strisunshinewentingwang/pytorch

Pull Requests Commits

[inductor] Revert channels-last support (#87588)

anijain2305 committed 1 year ago

ebe5aad4

Fixed minor typos in torch.flip and torch.rot90 (#87724)

scaomath committed 1 year ago

312628d2

[ONNX] Fix pad Circular Mode (#86984)

titaiwangms committed 1 year ago

52ac8adc

Use setup_instance script to enable conda and load cuda libraries (#87296)

xuzhao9 committed 1 year ago

e532fb9a

build: support DNNL_GRAPH_CPU_RUNTIME=TBB (#87512)

min-jean-cho committed 1 year ago

7a6808c5

Add prepend argument to nn.Module hooks (#87370)

mrshenli committed 1 year ago

82698b89

[ONNX] replace AT_ASSERT with TORCH_INTERTNAL_ASSERT take 2 (#86405)

titaiwangms committed 1 year ago

82dff8ee

[ONNX] Support quantized::conv1d_relu (#85997)

titaiwangms committed 1 year ago

65b4a633

Disable test_inductor_timm_shard (#87710)

desertfire committed 1 year ago

15370d32

Graph-break on FSDP in dynamo (#87420)

wconstab committed 1 year ago

874625e0

[pytorch] Layer norm backward speed gain with warp shuffles (#87445)

valentinandrei committed 1 year ago

b6f28334

Add named_buffers to torchdynamo nn_module (#87644)

tugsbayasgalan committed 1 year ago

7b597825

consider numel args when identifying aligned args (#87394)

stumpOS committed 1 year ago

8a2a4ed4

Add get_guard_expr to symbolic_shapes which returns all guards in a single expression (#87665)

Chillee committed 1 year ago

569eebb4

Prefer python meta function over c++ meta function (#87426)

SherlockNoMad committed 1 year ago

eb99c1ef

[ONNX] Add Support on 0d tensor Broadcast (#87211)

titaiwangms committed 1 year ago

65601f5e

Revert "Intercept aten._reshape_alias for nvFuser (#87072)"

pytorchmergebot committed 1 year ago

5308886e

Performance improvment to cumulative seq len (#87530)

drisspg committed 1 year ago

0cba7888

[inductor] Trivial smoke-test (#87598)

bertmaher committed 1 year ago

87163fe8

[ROCm] [FakeTensorTest] Enable test_fallback_memory_prop (#85760)

jaglinux committed 1 year ago

9efca7c0

Support `signbit` in MPS. (#87214)

dfalbel committed 1 year ago

e818574e

Intercept aten._reshape_alias for nvFuser (#87072)

IvanYashchuk committed 1 year ago

163a829c

[vision hash update] update the pinned vision hash (#87639)

pytorchmergebot committed 1 year ago

9bbdc7ab

[JIT] Fix return types of inputs/outputs method in Graph (#86349)

take-cheeze committed 1 year ago

e85230b8

Fix torch.testing.assert_close not exported from module (#87619)

bschnurr committed 1 year ago

0367c12b

remove unnecessary __syncthreads() in conv_depthwise2d_grad_weight_kernel (#84854)

s5u13b committed 1 year ago

ec159429

Fix `tensor.stride()` type hint (#84177)

soof-golan committed 1 year ago

874a94ce

Fix use after free in tensorpipe agent (#87627)

H-Huang committed 1 year ago

4ef5f5de

[Python] refactor slices on sorted (#86995)

ScholliYT committed 1 year ago

fd60b818

[Inductor] Truncate function expr str if it's too long at RecordLoadStore (#87248)

yanboliang committed 1 year ago

98f40af7

Newer Older