3 (#92239)

yanbing-j committed 2 years ago

85edb581

Fix typos introduced by 014ac7fda2d1e59796b1147221fb92f4377ca2f1

malfet committed 2 years ago

d62eff56

Add ROCm merge rules (#85762)

jithunnair-amd committed 2 years ago

014ac7fd

Fix CUDA error not getting captured by handler (#92227)

r-barnes committed 2 years ago

eadbf762

Don't raise error if job_id can't be fetched (#92192)

malfet committed 2 years ago

32937f39

[ROCm] disable NVFuser (#92182)

jeffdaily committed 2 years ago

301644d3

Unit test for is_causal Better Transformers (#91900) (#92102)

jcrousse committed 2 years ago

0b90ddac

Add missing conversion for to_sparse.sparse_dim (#92006)

alexsamardzic committed 2 years ago

b05f5096

Revert "[cuDNN][cuDNN V8 API] Always build assuming cuDNN >= 8.0 (#91527)"

pytorchmergebot committed 2 years ago

523d4f25

Revert "Add kwargs support to torch.export() API (#92013)"

pytorchmergebot committed 2 years ago

1a98c3e3

[inductor] Respect dtype argument in ops.constant (#92093)

peterbell10 committed 2 years ago

76c88364

Add MTIA DeviceType for Meta training and inference devices (#92232)

yuhc committed 2 years ago

5a0fa04a

[ONNX] Raise Unsupported for Grid Sample with volumetric 5D input (#92212)

leimao committed 2 years ago

9cf84347

[FSDP][BE] Improve `device_id` + CPU offload test (#92031)

awgu committed 2 years ago

85e0fd02

[FSDP][BE] Rename `prefixed_param_names` -> `fqns` for consolidation (#92028)

awgu committed 2 years ago

5a3b4dac

[FSDP][BE] Better error msg for incorrect device for training (#92027)

awgu committed 2 years ago

b0888cce

[DTensor] remove redundant device mesh test code (#92069)

XilunWu committed 2 years ago

b5d8fef9

[DTensor] check DeviceMesh ranks contiguity (#91802)

XilunWu committed 2 years ago

513c1e71

Older