Zero-shot Reasoning and Knowledge Evaluation on Reasoning and Common-Knowledge Tasks
80.02HellaSWAGSDM
Evaluation Results
| Method | Links | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| SDMScale=8B (L13), Local layer=SWA, Global layer=SDM2026.07 | 80.02 | 75.3 | 78.52 | 53.05 | 80.47 | 44.6 | 62.05 | 44.97 | 70.6 | 67.13 | 59.11 | 24.39 | 25.93 | 57.81 | 28.66 | 56.84 | |
| FullAttnScale=8B (L13), Local layer=SWA, Global layer=FullAttn2026.07 | 79.33 | 73.64 | 78.1 | 50.82 | 79.98 | 43.8 | 64.83 | 47.94 | 68.88 | 71.83 | 55.23 | 24.39 | 22.94 | 58.73 | 29.34 | 56.65 | |
| GDNScale=8B (L13), Local layer=SWA, Global layer=GDN2026.07 | 79.1 | 73.24 | 79.15 | 52.27 | 81.01 | 43 | 60.17 | 43.4 | 66.83 | 74.98 | 55.36 | 18.29 | 22.6 | 57.24 | 28.81 | 55.7 | |
| SDMScale=1.4B (L08), Local layer=SWA, Global layer=SDM2026.07 | 62.04 | 62.12 | 63 | 34.76 | 74.27 | 36 | 49.65 | 37.76 | 20.72 | 62.78 | 26.61 | 9.76 | 8.23 | 27.24 | 2.73 | 38.51 | |
| FullAttnScale=1.4B (L08), Local layer=SWA, Global layer=FullAttn2026.07 | 61.47 | 60.54 | 61.56 | 33.56 | 73.83 | 35.8 | 53.55 | 39.62 | 19.57 | 62.42 | 24.24 | 8.54 | 7.7 | 24.54 | 2.81 | 37.98 | |
| Mamba2Scale=1.4B (L08), Local layer=SWA, Global layer=Mamba22026.07 | 61.31 | 62.67 | 62.88 | 34.08 | 73.67 | 36.4 | 50.28 | 36.25 | 20.31 | 62.75 | 23.48 | 9.76 | 8.09 | 25.03 | 3.03 | 38 | |
| GDNScale=1.4B (L08), Local layer=SWA, Global layer=GDN2026.07 | 61.27 | 61.56 | 62.49 | 34.59 | 73.83 | 36.8 | 50.42 | 36.68 | 19.33 | 63.03 | 23.68 | 8.54 | 7.76 | 26.02 | 2.96 | 37.93 |