strisunshinewentingwang/pytorch

Pull Requests Commits

[QNNPACK] Fix Memory Leak in QNNPACK QSoftmax Op (#89544)

salilsdesai committed 2 years ago

f628f2ed

Add non-reentrant checkpoint to composable APIs (#90015)

mrshenli committed 2 years ago

7bd28449

[UCC] Properly finalize unsuccessful collective posts (#89306)

Aidyn-A committed 2 years ago

a5430e10

Revert "[quant] Explictly set default quantized engine instead of relying on the order of supported_qengines (#89804)"

pytorchmergebot committed 2 years ago

063bbeb3

[doc] update dtensor readme (#89991)

feifeibear committed 2 years ago

29ea1c9c

triton supports devices < 7.0, not 6.0 (#90020)

soumith committed 2 years ago

6f5945e4

[quant] Explictly set default quantized engine instead of relying on the order of supported_qengines (#89804)

jerryzh168 committed 2 years ago

607ff6f4

[Vulkan][TCC] Add tests for quantized add, sub, mul and div (#89578)

manuelcandales committed 2 years ago

d04480a6

[quant][be] Merge qconfig_mapping_utils.py in quantization and fx folders (#89979)

jerryzh168 committed 2 years ago

8aee7680

[aarch64] add sleef_arm dependency (#89988)

ajaymh committed 2 years ago

0ad6715b

[chalf] relax tolerance : conv_transpose2d (#89993)

kshitij12345 committed 2 years ago

07be48de

[tp] ufmt test/distributed/tensor (#89970)

wanchaol committed 2 years ago

ca5526cf

[tp] umft distributed.tensor.parallel (#89969)

wanchaol committed 2 years ago

9b5e6b02

[dtensor] ufmt test/distributed/_tensor (#89968)

wanchaol committed 2 years ago

c37c5163

[dtensor] ufmt distributed._tensor (#89967)

wanchaol committed 2 years ago

bf23e0bd

Add `torch.compile` implementation (#89607)

malfet committed 2 years ago

768bd3fb

Revert "Dynamo, FX, Inductor Progress Bars (#88384)" (#90018)

seemethere committed 2 years ago

bcf4292f

Editorial pass on Dyamo docs (#89921)

Svetlana Karslioglu committed 2 years ago

015b05af

[ONNX] Supports scatter_add with different static shape of src and index (#89787)

titaiwangms committed 2 years ago

b2f34055

[Quant][fx][bc-breaking] Rename fx/*patterns.py (#89872)

andrewor14 committed 2 years ago

d8005631

fix citation file in MANIFEST (#89994)

pmeier committed 2 years ago

314e7c37

Remove DDP import (#89982)

rohan-varma committed 2 years ago

a5532929

Add error repro test for FSDP ignored modules with mixed precision (#89971)

mrshenli committed 2 years ago

5a36d998

add vjp test with non-contig inputs (#89375)

kshitij12345 committed 2 years ago

dfb533ca

Type torch._dynamo.guards (#89919)

ezyang committed 2 years ago

99dac4dd

Guarantee symbol allocation for all sizes/strides/storage offset (#89879)

ezyang committed 2 years ago

e03cde07

Add definitely_not_01 set to ShapeEnv. (#89871)

ezyang committed 2 years ago

74bcf2b6

When dealing with dupe arguments, prefer leafifying if possible (#89896)

ezyang committed 2 years ago

8d333761

[FSDP][Dynamo] Define annotation attributes as globals (#89913)

awgu committed 2 years ago

808cb2e8

remove torch.equal usages (#89527)

pmeier committed 2 years ago

4095ef8b

Newer Older