List your company

Multiple Choice Question Answering on MMLU, ARC-Easy, HellaSwag, PIQA, OpenBookQA, WinoGrande

24.3MMLU AccuracyBaseline
22.84423.22223.623.978May 13, 2026

Evaluation results

MethodLinks
BaselineDepth=6, Hidden width=7682026.0524.333.127.658.925.252.4
4-viumDepth=24, Hidden width=768, beta=0.152026.0523.13732.363.928.453.5
4-viumDepth=6, Hidden width=768, beta=0.152026.052332.828.559.123.250.7
BaselineDepth=48, Hidden width=7682026.052337.435.365.328.850.1
BaselineDepth=12, Hidden width=7682026.0522.935.128.861.824.451.9
4-viumDepth=12, Hidden width=768, beta=0.152026.0522.933.529.261.42848.2
BaselineDepth=24, Hidden width=7682026.0522.935.831.363.927.850.6
BaselineDepth=36, Hidden width=7682026.0522.938.333.863.328.251.4
4-viumDepth=36, Hidden width=768, beta=0.152026.0522.939.335.564.928.451.7
4-viumDepth=48, Hidden width=768, beta=0.152026.0522.939.73766.629.653.7