strisunshinewentingwang/pytorch

Pull Requests Commits

Automated submodule update: kineto (#89846)

facebook-github-bot committed 1 year ago

dbe85265

[Composable API]Common _State parent class for composable and wrapper FSDP (#89147)

fegin committed 1 year ago

d52f121d

[FSDP] Fix `use_orig_params=True` + `no_sync()` (#90546)

awgu committed 1 year ago

b66cedd9

Fix forward AD custom Function non-differentiable outputs (#90787)

soulitzer committed 1 year ago

6d425a7c

[LTC] Make some LazyTensor interfaces virtual (#90686)

alanwaketan committed 1 year ago

9575f2ca

Add support for kineto in memory viz (#90567)

zdevito committed 1 year ago

bf2668a8

Doc for Canonical Aten and Prims IR (#90644)

SherlockNoMad committed 1 year ago

b4b8a565

[FSDP][optim_state_dict][5/N] Remove optim_inputs for sharded state_dict. (#89981)

fegin committed 1 year ago

65e762ac

[FSDP][optim_state_dict][4/N] Remove the unused _get_flat_param_to_fsdp_module API (#89980)

fegin committed 1 year ago

4a2d6499

[fix] `adaptive_{avg, max}_pool` variants : cuda & cpu (#88906)

khushi-411 committed 1 year ago

7cd900eb

[FSDP][optim_state_dict][3/N] Support use_orig_param optim_state_dict (non-broadcast version) (#89900)

fegin committed 1 year ago

043de8d1

[ROCm] add case for FP32MatMulPattern skip property (#84077)

jaglinux committed 1 year ago

0a4e4de5

[ao][fx] fixing public v private match_utils.py (#88396)

HDCharles committed 1 year ago

79156c11

[ao][fx] public v private convert.py (#88394)

HDCharles committed 1 year ago

a856557b

[FSDP][1/N] `fully_shard` state dict (#90767)

awgu committed 1 year ago

b3d49c2f

[reland][inductor] Update TIMM skip list (#90762)

desertfire committed 1 year ago

ad4189c8

[Dynamo] Supports two torch.distributed.* functions (#90683)

yanboliang committed 1 year ago

5c133c57

[stateless] fix functional call docs (#90476)

samdow committed 1 year ago

21fc2828

Make at::outer SymInt-aware (#90714)

jbschlosser committed 1 year ago

4a5f4416

Make functional inverse for squeeze_copy SymInt-aware (#90697)

jbschlosser committed 1 year ago

3f14c705

Revert "Reland "Add heirachical module names to torchFX graph.node" (#90205)"

pytorchmergebot committed 1 year ago

1119d2fa

Enable inductor perf test on GCP A100 (#90322)

weiwangmeta committed 1 year ago

1439ebd8

Fix mps constant pad (#89864)

qqaatw committed 1 year ago

544756ae

[inductor] Enable test_torch (#90518)

desertfire committed 1 year ago

7035bcdd

Update cuSPARSE usage for CUDA 12.0 (#90765)

IvanYashchuk committed 1 year ago

0d5c849d

Fix FSDP checkpoint tests (#90745)

mrshenli committed 1 year ago

d4dda519

[inductor] Pattern match cat->view*->pointwise and hoist pointwise (#90743)

bertmaher committed 1 year ago

a76032d8

[Fix]: Add missing std::vector reserve in aten and torch/csrc (#90627)

Skylion007 committed 1 year ago

da8f539e

[functorch] Refactor life handle storage (#90317)

zou3519 committed 1 year ago

4d494986

Move private forward grad mode helpers to torch.autograd.forward_ad (#90240)

zou3519 committed 1 year ago

24c3ad78

Newer Older