Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

Long-context language modeling evaluation on HELMET

0Average Sparsity

FullAttention

-2.642815.196133.03550.8739Feb 5, 2026
Updated 1mo ago

Evaluation Results

MethodLinks
2026.02
050.3348.8746.1143.8341.0746.04
2026.02
067.0967.4266.7865.7362.865.96
2026.02
061.1758.6756.7155.8149.7456.42
2026.02
054.7550.6648.444.9839.5947.68
2026.02
37.8853.1849.0546.4843.0135.5245.45
2026.02
38.6360.7358.1556.2255.1549.8756.02
2026.02
40.1650.8750.014643.8740.9246.33
2026.02
41.9666.5167.2266.3863.7561.1765.01
2026.02
43.0166.9667.6366.5564.4661.6665.45
2026.02
47.3553.949.7746.9644.4938.0946.64
2026.02
47.560.7958.155.5754.7849.4555.74
2026.02
47.5654.2250.0247.9245.1738.5147.17
2026.02
48.2266.1565.6164.9763.260.864.15
2026.02
48.6861.2258.2156.0755.3350.3756.24
2026.02
50.850.4148.5144.7343.6941.1745.7
2026.02
51.8650.9549.4547.1344.1241.0646.54
2026.02
55.266.7867.1965.2563.1260.0464.48
2026.02
56.2667.0166.665.7162.9760.764.6
2026.02
58.4264.0864.1163.1561.6459.362.46
2026.02
59.1552.5347.7245.7243.3537.0445.27
2026.02
59.5353.3749.3147.3544.6238.4846.63
2026.02
59.8859.9257.1555.5453.6647.8254.82
2026.02
61.0260.9558.4355.8654.2748.8555.67
2026.02
63.2144.1140.5940.636.3929.6638.27
2026.02
63.8550.2847.7944.1743.3140.1945.15
2026.02
64.0550.7948.5846.4843.8940.7746.1
2026.02
65.9844.9944.5143.2741.3138.7342.56
2026.02
66.0756.0254.4853.0150.7748.4252.54