strisunshinewentingwang/pytorch

Pull Requests Commits

Added all bits types

angelayi committed 2 years ago

bd4e5eac

Add bit16 tensor type

angelayi committed 2 years ago

879b9bf5

Add bit16 tensor type

angelayi committed 2 years ago

f49d8e51

Add bit16 tensor type

angelayi committed 2 years ago

eba0ebb8

Add bit16 tensor type

angelayi committed 2 years ago

b4f55ab6

Ci andriod cache conda (#89554)

PratsBhatt committed 2 years ago

6ba6b64a

Include test/distributed/test_dynamo_distributed.py for ciflow/inductor (#89755)

ezyang committed 2 years ago

2661ff10

Revert "Add single process version of dynamo distributed hf_Bert tests (#89721)"

pytorchmergebot committed 2 years ago

0d9a615a

Support set_rng_state with fake tensor (#89642)

ezyang committed 2 years ago

2f8769d6

Guard traceable_tensor_subclasses patching with finally (#89689)

ezyang committed 2 years ago

856e2fa5

Don't modify log level in dynamo distributed test (#89655)

ezyang committed 2 years ago

49eb43fc

supress Werror introduced by lack of override by #86786 on `bool initialized()` (#89687)

jeanschmidt committed 2 years ago

Verified d089fbdc

Add mypy checking for a few files in torch/_dynamo (#89731)

ezyang committed 2 years ago

f45fe7de

[xla hash update] update the pinned xla hash (#89405)

pytorchmergebot committed 2 years ago

55e8b5c1

Add simple assert to detect fake tensors on modules (#89723)

voznesenskym committed 2 years ago

b5616cd5

Beef up AOTAutograd logging with aot_id and input descriptions (#89710)

ezyang committed 2 years ago

db1f1144

Don't suppress log messages for dynamo CI config (#89653)

ezyang committed 2 years ago

5f8848f3

Add single process version of dynamo distributed hf_Bert tests (#89721)

ezyang committed 2 years ago

1a2dd6b1

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

ezyang committed 2 years ago

0e7c100c

Factor input deduplication into a separate function (#89701)

ezyang committed 2 years ago

1f95f24d

Implement guard_source on RandomValueSource (#89711)

ezyang committed 2 years ago

dcefc8f9

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

ezyang committed 2 years ago

1da633f9

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

Chillee committed 2 years ago

e36d68af

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

Taylor Robie committed 2 years ago

b709078d

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

Taylor Robie committed 2 years ago

143d2881

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

Taylor Robie committed 2 years ago

ae725d50

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

ppwwyyxx committed 2 years ago

56e40fe0

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

pytorchmergebot committed 2 years ago

39449ea6

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

Taylor Robie committed 2 years ago

483d3a3d

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

Taylor Robie committed 2 years ago

0435894b

Older