Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

MAS

Benchmarks

Task NameDataset NameSOTA ResultTrend
Authorship segmentationMAS (entire corpus)
SBDA @ 0.345.7
18
Safety DetectionMAS Gemma-2-9B (Current)
Accuracy95.35
10
Safety DetectionMAS Qwen-2.5-7B (Current)
Accuracy97.46
10
Safety DetectionMAS LLaMA-3.1-8B (Current)
Accuracy99.15
10
Safety DetectionMAS Qwen-2.5-7B (Lookahead)
Accuracy96.77
8
Safety DetectionMAS LLaMA-3.1-8B (Lookahead)
Accuracy92.31
8
Showing 6 of 6 rows