pytorch/benchmark

Pull Requests Commits

Prevent eval of torchvision benchmarks in unsqueeze to dim 0 (#454)

aaronenyeshi committed 4 years ago

Verified 8c8deab5

Merge pull request #453 from eircfb/cudastreams

eircfb committed 4 years ago

Verified a5ed5b77

Remove the legacy directory that is not used anymore. (#457)

xuzhao9 committed 4 years ago

Verified 8176a031

Cleanup unnecessary files. (#455)

xuzhao9 committed 4 years ago

Verified 8f2d65b4

run.py cudastreams arg for testing utilization

eircfb committed 4 years ago

8021205a

Merge pull request #434 from eircfb/cudabert

eircfb committed 4 years ago

Verified 1ef02faa

Fix init parameter on jit path.

eircfb committed 4 years ago

29ca0dc7

Resolve Issues Raised By XuZhao9

eircfb committed 4 years ago

f31538ab

Increase corpus and batch for gpu utilization

eircfb committed 4 years ago

55ec549f

Add test to verify models and inputs are using the expected device (#443)

aaronenyeshi committed 4 years ago

Verified c8f21e99

Disable hf_DistilBert model cuda tests because of GPU contamination (#440)

xuzhao9 committed 4 years ago

Verified dfcad497

Update README.md for typos and code blocks (#432)

aaronenyeshi committed 4 years ago

Verified a26331df

Restore TorchBench reporting pipeline to working state. (#431)

Taylor Robie committed 4 years ago

Verified e1c6ccf8

Merge pull request #435 from pytorch/erichan1/resnet50_changebatchsize

erichan1 committed 4 years ago

Verified 6b1b0a6c

Merge `torch.jit.script` and `torch.jit._script_pdt` API. (#433)

xuzhao9 committed 4 years ago

Verified 0cf31e1e

change resnet50 benchmark batch size from 1 to 32

wconstab committed 4 years ago

3a6395ad

Use taskset when GOMP_CPU_AFFINITY is set (#430)

Taylor Robie committed 4 years ago

Verified a114b47c

Run tests and benchmarks in subprocesses for isolation (#423)

Taylor Robie committed 4 years ago

Verified b9ff79c6

fix yolo, and disable broken bert case (#425)

Taylor Robie committed 4 years ago

Verified 521cc4a1

Catch stdout and stderr to provide debug info when installs fail (and fix install) (#424)

Taylor Robie committed 4 years ago

Verified d6ce75fc

Add config readme to V1 score (#418)

xuzhao9 committed 4 years ago

Verified c56afbd5

Generate the bisection config and publish to the github actions artifact (#416)

xuzhao9 committed 4 years ago

Verified 4072951e

Improve the logs printed by the bisector. (#415)

xuzhao9 committed 4 years ago

Verified 5e81d8a4

Enable the V1 nightly run. (#414)

xuzhao9 committed 4 years ago

Verified 82e39f87

Add the sweep workflow for v1. (#413)

xuzhao9 committed 4 years ago

Verified 2cbcdee3

Fix CUDA version and the output dir. (#411)

xuzhao9 committed 4 years ago

Verified 791d115f

Add other deps to the v1 workflow (#410)

xuzhao9 committed 4 years ago

Verified e58e2e38

Fix errors in the workflow yaml file (#409)

xuzhao9 committed 4 years ago

Verified ebf5ead0

Add v1 nightly CI workflow (#407)

xuzhao9 committed 4 years ago

Verified 7b74f668

Fix pytorch_struct broken by more strict in-place operation restrictions. (#408)

xuzhao9 committed 4 years ago

Verified 1d88f16d

Newer Older