strisunshinewentingwang/pytorch

Pull Requests Commits

[Dynamo] Fix number of inputs in onnxrt and tvm backend (#95429)

YJ Shi committed 1 year ago

a10897a3

Optimize nn.Module __call__ fast path for dynamo (#95931)

wconstab committed 1 year ago

26045336

[Dynamo] Support call dict with list/tuple as input (#95928)

yanboliang committed 1 year ago

6ca286df

Revert "[inductor] Improve error messages (#95567)" (#96014)

jansel committed 1 year ago

43dd043e

Add various uninterpreted bit tensor data types (try 2) (#95860)

vkuzo committed 1 year ago

dc70e817

[vision hash update] update the pinned vision hash (#95932)

pytorchmergebot committed 1 year ago

5e1067bc

Improve error message for instance norm when channels is incorrect (#94624)

soulitzer committed 1 year ago

7ff9612e

[MPS] Error on unsupported types (#95982)

malfet committed 1 year ago

436993d5

[BE] Remind people there are rsts to update in docs/source (#95914)

janeyx99 committed 1 year ago

f4b33791

Make int unspecialization actually work (#95621)

ezyang committed 1 year ago

d303665d

Fix wrong handling of `grad_scale` & `found_inf` in fused optimizers (#95847)

crcrpar committed 1 year ago

7d765cdc

Prettify assert expr in self.symbol_to_source failure (#95972)

ezyang committed 1 year ago

d214d82a

[SPMD] Introduce the cross-iteration graph optimization framework (#94803)

fegin committed 1 year ago

4d9c499d

Remove fake inputs from control flow (#95988)

angelayi committed 1 year ago

5a07c3d3

Add flop formulas for sdpa (#95854)

Chillee committed 1 year ago

9a781ce3

Improve Discoverability of Inductor Optimizations (#95824)

msaroufim committed 1 year ago

7db5f8c7

Fix typo in RELEASE.md and CONTRIBUTING.md (#95965)

kiszk committed 1 year ago

7d02ecfa

Add export docs, improve asserts (#94961)

voznesenskym committed 1 year ago

ac07de4a

Don't use guardless contiguity/stride-like implementations (#95733)

ezyang committed 1 year ago

027ebca4

[EASY] Unindent some blocks (#95967)

ezyang committed 1 year ago

f8b57ba6

Fix: `validate_input_col` for partial functions (#95067)

BeckerFelix committed 1 year ago

e6f3e16d

memory viz: Add colors for categories and a legend (#90587)

zdevito committed 1 year ago

ee438425

[inductor] do not use `ceil` in `arange` ref (#95773)

nkaretnikov committed 1 year ago

c72fbf2e

[inductor] use FP div in CPP expr printer (#95698)

nkaretnikov committed 1 year ago

feffcafe

[DTensor][Shampoo] add _tenso.zero function (#95863)

minddrummer committed 1 year ago

6c061e51

[MTPG] Improve all_reduce and handle bwd thread support (#95524)

Rodrigo Kumpera committed 1 year ago

1d3c394d

[DCP] Add DCP FSDP sharded_state_dict checkpoint example to DCP .rst file (#95517)

wz337 committed 1 year ago

a7698a82

Revert "Don't use guardless contiguity/stride-like implementations (#95733)"

pytorchmergebot committed 1 year ago

4026c621

Run _nvfuser/test_torchscript serially (#95951)

clee2000 committed 1 year ago

7f5f0b36

Revert "[inductor] Add an AOT compilation mode for Inductor CPP backend (#94822)"

pytorchmergebot committed 1 year ago

879400e4

Newer Older