strisunshinewentingwang/pytorch

Pull Requests Commits

Add tensor post accumulate grad hook API (#107063)

janeyx99 committed 1 year ago

3f655277

Do not mutate `SymNode` expression. (#107492)

ysiraichi committed 1 year ago

bcede143

Fix: transactional translation validation insertion. (#107523)

ysiraichi committed 1 year ago

d2215f14

reduce header file to boost cpp_wrapper build. (#107585)

xuhancn committed 1 year ago

3f3479e8

Enable Mypy Check in torch/_inductor/triton_heuristics.py (#107135)

FFFrog committed 1 year ago

94d85f18

[export] Add save/load function (#107309)

angelayi committed 1 year ago

431d25a1

Unlift mutated buffers (#107643)

tugsbayasgalan committed 1 year ago

134d4156

Re-enable AVX512 ATen kernels for compute-intensive ops (#104165)

sanchitintel committed 1 year ago

8ed169b1

Avoid executing side-effectful graph_module as validation step (#107271)

tugsbayasgalan committed 1 year ago

ee72071f

Update utils.h and correct misleading error messages (#107602)

majercakdavid committed 1 year ago

155d1285

[ONNX] Add unittest for exporting embedding_bag (#105862)

BowenBao committed 1 year ago

f9f88f2d

[vision hash update] update the pinned vision hash (#107649)

pytorchupdatebot committed 1 year ago

849fbc69

[dynamo] Store originating source in the Guard object (#107634)

anijain2305 committed 1 year ago

a506d0ad

[dynamo] Continue on fbgemm import fail (#107622)

anijain2305 committed 1 year ago

12b0372a

Fix FP16Planner documentation (#107620)

b-chu committed 1 year ago

3361fae8

Quick return when there's nothing to bound in bound_sympy (#107549)

lezcano committed 1 year ago

f1310164

Wrap indirect indexing on CUDA (#105055)

lezcano committed 1 year ago

85c673e6

Use fast traceback for symbolic shapes (#107439)

ezyang committed 1 year ago

8292b03c

Change how caching/cleanup for CapturedTraceback works (#107471)

ezyang committed 1 year ago

072bb061

Move torch.export() to torch.export.export() (#107609)

gmagogsfm committed 1 year ago

bbb216bc

[fx][split] make sure we copy node.meta over during split (#107248)

842974287 committed 1 year ago

2e73c86d

[dtensor] group all dynamo tests together (#107487)

wanchaol committed 1 year ago

9c2b4a35

[dynamo] enable 2d torch.compile test (#107473)

wanchaol committed 1 year ago

42f25d49

Update reduce_scatter_tests to work for world_size > 1 (#104424)

H-Huang committed 1 year ago

8c10be28

[optim] FusedAdam/W accepts lr: Tensor without h2ds (#106916)

janeyx99 committed 1 year ago

1641d671

Add space to merge cancel comment (#107603)

janeyx99 committed 1 year ago

350fb16f

torch._numpy: remove noops and half-implemented nan-functions (#107596)

ev-br committed 1 year ago

da67b414

[1/N] Introduce init_device_mesh() (#107254)

wz337 committed 1 year ago

f5d1df3c

Make emit_metrics importable without having boto3 installed (#107070)

ZainRizvi committed 1 year ago

5ddb8ef8

[inductor] Adjust dynamic SMEM limit when above default in AOT (#107601)

aakhundov committed 1 year ago

3920ce2f

Newer Older