strisunshinewentingwang/pytorch

Pull Requests Commits

[benchmarks] Disabling gradscaler

anijain2305 committed 1 year ago

f1bf4241

Add debug asserts to AOTAutograd for input consistency with compilation (#89702)

ezyang committed 1 year ago

0e7c100c

Factor input deduplication into a separate function (#89701)

ezyang committed 1 year ago

1f95f24d

Implement guard_source on RandomValueSource (#89711)

ezyang committed 1 year ago

dcefc8f9

Access named parameters/buffers/etc via getattr rather than index (#89625)

ezyang committed 1 year ago

1da633f9

Don't allow recomputing a node that *must* be materialized in the backwards pass (#89171)

Chillee committed 1 year ago

e36d68af

[Profiler] Memory profiler part 11: Mark tensors created in the backward pass which don't correspond to parameters. (#88926)

Taylor Robie committed 1 year ago

b709078d

[Profiler] Memory profiler part 10: Mark optimizer state (#88925)

Taylor Robie committed 1 year ago

143d2881

[Profiler] Memory profiler part 9: Mark activations (#88924)

Taylor Robie committed 1 year ago

ae725d50

Let SyncBatchNorm fallback to BN if not using distributed training (#89706)

ppwwyyxx committed 1 year ago

56e40fe0

[vision hash update] update the pinned vision hash (#89692)

pytorchmergebot committed 1 year ago

39449ea6

[Profiler] E2E expecttests for category assignment (#88653)

Taylor Robie committed 1 year ago

483d3a3d

[Profiler] Memory profiler part 8: Mark parameters. (#87568)

Taylor Robie committed 1 year ago

0435894b

[Profiler] Memory profiler part 7: Mark inputs (#87567)

Taylor Robie committed 1 year ago

17fa6bf1

[Profiler] Memory profiler part 6: Mark gradients and temporary intermediates. (#87566)

Taylor Robie committed 1 year ago

64c5c77c

[Profiler] Memory profiler part 5: Data flow graph (#87006)

Taylor Robie committed 1 year ago

5f09a6d5

[Profiler] Memory profiler part 4: Select top level torch ops (#86880)

Taylor Robie committed 1 year ago

c3116dd7

[Inductor] Record cpp kernel in PyTorch Profiler (#89367)

jgong5 committed 1 year ago

bb77accb

Don't suppress exceptions from backends (#89656)

ezyang committed 1 year ago

36018a6e

put descriptive kernel names behind config (#89697)

ngimel committed 1 year ago

3e20d023

update docstring for torch.linalg.lstsq (#89383)

jlukehubbard committed 1 year ago

591dfffa

Simplify aot_module_simplified by removing top_args/top_kwargs (#89666)

ezyang committed 1 year ago

c9a0cc86

Don't support kwargs at runtime in aot_module_simplified (#89664)

ezyang committed 1 year ago

6168f22f

Delay verify correctness wrapping to call site. (#89662)

ezyang committed 1 year ago

b04dda42

make inductor correctly propagate nans for maximum and minimum (#89612)

ngimel committed 1 year ago

61a3fe4b

Fix typo in segment_reduction_op_gpu.cu (#89647)

eltociear committed 1 year ago

70c0a300

complex: register c10::complex with py::cast (#89680)

kshitij12345 committed 1 year ago

2c0bd85c

Implement old windows in Python (#87082)

alvgaona committed 1 year ago

abb446af

torchdynamo to torch._dynamo in aot_autograd.py (#89385)

jansel committed 1 year ago

95ea47ef

Remove fake_tensor_propagation (#89646)

ezyang committed 1 year ago

69043247

Older