strisunshinewentingwang/pytorch

Pull Requests Commits

Fix TorchInductor benchmarking in fbcode (#88689)

jansel committed 2 years ago

de53d414

[vulkan] Add option for buffer representations in vTensor (#87622)

SS-JIA committed 2 years ago

c4a3aa8f

Meta function for aten.sort and aten.scatter* (#88705)

ezyang committed 2 years ago

d81797e8

Mark dynamo torchbench dlrm as unsupported (#88712)

wconstab committed 2 years ago

100b5563

[fix] MathBits: serialization (#88182)

kshitij12345 committed 2 years ago

eb9b1560

[BE] Delete push_nightly_docker_ghcr (#88748)

malfet committed 2 years ago

525fe53a

[inductor] Handle nested tuple/list output in fallback kernel (#88495)

desertfire committed 2 years ago

f11f0e4a

extract out the clean workspace test to its own file (#88682)

mikey dagitses committed 2 years ago

3150c9dc

Add SherlockNoMad to symbolic-shapes reviewer list (#88739)

ezyang committed 2 years ago

c19bae9f

Add voznesenskym to symbolic-shapes group, move wconstab to listener (#88593)

ezyang committed 2 years ago

44de7cdb

Mark diag.out composite (#88670)

ezyang committed 2 years ago

c86cc68d

Add min cut partitioner for AOT+nvFuser (#88204)

IvanYashchuk committed 2 years ago

69b23522

Changing as_strided_scatter to deterministic inputs (#85583)

srossross committed 2 years ago

ff7c5b0d

[Inductor] fix c++ compile error with masked float value init (#88298)

blzheng committed 2 years ago

fca6ed02

upsample_*.vec ops are now CompositeImplicit (#85638)

Fabio Rocha committed 2 years ago

652af5ec

[primTorch] Improve `narrow` and `narrow_copy`: refs, tests, docs (#87045)

nkaretnikov committed 2 years ago

aa8279bc

[Quant] Support lowering of channel shuffle in FX (#83731)

Xia-Weiwen committed 2 years ago

f6192b75

[BE] Move `setup-ssh` step ahead of clone PyTorch (#88715)

malfet committed 2 years ago

ab9a19a9

Hopper (`sm90`) support (#87736)

eqy committed 2 years ago

a7420d2c

Fix Python-bound function signature (torch._C.Graph.addInput) (#88528)

wschin committed 2 years ago

19d7941e

Meta registrations for inplace operators (#88678)

ezyang committed 2 years ago

f0e6cea2

Meta implementation for unsqueeze_ (#88675)

ezyang committed 2 years ago

a880ddc1

Meta implementation for bernoulli (#88676)

ezyang committed 2 years ago

1dab35ca

Update gloo submodule (#88530)

malfet committed 2 years ago

6be426ca

[export] Preserve meta["val"] on placeholders in dynamo.export(). (#88651)

zhxchen17 committed 2 years ago

08b2a251

Reduce the number of shards inductor uses for model tests (#88610)

desertfire committed 2 years ago

5f876bfd

Add implementation for irregular dimension selection for nested tensors. (#88585)

Antoni Viros i Martin committed 2 years ago

9f58e027

[nn] add remove_duplicate flag to named_parameters (#759) (#88090)

samdow committed 2 years ago

87238e64

[Profiler] Memory profiler part 1: Gradient identification (#86802)

Taylor Robie committed 2 years ago

cef13ebe

[dynamo] only error out on nested fx trace if dynamo is optimizing (#88640)

suo committed 2 years ago

c0e6b432

Newer Older