Multi-hop Question Answering
Benchmarks
Dataset NameSOTA methodMetricTrendResultsLast Updated
69.7F1
6
Jun 30, 2026
73.2F1
6
Jun 30, 2026
30EM
6
Jun 25, 2026
75Initial Token-level F1
6
Jun 23, 2026
58.2Exact Match (EM)
6
Jun 4, 2026
18.5EM
6
Jun 4, 2026
50Exact Match (EM)
6
Jun 4, 2026
46.3Exact Match
6
Jun 4, 2026
67.6Exact Match (EM)
6
Jun 4, 2026
38.36Accuracy
6
Jun 4, 2026
35.18Accuracy
6
Jun 4, 2026
57.21Accuracy
6
Jun 4, 2026
0.45P-hat
6
May 22, 2026
0.5p-hat
6
May 22, 2026
68p-hat
6
May 22, 2026
34.84Strict Hit Rate
6
May 8, 2026
57.9F1 (token-overlap)
6
Apr 30, 2026
25.5F1 (token-overlap)
6
Apr 30, 2026
43.6F1 (token-overlap)
6
Apr 30, 2026
70.3F1 (Token Overlap)
6
Apr 30, 2026
70.11LM Score
6
Apr 7, 2026
57.7Recall
6
Mar 4, 2026
10.6MBE
6
Mar 4, 2026
28.4MBE
6
Mar 4, 2026
49.6Exact Match (EM)
6
Feb 27, 2026