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

Recall-intensive Retrieval on SWDE, SQuAD, FDA, TQA, NQ, and Drop Suite

76.85Performance on SWDE

Qwen3-8B

5.55824.066542.57561.0835Oct 8, 2025Nov 15, 2025Dec 24, 2025Jan 31, 2026Mar 11, 2026Apr 18, 2026May 27, 2026
Updated 6d ago

Evaluation Results

MethodLinks
2025.10
76.8559.8973.4872.6345.0136.6160.75
2025.10
75.4554.2882.3877.4948.4338.2862.72
2025.10
72.0754.2282.9274.1743.2433.8860.08
2025.10
71.750.6976.0270.5642.125.7856.14
2025.10
69.8253.8175.371.8642.4132.3457.59
2025.10
67.6755.1673.8473.0542.134.0257.64
2025.10
65.3254.0175.373.1642.7627.5556.35
2025.10
64.4850.1571.4871.8640.1329.2354.56
2025.10
59.5149.3148.3275.0637.8530.0950.02
2025.10
51.5546.6256.8669.8537.0328.748.44
2026.05
50.141.757.663.928.322.544
2025.10
44.6147.2662.5871.9837.4737.4750.23
2026.05
44.541.246.464.827.821.741.1
2025.10
43.346.9337.0671.5633.3925.342.92
2026.05
42.342.855.265.926.722.942.6
2025.10
38.240.450.763.324.823.340.1
2025.10
37.84151.162.824.822.940.1
2025.10
35.639.75260.124.622.239
2025.10
35.139.850.56022.321.738.2
2025.10
3339.250.557.723.520.237.3
2025.10
29.53852.258.322.521.637
2026.05
29.337.918.26423.52232.5
2026.05
27.436.424.263.323.821.632.8
2025.10
25.434.823.7602019.830.6
2026.05
22.236.419.463.721.822.331
2026.05
20.636.12161.722.420.330.3
2025.10
19.133.625.36120.819.229.8
2025.10
17.930.918.453.917.318.626.2
2025.10
1428.5754.416.217.322.9
2025.10
9.825.83.754.314.917.421
2025.10
8.325.34.851.214.216.920.1