strisunshinewentingwang/pytorch

Pull Requests Commits

CUDA Graph Trees - error on deallocated access (#100927)

eellison committed 1 year ago

0ec46465

[Dynamo] Remove cross import in dynamo unit tests (#100851)

yanboliang committed 1 year ago

369a2563

Update cpuinfo submodule to include AVX512-FP16 detection (#100865)

CaoE committed 1 year ago

502e7912

[inductor] Test for shape padding (#100493)

bertmaher committed 1 year ago

4a4854f6

Reduce fake_tensor create_mode logging (#101074)

bertmaher committed 1 year ago

d2830752

Check canary_models for models too in torchbench.py (#101081)

ezyang committed 1 year ago

ad070b6d

Revert "Fix header inclusions in c10 by iwyu (#100304)"

pytorchmergebot committed 1 year ago

4eaaa086

Add dummy CUDA kernel for assert_async.msg (#101130)

tugsbayasgalan committed 1 year ago

683adb20

Revert "CUDA Graph Trees - error on deallocated access (#100927)"

pytorchmergebot committed 1 year ago

cbfed470

bsr_dense_bmm(): enable more precise float32 support with float64 accumulators (#100882)

nikitaved committed 1 year ago

dd2c22f4

implementation of DataPtr context for copy-on-write tensors (#100818)

dagitses committed 1 year ago

979f55d3

[CI][MPS] Actually make grid_sampler_2d available (#101108)

malfet committed 1 year ago

87084643

[MPS] Rename metalIndexingFunction to metalIndexingPSO (#101156)

qqaatw committed 1 year ago

c4752b1a

Fix microbenchmarks (#101065)

lezcano committed 1 year ago

8b4e28d6

Remove NullContext() from benchmark runners (#100309)

jansel committed 1 year ago

036a8d6b

[cuBLAS][cuBLASLt] Allow user-specified cuBLASLt workspace size via `CUBLASLT_WORKSPACE_SIZE` (#101145)

eqy committed 1 year ago

c25fdc20

[inductor] simplify test_cpu_repro with self.common (#101050)

jgong5 committed 1 year ago

b46553f6

inductor: fallback ConvTranspose when output_padding is big (#100846)

chunyuan-w committed 1 year ago

ea86eb31

[Quant][PT2E]Fix pt2e quantization maxpool input observer issue (#100961)

leslie-fang-intel committed 1 year ago

a66de845

Fix header inclusions in c10 by iwyu (#100304)

cyyever committed 1 year ago

6037ee8c

Revert "PyTorch -> C++17 (#98209) (#100557)"

pytorchmergebot committed 1 year ago

da02ccc6

Turn on anomaly detection for AOTAutograd backward tracing (#101047)

ezyang committed 1 year ago

2621fbda

simplify sdpa backward meta registration (#101128)

Natalia Gimelshein committed 1 year ago

15a51e20

[vision hash update] update the pinned vision hash (#101142)

pytorchmergebot committed 1 year ago

5f89d89a

[Dynamo] Fix nested function resume execution (#100426)

yanboliang committed 1 year ago

075d36d3

benchmarks: make --amp works for cpu path (#101057)

XiaobingSuper committed 1 year ago

c84627c2

[dtensor] tensor ops to use strategy based sharding prop (#100607)

wanchaol committed 1 year ago

a1aa32e2

Run C++ test_api binary directly in CI slow jobs (#101088)

huydhn committed 1 year ago

d1f0c8e2

[c10d] Figure out device to use for object collectives (#100954)

kwen2501 committed 1 year ago

0848ed21

Restore Vulkan tests to periodic (#101026)

huydhn committed 1 year ago

a0e6ae2c

Newer Older