MAS
Benchmarks
TaskDatasetSOTA resultTrendResults
MAS (entire corpus)
45.7SBDA @ 0.3
18
MAS Gemma-2-9B (Current)
95.35Accuracy
10
MAS Qwen-2.5-7B (Current)
97.46Accuracy
10
MAS LLaMA-3.1-8B (Current)
99.15Accuracy
10
MAS Qwen-2.5-7B (Lookahead)
96.77Accuracy
8
MAS LLaMA-3.1-8B (Lookahead)
92.31Accuracy
8