Zero-shot Evaluation on 8 Downstream Tasks (ARC-E, ARC-C, HellaSwag, PIQA, WG, OBQA, SciQ, BoolQ)
74.82ARC-Easy AccuracyLooped Hybrid (Full+GDN)
Evaluation results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Looped Hybrid (Full+GDN)Parameters=1.3B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 74.82 | 41.63 | 61.04 | 75.93 | 69.52 | 41.3 | 75.4 | 62.04 | 62.89 | |
| Looped Hybrid (GDN+DSA)Parameters=1.3B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 72.44 | 39.33 | 58.84 | 73.98 | 67.13 | 39.7 | 74.9 | 59.77 | 60.73 | |
| Looped KDAParameters=1.3B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 71.57 | 38.62 | 57.99 | 73.53 | 66.42 | 39.3 | 74.6 | 58.98 | 60.14 | |
| Looped GDNParameters=1.3B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 71.28 | 38.33 | 57.73 | 73.37 | 66.26 | 39.1 | 74.3 | 58.78 | 59.92 | |
| Looped Hybrid (Full+DSA)Parameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 71.13 | 37.28 | 56.84 | 73.24 | 64.52 | 38.9 | 73.4 | 58.73 | 59.28 | |
| Looped Hybrid (Full+Window)Parameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 70.93 | 37.12 | 56.68 | 73.12 | 64.34 | 38.8 | 73.3 | 58.56 | 59.13 | |
| Looped Transformer (ref)Parameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 70.83 | 37.54 | 57.06 | 72.43 | 65.83 | 38.6 | 74.1 | 57.83 | 59.27 | |
| Looped DSAParameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 69.93 | 36.93 | 56.38 | 71.94 | 64.87 | 37.7 | 72.9 | 57.42 | 58.54 | |
| Looped Hybrid (Full+GDN)Parameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 69.82 | 37.34 | 55.83 | 72.62 | 64.61 | 38.9 | 73.3 | 57.74 | 58.65 | |
| Looped Mamba2Parameters=1.3B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=false2026.05 | 69.47 | 36.63 | 55.94 | 72.68 | 64.37 | 38.2 | 73 | 57.03 | 58.43 | |
| Looped NSAParameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 69.02 | 35.97 | 55.08 | 71.52 | 64.03 | 37.2 | 72.2 | 56.53 | 57.72 | |
| Looped WindowParameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 68.43 | 35.47 | 54.87 | 71.32 | 63.23 | 36.9 | 71.7 | 55.87 | 57.23 | |
| TransformerParameters=1.3B, Token Budget=100B, Loops (K)=42026.05 | 67.52 | 33.84 | 52.47 | 71.03 | 61.48 | 36.6 | 71.3 | 54.02 | 56.04 | |
| Looped Hybrid (GDN+DSA)Parameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 67.43 | 34.53 | 53.42 | 70.63 | 62.92 | 37 | 71.2 | 55.13 | 56.53 | |
| Looped Transformer (ref)Parameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 67.13 | 34.67 | 53.29 | 70.58 | 62.83 | 38.2 | 73.6 | 54.87 | 56.42 | |
| Looped GDNParameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 66.43 | 33.89 | 52.62 | 70.27 | 61.48 | 36.4 | 70.5 | 54.13 | 55.74 | |
| Looped DSAParameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 66.37 | 33.82 | 52.53 | 70.23 | 61.42 | 36.4 | 70.4 | 54.07 | 55.67 | |
| Looped KDAParameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=true2026.05 | 66.12 | 33.63 | 52.37 | 70.13 | 61.22 | 36.2 | 70.2 | 53.92 | 55.49 | |
| Looped NSAParameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 65.57 | 32.74 | 51.43 | 70.04 | 60.32 | 36 | 69.5 | 53.13 | 54.84 | |
| Looped Hybrid (Full+DSA)Parameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 65.53 | 32.34 | 51.42 | 70.04 | 58.63 | 36.2 | 69.4 | 53.32 | 54.62 | |
| Looped Hybrid (Full+Window)Parameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 65.32 | 32.13 | 51.23 | 69.86 | 58.42 | 36 | 69.2 | 53.13 | 54.43 | |
| Looped Mamba2Parameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=false2026.05 | 64.53 | 31.82 | 49.87 | 69.74 | 58.63 | 35.6 | 68.8 | 51.83 | 53.86 | |
| Looped WindowParameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 64.23 | 31.53 | 48.83 | 69.82 | 57.34 | 35.8 | 68.5 | 51.23 | 52.17 | |
| TransformerParameters=0.6B, Token Budget=100B, Loops (K)=42026.05 | 63.09 | 30.72 | 47.43 | 69.53 | 56.24 | 35.6 | 68.2 | 50.07 | 51.34 | |
| Looped DeltaNetParameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=false, Δ=true2026.05 | 60.47 | 28.53 | 44.22 | 67.87 | 53.24 | 33.8 | 65.5 | 49.13 | 50.12 | |
| Looped HGRN2Parameters=0.6B, Token Budget=100B, Loops (K)=4, D-Gate=true, Δ=false2026.05 | 59.82 | 27.93 | 43.17 | 67.34 | 52.13 | 33.4 | 65.2 | 48.53 | 49.69 |