strisunshinewentingwang/pytorch

Pull Requests Commits

Add lowering for cummax, cummin (#120429)

isuruf committed 187 days ago

409b1a60

[dynamo][compile-time] Remove preserve rng state per op (#121923)

anijain2305 committed 187 days ago

d04faf45

Fix FakeTensorUpdater logic for updating fake tensors (#116168)

isuruf committed 187 days ago

67ec8702

combine loops so fn_name correct in error message (#121601)

drewd789 committed 187 days ago

239d87af

[release] Increase version 2.3.0->2.4.0 (#121974)

atalman committed 187 days ago

39fdde7f

update kineto submodule commit id (#121843)

shengbao-zheng committed 187 days ago

565d1e28

Disable inductor (default) and inductor (dynamic) by default in the perf run launcher (#121914)

Chillee committed 187 days ago

3c3d7455

[export] Add serialization support for tokens (#121552)

angelayi committed 188 days ago

ef25d83a

[FSDP2] implement HSDP (#121569)

weifengpy committed 188 days ago

014f91a9

[BE] Migrate pull.yml to use S3 pytorch-ci-artifacts bucket for linux-jammy-py3_8-gcc11 and docs builds/tests (#121908)

jeanschmidt committed 188 days ago

4cbf9638

Force upsample to be float32 (#121324)

bhack committed 188 days ago

2770e3ad

[compiled autograd] free stack objects before calling compiled graph (#121707)

xmfan committed 188 days ago

e25054b2

[dynamo] Convert invalid args into graph breaks (#121784)

jansel committed 188 days ago

5a2b4fc8

better support set_default_dtype(torch.float16), update doc (#121730)

LamForest committed 188 days ago

fc33bbf8

[executorch hash update] update the pinned executorch hash (#121871)

pytorchupdatebot committed 188 days ago

8fdd8125

[Clang-tidy header][24/N] Fix clang-tidy warnings on c10/cuda/*.{cpp,h} (#120781)

cyyever committed 188 days ago

fb10e130

DTensor: add comm tests to test_tp_examples (#121669)

d4l3k committed 188 days ago

e4fda049

[DCP][DSD] Add AdamW to distributed state dict unit tests (#121774)

wz337 committed 188 days ago

02083f54

[executorch] Make trymerge force merges actually work with executorch (#121920)

PaliC committed 188 days ago

efbeefbb

[dynamo][compile-time] Make output_graph new_var linear (#121858)

anijain2305 committed 188 days ago

a6236660

use two pass reduction for deterministic reduction order (#115620)

zhuhaozhe committed 188 days ago

3bc2bb67

Revert "[aoti] Fix compilation bug for buffer mutations (#121688)"

pytorchmergebot committed 188 days ago

0cd094a4

Make torch/_inductor/comms.py recognize native funcol IRs as collective IRs (#118498)

yifuwang committed 188 days ago

01d7c948

[dynamo] Refactor update_block_stack into a seperate function (#121810)

jansel committed 188 days ago

60ccf814

[dynamo] Compile time optimizations in tx.step() (#121790)

jansel committed 188 days ago

1e9a7df8

Fix #83153: torch.nn.hardtahn allowed min_val to be greater than max_val (#121627)

joaotgouveia committed 188 days ago

1afa8e09

[dtensor] refactor and generalize stack strategy (#121869)

wanchaol committed 188 days ago

710446b1

Revert "Switch cudagraph backend to cudagraph trees (#121019)" and "Add Cudagraphs disable checking (#121018)" (#121864)

anijain2305 committed 188 days ago

92ed8553

[PyTorch] Fix static runtime sigrid_hash precomputed multiplier pass (#120851)

swolchok committed 188 days ago

d604ab81

[jit] ClassType hashing: hash on compilation_unit as well (#121928)

davidberard98 committed 188 days ago

cceabe87

Newer Older