List your company

Multiple-choice QA on PIQA, OpenBookQA, HellaSwag, and ARC

68.39PIQA AccuracyNexusFormer
50.117254.861159.60564.3489Apr 21, 2026

Evaluation results

MethodLinks
NexusFormerSize=640M2026.0468.3933.64559.6730.7247.48
TokenFormerSize=640M2026.0466.533439.9351.5127.943.97
NexusFormerSize=240M2026.0466.273339.8950.828.6743.73
TokenFormerSize=240M2026.0465.9932.639.9452.3528.6743.91
TokenFormerSize=170M2026.0465.232.638.446.827.242.04
NexusFormerSize=480M2026.0464.4730.641.4150.9727.7343.04
NexusFormerSize=170M2026.0464.4232.435.5349.3326.8841.71
Qwen3Size=640M2026.0460.342929.7244.7424.8337.73
Qwen3Size=480M2026.0459.7430.432.1842.9725.3438.13
TokenFormerSize=480M2026.0456.9128.831.5340.3625.9336.71
Qwen3Size=170M2026.0453.322526.4234.8922.4432.41
Transformer++Size=240M2026.0451.92826.2727.5726.1931.99
Qwen3Size=240M2026.0451.412526.2632.8323.8131.86
Transformer++Size=640M2026.0451.1425.425.928.224.230.97
Transformer++Size=480M2026.0451.0327.427.2829.2924.1431.83
Transformer++Size=170M2026.0450.822625.527.5726.8831.35