pytorch/benchmark

Pull Requests Commits

remove memory copies; sync dalle2 with upstream

FindHao committed 2 years ago

36e0ceb0

Create the new pytorch nightly userbenchmark for V3 (#1482)

xuzhao9 committed 2 years ago

2ad17632

Improve the run.py output message

xuzhao9 committed 2 years ago

268fa024

Improve the model-stableness workflow to dump stableness result in YAML (#1483)

xuzhao9 committed 2 years ago

355dc800

Upload benchmark metrics to S3 bucket (#1484)

xuzhao9 committed 2 years ago

cc8b9f56

Enable the a10g instance ci (#1476)

xuzhao9 committed 2 years ago

83a316df

Add LLAMA (#1446)

msaroufim committed 2 years ago

c78f1f32

Replace the numpy.meshgrid() with more efficient torch.meshgrid() (#1475)

CuiJinku committed 2 years ago

d8e5325b

Fix a bug in hf_bert distributed (#1477)

xuzhao9 committed 2 years ago

0f02ca65

Remove all git-lfs files (#1474)

xuzhao9 committed 2 years ago

500701aa

Remove git lfs and add readme on Amazon S3 data host (#1468)

xuzhao9 committed 2 years ago

deb23fdc

Add stable diffusion model benchmark (#1204)

xuzhao9 committed 2 years ago

db7b5ad8

Make install.py work on canary models (#1456)

xuzhao9 committed 2 years ago

afd41515

Update optimizer accessors to influence dependencies (#1460)

janeyx99 committed 2 years ago

aafc8507

Remove old generating files from util (#1455)

janeyx99 committed 2 years ago

fc9fb89b

Ignore pkls and zips and pts in .gitignore (#1462)

janeyx99 committed 2 years ago

74a0d3fe

Add get/set_optimizer to E2EBenchmarkModel API (#1458)

janeyx99 committed 2 years ago

5b4b5327

Enabled fuser for jit backend and added fuser3 for llga path (#1449)

chuanqi129 committed 2 years ago

0990ae7c

Fix detectron2 models on cpu-only torch (#1430)

YJ Shi committed 2 years ago

a22c89c1

Add instructpix2pix model (#1451)

xuzhao9 committed 2 years ago

560a4ee5

Add instructions to test model after submitting (#1447)

msaroufim committed 2 years ago

e080877d

GeForce clock query fix Issue #1440 (#1445)

Mike Schneider committed 2 years ago

9ab7a6f2

enable peak memory measurement by default (#1392)

FindHao committed 2 years ago

1604885b

Add the MPS device to eval mode. (#1435)

kulinseth committed 2 years ago

e9941f39

Clarify instllation step (#1426)

spzala committed 2 years ago

f33ce935

Fix the case when workflow is empty (#1438)

xuzhao9 committed 2 years ago

c9e3dd95

Add the --mps_only option. (#1432)

kulinseth committed 2 years ago

1635a749

Disable the nvfuser nightly CI (#1431)

xuzhao9 committed 2 years ago

4fb6838c

Fix torchrec_dlrm model and enable fbcode test (#1428)

xuzhao9 committed 2 years ago

08fc5914

Use the correct batch size for timm ViT to improve stableness (#1419)

xuzhao9 committed 2 years ago

06538fd1

Older