Arc, HellaSwag, LAMBADA, PIQA, WinoGrande
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
Arc-Challenge, Arc-Easy, HellaSwag, LAMBADA-openai, LAMBADA-standard, PIQA, and WinoGrande Zero-Shot
76.71Accuracy (0-shot)
123
ARC-C, ARC-E, HellaSwag, LAMBADA, PIQA, WinoGrande (test)
56.81Mean Score
5