HotpotQA
Benchmarks
Task NameDataset NameSOTA ResultTrendResults
HotpotQA
60.9Top-1 Accuracy
4
HotpotQA cross-episode setup
41.1Success Rate
4
HotpotQA
714Max System Prompt Tokens
4
HotPotQA
97.8Accuracy
4
HotpotQA
0.562Latency (s)
4
HotpotQA
61Accuracy
4
HotpotQA
15.2Joint Answer-Evidence Score
4
HotpotQA
33.2Evidence Score
4
HotpotQA Hard tier distractor (val)
34.7Mean F1
4
HotpotQA Easy tier distractor (val)
60.1Mean F1
4
HotpotQA Overall distractor (val)
47.3Mean F1
4
HotpotQA
51.53Accuracy (56K Context)
4
HotpotQA (test)
14.2Accuracy
4
HotpotQA
63Score
4
HotpotQA (val)
63.3Score (%)
4
HotpotQA
38.8Mean Performance
4
HotpotQA RAG v3
0.164Risk@30
4
HotpotQA
1Steps
4
HotpotQA
836Helpful Citations Count
4
HotpotQA
8,170ITC
4
HotpotQA
73.5Recall@10
4
HotpotQA
45CoverEM
4
HotpotQA
89.5Pairwise BLEU
4
HotpotQA
3.497R@1
4
HotpotQA
0.91Latency (s/query)
4