strisunshinewentingwang/pytorch

Pull Requests Commits

fix xla failures

samdow committed 2 years ago

25309c2b

use detach, add tests

samdow committed 2 years ago

cf56b29d

make embedding work with forward mode AD

samdow committed 2 years ago

6f3457b0

Fix distributed_test.py flakiness (#78797)

rohan-varma committed 2 years ago

79931889

More forward-over-reverse implementations. (#78740)

nikitaved committed 2 years ago

a4509f5b

[ci] harden test stats reporting

suo committed 2 years ago

298b9ad7

Run MPS tests (#78723)

malfet committed 2 years ago

f7ac389e

to_padded_tensor doc v0 (#78657)

YifanShenSZ committed 2 years ago

6ba1d05f

Add Caching of Conversion to Fake/Meta tensors in FakeTensorMode

eellison committed 2 years ago

26d27395

[FSDP] Allow different `optim_input` orders across ranks

awgu committed 2 years ago

4615738a

[FSDP][Docs] Fix typo in `full_optim_state_dict()`

awgu committed 2 years ago

d4d8aaf7

rename BUILD.buck to BUCK.oss (#78792)

linbinyu committed 2 years ago

1683a261

[cuDNN] [cuDNN v8 API] Support cuDNN Errata Filter (#73934)

eqy committed 2 years ago

fc66521e

[FSDP] Return original module when fsdp wrapped model call .module (#78671)

LinjianMa committed 2 years ago

c29df68f

Avoid CPU Sync in SyncBatchNorm When Capturing CUDA Graphs

mrshenli committed 2 years ago

1884d7fb

Remove non-existing code_template.py glob (#78773)

kit1980 committed 2 years ago

1eab34d1

Add linux-focal-py3.7-gcc7 to merge_rules.json (#78785)

kit1980 committed 2 years ago

dd45e8d3

Revert "Autogen Tags enum, and allow specifying tags while defining an op"

pytorchmergebot committed 2 years ago

954522a4

[ONNX] Variable length argument support for quantized_args (#78775)

justinchuby committed 2 years ago

c0814bff

Ported t decomp to become a ref (#78686)

Chillee committed 2 years ago

1ea4075b

Autogen Tags enum, and allow specifying tags while defining an op

anjali411 committed 2 years ago

9476a78f

[quant] follow up fixes for prepare_fx/prepare_qat_fx calls in classyvision (#105) (#78660)

jerryzh168 committed 2 years ago

063c9366

[TF32] Fix typo in tf32 wrapper function (#78438)

eqy committed 2 years ago

ad1bff1b

[cuDNN][TF32] Threshold adjustments for TF32 on `>=sm80` (#78437)

eqy committed 2 years ago

b740a99b

Updating torch.log example

Gamrix committed 2 years ago

416f581e

Revert "Reenable assert after test update"

pytorchmergebot committed 2 years ago

8047d2a5

Migrate off Xenial gcc5.4 for trunk jobs (#78734)

kit1980 committed 2 years ago

76392c67

[JIT] enable autocasting + freezing test

davidberard98 committed 2 years ago

4ab62ecf

using new rockset commit_jobs_batch_query to print check statuses (#78750)

swang392 committed 2 years ago

88f4a124

Allow relocatable device code linking in pytorch CUDA extensions (#78225)

xwang233 committed 2 years ago

ef0332e3

Older