General Ability Suite
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
General Ability Suite (MMLU, PIQA, ARC-E, ARC-C, BoolQ, WinoGrande, HellaSwag, TruthfulQA)
65MMLU Accuracy
20
General Ability Suite (ARC, HellaSwag, PIQA, BoolQ, WinoGrande, COPA, OBQA, SciQ) various (test)
36.4ARC-C Accuracy
19
General Ability Suite (C-QA, T-QA, LAM, MMLU, L-Code)
48.1Average Score
16
General Ability Suite ARC, HellaSwag, PIQA, BoolQ, WinoGrande, COPA, OBQA, SciQ
—ARC-C Accuracy
0