Our new X account is live! Follow @wizwand_team for updates
WorkDL logo mark

Long-context language understanding on LongBench v2

46.32Overall Accuracy

HyLRA

8.900818.615428.3338.0446May 28, 2025Jul 9, 2025Aug 21, 2025Oct 2, 2025Nov 14, 2025Dec 26, 2025Feb 7, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.01
46.3249.4844.3752.7843.7240.74-----------
2026.01
45.7348.9643.7352.2244.1937.96-----------
2026.01
45.0348.9642.615042.7941.2-----------
2026.01
44.3348.4441.848.894045.37-----------
2026.01
42.5443.7541.85039.0737.04-----------
2026.01
40.7643.7538.9145.5638.1437.96-----------
2026.02
33.59-----50.8534.7132.9525.9837.7828.57-----
2026.02
33.07-----52.5434.7134.0926.7737.2119.67-----
2025.05
28.6-----------41.736.43526.526.5
2026.02
28.43-----37.2925.6228.4125.231.1130.16-----
2026.02
25.84-----27.1216.5336.3625.226.6728.57-----
2026.02
25.5-----23.731.628.2242520-----
2026.02
24.6-----35.5919.4926.1422.0528.8923.81-----
2025.05
24.2-----------2530.33525.318.8
2026.02
19.9-----18.621.521.622.215.614.3-----
2026.02
17.91-----16.951913.6192019.05-----
2026.02
16.7-----18.6410.7414.7720.4728.8912.7-----
2026.02
13.92-----15.2511.5711.3614.1717.7817.46-----
2026.02
13.12-----18.6415.712.58.6611.1114.29-----
2026.02
10.34-----8.4715.77.957.8713.337.94-----