pytorch/benchmark

Pull Requests Commits

Krovatkin committed 4 years ago

d3426314

add a timeout arg

Krovatkin committed 4 years ago

0ae0d2f3

Krovatkin committed 4 years ago

0697dee6

Merge branch 'main' into wconstab/ltc

Jiewen Tan committed 4 years ago

aa45d35d

quick fix (#599)

Krovatkin committed 4 years ago

Verified 2204b175

hack demucs (#601)

Krovatkin committed 4 years ago

Verified d5ca0c71

Added lazy tensor testing CI. (#600)

xuzhao9 committed 4 years ago

bb7256f2

Fixes squeezenet hierarchical batching. (#609)

xuzhao9 committed 4 years ago

b0a71ba7

Set the correct train batch size for squeezenet. (#574)

xuzhao9 committed 4 years ago

cee0ce46

Fix the runner temp interface. (#607)

xuzhao9 committed 4 years ago

c2e36b87

Remove V0 workflow. Use CUDA 11.3 for V1 workflow. (#596)

xuzhao9 committed 4 years ago

26229b6e

Use opacus version < 1.0 to mitigate BC-breaking API change (#605)

xuzhao9 committed 4 years ago

3e7bf736

Update the train_bs and eval_bs for super_slomo. (#595)

xuzhao9 committed 4 years ago

1dbe4bc7

Improve test.py EXCLUDELIST to disable by model, test, and device (#592)

aaronenyeshi committed 4 years ago

5e898020

Fix train batch size for attention model. (#593)

xuzhao9 committed 4 years ago

488d2071

Fix train batch size for pytorch_struct. (#591)

xuzhao9 committed 4 years ago

121ea023

Fix background matting batch size. (#594)

xuzhao9 committed 4 years ago

5b9b32f9

Fix the train options of LearningToPaint (#581)

aaronenyeshi committed 4 years ago

a5213f1c

Fix the train arch of DLRM (#580)

aaronenyeshi committed 4 years ago

149410d8

Support skipping tests on a blacklist in test.py (#584)

aaronenyeshi committed 4 years ago

75c0da2e

Only load the module specified in run.py argument (#577)

xuzhao9 committed 4 years ago

1941a66f

Use the original batch size from mobilenet_v2 paper. (#576)

xuzhao9 committed 4 years ago

a8023c67

Fix the train batch size of alexnet. (#575)

xuzhao9 committed 4 years ago

b4c316b5

Adding maskrcnn model from torchvision (#547)

xuzhao9 committed 4 years ago

39b5b6d3

Fix the fastNLP model train batch size. (#571)

xuzhao9 committed 4 years ago

0d1da2b8

Fix the jit code for timm models. (#572)

xuzhao9 committed 4 years ago

dc342679

Correct LearningToPaint to match the wild (#560)

aaronenyeshi committed 4 years ago

17cd2f96

Add time_lazy.py script for lazy tensor measurement

wconstab committed 4 years ago

4fbf86a1

Make bert_pytorch support lazy device

wconstab committed 4 years ago

6f602558

Add check_lazy.py to sweep models and collect stats

wconstab committed 4 years ago

31e733d3

Older