Generative Retrieval
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
0.003Reconstruction Loss
14
Mar 12, 2026
0.772R@10
11
Mar 4, 2026
0.775R@10
11
Mar 4, 2026
69.3R@1
11
Mar 4, 2026
69.6R@1
11
Mar 4, 2026
0.033NDCG@10
10
May 14, 2026
30.13Recall@10
8
Jun 8, 2026
5.75Hallucination Rate
8
Apr 29, 2026
4.32Recall@10
8
Apr 16, 2026
1.44Recall@10
8
Apr 16, 2026
2.54Recall@10
8
Apr 16, 2026
2.23Hit Rate @ 1
8
Mar 12, 2026
50.52Hit Rate @1
8
Mar 12, 2026
35.72Average Text Performance
7
May 13, 2026
35.72Avg Text Performance
7
May 13, 2026
35.72Avg Text Perf.
7
May 13, 2026
0.447L_recon
7
Mar 4, 2026
35.2Unordered Recall@10
6
Feb 26, 2026
53.9Unordered Recall@10
6
Feb 26, 2026
62.5Unordered Recall@10
6
Feb 26, 2026
42.5nDCG@10 (ArguAna)
5
Apr 28, 2026
15.79HR@50
4
Jul 7, 2026
0.622AUC
4
Jun 26, 2026
0.656AUC
4
Jun 26, 2026
58.1AUC
4
Jun 26, 2026