List your company

MQMTAR on MQMTAR OOD lengths 2x 4x 16x 64x 256x 1024x

100Exact Match AccuracySoftmax
-4235077Jun 19, 2025

Evaluation results

MethodLinks
SoftmaxLayers (L)=4, OOD Sequence Length Scale=2x2025.06100
SoftmaxLayers (L)=4, OOD Sequence Length Scale=4x2025.06100
SSMaxLayers (L)=4, OOD Sequence Length Scale=2x2025.06100
EntmaxLayers (L)=4, OOD Sequence Length Scale=2x2025.06100
EntmaxLayers (L)=4, OOD Sequence Length Scale=4x2025.06100
ASEntmaxLayers (L)=4, OOD Sequence Length Scale=2x2025.06100
ASEntmaxLayers (L)=4, OOD Sequence Length Scale=4x2025.06100
Top-K (K=32)Layers (L)=4, OOD Sequence Length Scale=2x2025.0699.9
SSMaxLayers (L)=4, OOD Sequence Length Scale=4x2025.0699.9
ASEntmaxLayers (L)=4, OOD Sequence Length Scale=16x2025.0699.7
SSMaxLayers (L)=4, OOD Sequence Length Scale=16x2025.0699.6
ASEntmaxLayers (L)=4, OOD Sequence Length Scale=64x2025.0699.6
SoftmaxLayers (L)=4, OOD Sequence Length Scale=16x2025.0699.5
EntmaxLayers (L)=4, OOD Sequence Length Scale=16x2025.0699.2
ASEntmaxLayers (L)=4, OOD Sequence Length Scale=256x2025.0699
SSMaxLayers (L)=4, OOD Sequence Length Scale=64x2025.0698.3
SoftmaxLayers (L)=4, OOD Sequence Length Scale=64x2025.0697.8
ASEntmaxLayers (L)=4, OOD Sequence Length Scale=1024x2025.0695.3
EntmaxLayers (L)=4, OOD Sequence Length Scale=64x2025.0692.7
SSMaxLayers (L)=4, OOD Sequence Length Scale=256x2025.0690.6
SoftmaxLayers (L)=4, OOD Sequence Length Scale=256x2025.0680.2
EntmaxLayers (L)=4, OOD Sequence Length Scale=256x2025.0666.8
SSMaxLayers (L)=4, OOD Sequence Length Scale=1024x2025.0626.7
EntmaxLayers (L)=4, OOD Sequence Length Scale=1024x2025.069.3
Top-K (K=32)Layers (L)=4, OOD Sequence Length Scale=4x2025.066.2
SoftmaxLayers (L)=4, OOD Sequence Length Scale=1024x2025.063
Top-K (K=32)Layers (L)=4, OOD Sequence Length Scale=16x2025.060
Top-K (K=32)Layers (L)=4, OOD Sequence Length Scale=64x2025.060
Top-K (K=32)Layers (L)=4, OOD Sequence Length Scale=256x2025.060
Top-K (K=32)Layers (L)=4, OOD Sequence Length Scale=1024x2025.060