strisunshinewentingwang/pytorch

Pull Requests Commits

Higher order operator util for raising if inputs require grads (#106078)

drisspg committed 1 year ago

788c8258

aot_inductor_interface: surface previously eaten error messages (#105366)

kflu committed 1 year ago

57d0bec3

[inductor] Make autotune_process.py pass mypy (#105791)

jkseppan committed 1 year ago

186352a6

fix x.numpy() breaks in #106211 (#106327)

williamwen42 committed 1 year ago

018ac763

[onnx] Fix output shape mismatch issue of max_pool (#106270)

CYuxian committed 1 year ago

2f35715f

[FSDP] Validate buffer dtype in pre-forward hook for FSDP mixed precision tests (#106231)

vnalla-meta committed 1 year ago

2138aaa9

[optim][rprop] Minimize intermediates=1 for foreach to save memory (#105193)

janeyx99 committed 1 year ago

23f47f74

Correct URL Link for torchDynamo (#105903)

wangxiyuan committed 1 year ago

4eeda661

[ROCm] use hipblas instead of rocblas (#105881)

jeffdaily committed 1 year ago

5379b5f9

Move more TCPStorestate from BackgroundThread to TCPStoreMasterDaemon as it won't be used by the libuv backend. (#105674)

Rodrigo Kumpera committed 1 year ago

c9c66819

[ONNX] Log instead of crash when 'tabulate' is not installed (#106228)

BowenBao committed 1 year ago

57a47ed9

Use nodejs-16 for docs builds (#106312)

malfet committed 1 year ago

19637241

Enable distributed cpp test for rocm (#106132)

hongxiayang committed 1 year ago

1b757fb6

AOTAutograd: correctness fix when tracing custom autograd functions that alias inputs (#102992)

bdhirsh committed 1 year ago

4a549dd5

Remove @penguinwu from distributed codeowners (#106322)

awgu committed 1 year ago

a0b6c0d1

Deallocate constant when it is no longer needed in constant folding (#106216)

eellison committed 1 year ago

0010a8f7

[ONNX] Perform Shape inference on added "Cast" node (#106093)

ilyasher committed 1 year ago

cdc91277

[export] Move attrs to properties and add BC decorator (#106170)

angelayi committed 1 year ago

66c53742

`torch::nn::functional::batch_norm()`: add a shape check of input tensor (#105930)

Sehun0819 committed 1 year ago

0dc25132

sm90 efficient attention test fixes (#105978)

Fuzzkatt committed 1 year ago

1cebfef8

Reland "Make adding buffers more like adding parameters (#104069)" (#106224)

mikaylagawarecki committed 1 year ago

d8e5f2aa

[ROCm] HIP stream priority fix post #101956 (#106157)

jeffdaily committed 1 year ago

50e3f9cb

Revert "Reland "Add forward mode AD to out-place foreach functions (#102409) (#106043)"

pytorchmergebot committed 1 year ago

2b427ae3

Optimize stack frame inspection in `torch._custom_op.impl:CustomOp._register_impl` (#105940)

jeffdunn committed 1 year ago

c5b9dc1f

faketensor: prevent deepcopy from cloning FakeTensorMode (#104476)

bdhirsh committed 1 year ago

c54afea6

Fix typo which suppresses user exception reporting (#106289)

ezyang committed 1 year ago

d3b508d0

fix torchrun script for custom device (#105443)

heidongxianhua committed 1 year ago

0af3203c

Revert "Update kineto submodule to 465ff4cd7 (#106154)"

pytorchmergebot committed 1 year ago

0a0abd0e

UFMT torch/jit/frontend.py, manually fix mypy suppression (#106268)

ezyang committed 1 year ago

3c70d4bd

UFMT torch/jit/_script.py, manually move mypy suppressions (#106267)

ezyang committed 1 year ago

af88e6d0

Newer Older