pytorch/benchmark

Pull Requests Commits

fix bug for gpu detection

FindHao committed 3 years ago

711b1cfd

replace torchbench impelmentation with torchrec that shards the module (#1401)

colin2328 committed 3 years ago

9806a7ec

Fix vision model on inductor and profile on CPU only machines (#1415)

xuzhao9 committed 3 years ago

7bd50b01

change fastNLP_Bert's task name (#1410)

FindHao committed 3 years ago

3fe6446f

Add detailed task names (#1411)

Yueming Hao committed 3 years ago

74895429

Add channels_last for all torchbench models and enable it with run.py (#1371)

chuanqi129 committed 3 years ago

d64579cc

Pin pygame version to fix soft_actor_critic (#1412)

xuzhao9 committed 3 years ago

43ee38aa

Match AITModule on ait inputs and fx inputs (#262)

tissue3 committed 3 years ago

1bbb66c4

Fix type in iter rounds (#1408)

xuzhao9 committed 3 years ago

411117d8

Add mimo_cmf_30x

bertmaher committed 3 years ago

bf02c40a

Upgrade to upload-artifact v3 to resolve GitHub warnings (#1407)

xuzhao9 committed 3 years ago

91d91104

Force creation of the conda environment and upgrade dependencies (#1406)

xuzhao9 committed 3 years ago

adce3da2

Add CI to main branch commits (#1404)

xuzhao9 committed 3 years ago

916ab70b

Fix A100 workflow (#1403)

xuzhao9 committed 3 years ago

4a8f9927

Upgrade to CUDA 11.7 and remove V1 workflows (#1398)

xuzhao9 committed 3 years ago

11f7141d

Fix a bug in train correctness checking

xuzhao9 committed 3 years ago

0c99340f

Print grad values when not matching for debugging purpose (#1399)

brad-mengchi committed 3 years ago

1793f37f

Add model torchrec dlrm (#1397)

xuzhao9 committed 3 years ago

bd90678d

check for fx2trt only for cuda device (#1395)

snadampal committed 3 years ago

9bff3704

Fix cudagraph backend (#1393)

xuzhao9 committed 3 years ago

a9771829

Migrate fx2trt and torch_trt to use the backend option. (#1388)

xuzhao9 committed 3 years ago

d5801978

fix bug for short model executions (#1391)

FindHao committed 3 years ago

d166639f

Add support to AITemplate engine (#1386)

xuzhao9 committed 3 years ago

71440ac8

Run 15 rounds and collect the max delta (#1352)

xuzhao9 committed 3 years ago

597a6443

fix bug for record collections (#1366)

FindHao committed 3 years ago

7ab73738

Update the backend interface (#1384)

xuzhao9 committed 3 years ago

eb49327b

remove outdated argument (#1373)

FindHao committed 3 years ago

9b9bcbcb

call eval() for inference tests (#1380)

snadampal committed 3 years ago

410ab490

Remove the use of circleci. (#1258)

xuzhao9 committed 3 years ago

0b8aa0a8

drq: change view to reshape to support channels last (#1367)

chunyuan-w committed 3 years ago

ef39f8ca

Older