Information Retrieval on BRIGHT v1 (test)
49.1nDCG@10 (Avg)LATTICE++ (G-3.0-f)
Evaluation results
| Method | Links | ||||||||||||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| LATTICE++ (G-3.0-f)Ensemble=true, Strategy=LLM-guided Hierarchical Search, Search LLM=Gemini-3-Flash2025.10 | 49.1 | — | — | — | — | — | — | — | — | — | — | — | — | 55.7 | 73.2 | 67.1 | 45.3 | 58.3 | 52.4 | 45.5 | 48.2 | 41.9 | 32.6 | 51.2 | 38.3 | 16.2 | 46.9 | 51.9 | |
| DIVER v3Ensemble=true, Strategy=Iterative-retrieve-then-rerank2025.10 | 46.8 | — | — | — | — | — | — | — | — | — | — | — | — | 51.8 | 66 | 63.7 | 42.4 | 55 | 40.6 | 44.7 | 50.4 | 39.9 | 32.5 | 47.3 | 39.7 | 17.2 | 46.4 | 55.6 | |
| LATTICE (G-3.0-f)Ensemble=false, Strategy=LLM-guided Hierarchical Search, Search LLM=Gemini-3-Flash2025.10 | 46.7 | — | — | — | — | — | — | — | — | — | — | — | — | 55.8 | 69.4 | 65 | 48.8 | 61 | 53.9 | 42.2 | 50.4 | 33 | 22.9 | 43.2 | 34.6 | 13.7 | 44 | 46 | |
| LATTICE++ (G-2.5-f)Ensemble=true, Strategy=LLM-guided Hierarchical Search, Search LLM=Gemini-2.5-Flash2025.10 | 46.5 | — | — | — | — | — | — | — | — | — | — | — | — | 53.6 | 71.4 | 66.9 | 46.5 | 55.3 | 45.4 | 43.7 | 45.9 | 36.7 | 28.5 | 44.8 | 36.7 | 14.2 | 43.6 | 52.4 | |
| BGE-Reasoner-0928Ensemble=true, Strategy=Iterative-retrieve-then-rerank2025.10 | 46.4 | — | — | — | — | — | — | — | — | — | — | — | — | 52 | 68.5 | 66.4 | 40.6 | 53.1 | 43.2 | 44.1 | 47.8 | 35.3 | 29 | 41.6 | 40.7 | 17.2 | 46.5 | 58.4 | |
| DIVER v2Ensemble=true, Strategy=Iterative-retrieve-then-rerank2025.10 | 45.7 | — | — | — | — | — | — | — | — | — | — | — | — | 52.2 | 68 | 62.5 | 42 | 58.2 | 41.5 | 44.3 | 49.2 | 33.8 | 34.8 | 32.9 | 38.6 | 19.1 | 44.3 | 52.6 | |
| LATTICE (G-2.5-f)Ensemble=false, Strategy=LLM-guided Hierarchical Search, Search LLM=Gemini-2.5-Flash2025.10 | 43 | — | — | — | — | — | — | — | — | — | — | — | — | 52 | 66.3 | 63 | 47.4 | 54 | 47.6 | 37.6 | 48.2 | 26.9 | 19.9 | 34 | 32.6 | 12 | 38 | 47.9 | |
| BGE-Reasoner (w/o ensembling)Ensemble=false, Strategy=Iterative-retrieve-then-rerank2025.10 | 42.2 | — | — | — | — | — | — | — | — | — | — | — | — | 51.1 | 64.3 | 62.6 | 45.3 | 52.7 | 44.3 | 41.5 | 46.9 | 22.3 | 14.4 | 30.2 | 34.5 | 14.2 | 41.7 | 47.5 | |
| ReasonRankEnsemble=false, Strategy=Retrieve-then-rerank2025.10 | 40.8 | — | — | — | — | — | — | — | — | — | — | — | — | 46.8 | 62.7 | 55.5 | 36.7 | 54.6 | 35.7 | 38 | 44.8 | 27.5 | 29.5 | 25.6 | 35.5 | 14.4 | 42 | 50.1 | |
| XRR2Ensemble=false, Strategy=Retrieve-then-rerank, Pipeline=GPT-4 + BM25 → Gemini-2.5-Flash2025.10 | 40.3 | — | — | — | — | — | — | — | — | — | — | — | — | 47.4 | 63.1 | 58.2 | 38.5 | 52.9 | 37.1 | 37.6 | 44.6 | 28.4 | 21.9 | 35 | 31.8 | 15.7 | 34.4 | 45.5 | |
| RaDeR-7BEvaluation Protocol=Retriever with QwenRerank [21]2026.03 | 39.2 | 58 | 59.2 | 33 | 49.4 | 31.8 | 39 | 36.4 | 33.3 | 33.5 | 10.8 | 34.2 | 51.6 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ReasonIREnsemble=false, Strategy=Retrieve-then-rerank2025.10 | 37.3 | — | — | — | — | — | — | — | — | — | — | — | — | 41.7 | 59.8 | 53.2 | 32 | 43.6 | 28.8 | 38.7 | 36 | 34 | 33.2 | 34.8 | 29.4 | 7.9 | 32.6 | 47.7 | |
| T1-4BEvaluation Protocol=Retriever with Original Query2026.03 | 37.1 | 57.4 | 54.8 | 30.6 | 48.2 | 33.1 | 36.4 | 35.6 | 31.9 | 14.9 | 11.9 | 41.6 | 48.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ReasonIR-8BEvaluation Protocol=Retriever with QwenRerank [21]2026.03 | 36.9 | 58.2 | 53.2 | 32 | 43.6 | 28.8 | 37.6 | 36 | 34.8 | 33.2 | 7.9 | 32.6 | 45 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ReasonIR-8BEvaluation Protocol=Retriever with GPT-4 REASON-query2026.03 | 29.9 | 43.6 | 42.9 | 32.7 | 38.8 | 20.9 | 25.8 | 27.5 | 19.6 | 31.5 | 7.4 | 33.1 | 35.7 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| GPT-4 + Rank1-32BEnsemble=false, Strategy=Retrieve-then-rerank2025.10 | 29.4 | — | — | — | — | — | — | — | — | — | — | — | — | 32 | 49.7 | 35.8 | 22 | 37.5 | 22.5 | 21.7 | 35 | 25.7 | 18.8 | 32.5 | 25.8 | 10.8 | 22.9 | 43.7 | |
| RaDeR-7BEvaluation Protocol=Retriever with GPT-4 REASON-query2026.03 | 29.2 | 36.1 | 42.9 | 25.2 | 37.9 | 16.6 | 27.4 | 25 | 11.9 | 34.8 | 12 | 37.7 | 43.4 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Qwen1.5-7B + InteRank-3BEnsemble=false, Strategy=Retrieve-then-rerank2025.10 | 27.4 | — | — | — | — | — | — | — | — | — | — | — | — | 31.9 | 51.2 | 51.4 | 22.4 | 31.9 | 17.3 | 26.6 | 22.4 | 23.8 | 24.5 | 23.1 | 19.4 | 13.5 | 19.3 | 25.5 | |
| Seed1.5-EmbeddingEvaluation Protocol=Retriever with Original Query2026.03 | 27.2 | 34.8 | 46.9 | 23.4 | 31.6 | 19.1 | 25.4 | 21 | 4.9 | 43.2 | 12.2 | 33.3 | 30.5 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| RaDeR-7BEvaluation Protocol=Retriever with Original Query2026.03 | 25.5 | 34.6 | 38.9 | 22.1 | 33 | 14.8 | 22.5 | 23.7 | 5 | 37.3 | 10.2 | 28.4 | 35.1 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| ReasonIR-8BEvaluation Protocol=Retriever with Original Query2026.03 | 24.4 | 26.2 | 31.4 | 23.3 | 30 | 18 | 23.9 | 20.5 | 10.5 | 35 | 14.7 | 31.9 | 27.2 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | |
| Rank-R1-14BEnsemble=false, Strategy=Retrieve-then-rerank2025.10 | 20.5 | — | — | — | — | — | — | — | — | — | — | — | — | 26.6 | 31.2 | 38.5 | 21.2 | 26.4 | 22.6 | 18.9 | 27.5 | 14.7 | 9.2 | 20.2 | 10.3 | 9.7 | 11.9 | 9.2 |