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

Reward Modeling on JudgeBench (test)

82Overall

Qwen3-30B-A3B

40.451.26272.8Jan 28, 2026Jan 29, 2026Jan 30, 2026Jan 31, 2026Feb 1, 2026Feb 2, 2026Feb 4, 2026
Updated 4d ago

Evaluation Results

MethodLinks
2026.02
8273.489.882.195.284.6
2026.02
81.474.685.785.790.583.8
2026.02
8190.983.787.561.983.4
2026.02
79.166.986.791.190.582.9
2026.02
78.859.182.780.492.975.8
2026.02
78.87680.688.170.581.4
2026.02
78.168.172.487.884.279.8
2026.02
77.271.473.587.576.280
2026.01
76.3-----
2026.02
75.779.970.487.56574.4
2026.02
75.765.687.882.176.280.3
2026.01
75.7-----
2026.01
75.1-----
2026.01
73.7-----
2026.01
70.4-----
2026.02
7055.880.678.685.776.8
2026.01
68.6-----
2026.01
68.2-----
2026.01
67.5-----
2026.01
66.9-----
2026.01
66.7-----
2026.01
66.3-----
2026.01
65.5-----
2026.02
6559.766.383.95067.8
2026.02
64.862.366.366.164.362.9
2026.01
63.1-----
2026.02
62.66157.173.266.772.9
2026.02
62.559.164.376.85066.3
2026.01
60.9-----
2026.01
60-----
2026.01
60-----
2026.02
59.850.654.17559.566.2
2026.01
59.7-----
2026.01
59.1-----
2026.01
57.7-----
2026.01
56.6-----
2026.01
54.6-----
2026.01
48.6-----
2026.01
42.3-----
2026.01
42-----