Our new X account is live! Follow @wizwand_team for updates
Home
/
Benchmarks
Scientific Reasoning on ARC Easy
Loading...
73.4
Accuracy
Sink Attention
62.896
65.623
68.35
71.077
Feb 1, 2026
Accuracy
Updated 3d ago
Evaluation Results
Method
Method
Links
Accuracy
Sink Attention
Model Size=2B, Trainin...
2026.02
73.4
Gated Attention
Model Size=2B, Trainin...
2026.02
73.2
Sink Attention
Model Size=2B, Trainin...
2026.02
73.06
Vanilla Attention
Model Size=2B, Trainin...
2026.02
72.8
Gated Attention
Model Size=2B, Trainin...
2026.02
72.6
Vanilla Attention
Model Size=2B, Trainin...
2026.02
72.26
Sink Attention
Model Size=1B, Trainin...
2026.02
68.81
Vanilla Attention
Model Size=1B, Trainin...
2026.02
68.8
Sink Attention
Model Size=1B, Trainin...
2026.02
68.73
Vanilla Attention
Model Size=1B, Trainin...
2026.02
68.56
Gated Attention
Model Size=1B, Trainin...
2026.02
68.35
Gated Attention
Model Size=1B, Trainin...
2026.02
67.25
Vanilla Attention
Model Size=0.6B, Train...
2026.02
65.28
Gated Attention
Model Size=0.6B, Train...
2026.02
64.86
Gated Attention
Model Size=0.6B, Train...
2026.02
63.97
Vanilla Attention
Model Size=0.6B, Train...
2026.02
63.8
Sink Attention
Model Size=0.6B, Train...
2026.02
63.8
Sink Attention
Model Size=0.6B, Train...
2026.02
63.3
Feedback
Search any
task
Search any
task