strisunshinewentingwang/pytorch

Pull Requests Commits

Add a timeout to benchmark script (#90634)

ezyang committed 1 year ago

e1ed5ad5

Some memory saving in large unittests (#90148)

ppwwyyxx committed 1 year ago

5d8618df

[Fix]: Add some missing moves in 90442 (#90661)

Skylion007 committed 1 year ago

995d39c2

SymIntify resize_ and deduplicate memory format logic (#90442)

ezyang committed 1 year ago

e33f1eee

Simple fix: add missing positional arg in init_optimizer() call (#90641)

jgong5 committed 1 year ago

181d3747

Revert "Make torch._guards, shuffle structures around for migration (#90636)"

pytorchmergebot committed 1 year ago

15a4c603

[FSDP] Fix _pre_forward type annotation (#90621)

mrshenli committed 1 year ago

7ec1cb85

[FSDP] Allow MixedPrecision to skip inputs (#90620)

mrshenli committed 1 year ago

80542add

[FSDP] Tighten post-bwd cast to `reduce_dtype` (#90615)

awgu committed 1 year ago

31351c61

Make torch._guards, shuffle structures around for migration (#90636)

voznesenskym committed 1 year ago

933b6c4e

Adopt state_dict_pre_hook in FSDP (#90436)

rohan-varma committed 1 year ago

c7d2fb7f

[FSDP][Easy] Move to `_storage()` in test file (#90622)

awgu committed 1 year ago

746c773d

[FSDP] Uncomment test for `use_orig_params=True` (#90610)

awgu committed 1 year ago

68455986

[FSDP] Save `_stream_to_name` for debugging (#90611)

awgu committed 1 year ago

e7efeb52

Fix perf bug in #90528 (#90630)

Skylion007 committed 1 year ago

184f6b57

[Reland][FSDP] Another fix for `DTensor`, `use_orig_params=True` (#90562)

awgu committed 1 year ago

9eccfedc

Add global registry to composable API contract (#90579)

mrshenli committed 1 year ago

a69cdd9c

Reserve space for std::vector output in extract_tensors for nccl python bindings (#88203)

Skylion007 committed 1 year ago

12671fe6

Fix: [ATen] add more missing moves - part 2 (#89000)

Skylion007 committed 1 year ago

583d216c

Fix non-existing parameters in docstrings in torch/nn (#90596)

kit1980 committed 1 year ago

9ef1d55e

Completely redo how ShapeEnv guards are generated (#90528)

ezyang committed 1 year ago

45109ec3

Revert guaranteed symint allocation (#90381)

ezyang committed 1 year ago

49c674e1

Keep track of source name on all allocated SymInts (#90295)

ezyang committed 1 year ago

b68dead2

[Inductor] fix issue: redeclaration of float g_tmp_buffer_xxx (#90270)

blzheng committed 1 year ago

f9aa0990

[LTC] Make some LazyGraphExecutor private data structures protected (#90598)

alanwaketan committed 1 year ago

5a665a39

[torchgen] Introduce Executorch types and signatures (#90591)

larryliu0820 committed 1 year ago

ddf00c80

[torchgen] Let native function declaration generation logic take a callable (#90590)

larryliu0820 committed 1 year ago

de6beca8

[torchgen] Refactor types (#90589)

larryliu0820 committed 1 year ago

453ff960

[dims] Fix large array inputs (#88596)

zdevito committed 1 year ago

0457020d

[vision hash update] update the pinned vision hash (#90586)

pytorchmergebot committed 1 year ago

bb9fc32f

Newer Older