List your company

Zero-shot Task Evaluation on 9 Downstream Tasks Utility

54.7Average AccuracyDAPT (nontoxic)
44.50847.15449.852.446Feb 8, 2022

Evaluation results

MethodLinks
DAPT (nontoxic)Backbone=1.3B LM, Evaluation Protocol=zero-shot2022.0254.7
SGEAT (heuristic)Backbone=1.3B LM, Evaluation Protocol=zero-shot, Prompt Design=heuristic2022.0254.7
Jigsaw (nontoxic)Backbone=1.3B LM, Evaluation Protocol=zero-shot2022.0254.6
SGEAT (standard)Backbone=1.3B LM, Evaluation Protocol=zero-shot, Prompt Design=standard2022.0254.6
SGEAT (augmented)Backbone=1.3B LM, Evaluation Protocol=zero-shot, Prompt Design=augmented2022.0254.4
SGEAT + Rejection SamplingBackbone=1.3B LM, Evaluation Protocol=zero-shot2022.0254.4
Word BanningBackbone=1.3B LM, Evaluation Protocol=zero-shot2022.0254.3
Rejection SamplingBackbone=1.3B LM, Evaluation Protocol=zero-shot, Decoding Speed=4x slow2022.0254.3
DEXPERTSBackbone=1.3B LM, Evaluation Protocol=zero-shot, Decoding Speed=3x slow2022.0246.2
SGEAT + DEXPERTSBackbone=1.3B LM, Evaluation Protocol=zero-shot2022.0244.9