Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Weak Scaling on BERT-Base (train)

3,707.01Memory (MB)

Sequence parallelism

3,260.80686,272.67849,284.5512,296.4216May 26, 2021
Updated 4d ago

Evaluation Results

MethodLinks
2021.05
3,707.019,340.13
2021.05
3,707.399,752.61
2021.05
4,670.6413,144.16
2021.05
4,993.4314,195.17
2021.05
6,601.8818,243.82
2021.05
8,175.9319,879.27
2021.05
8,477.289,946.15
2021.05
8,477.539,261.04
2021.05
8,478.7613,938.22
2021.05
8,481.2621,269.91
2021.05
8,490.7526,401.64
2021.05
9,520.4715,510.19
2021.05
10,536.3821,625.51
2021.05
12,232.5220,701.96
2021.05
14,862.0922,330.5