Agentic Retrieval on BRIGHT-PRO Round 1 (test)
56.66alpha-nDCGBGE-Reasoner-8B
Evaluation results
| Method | Links | |||
|---|---|---|---|---|
| BGE-Reasoner-8BEvaluation Agent=GPT-5-mini, Fine-tuning=none, Hardware=NVIDIA H200 GPUs2026.05 | 56.66 | 4.14 | 3.95 | |
| DIVER-4BEvaluation Agent=GPT-5-mini2026.05 | 46.75 | 3.95 | 3.78 | |
| GTE-7BEvaluation Agent=GPT-5-mini2026.05 | 45.88 | 3.93 | 3.84 | |
| BM25Evaluation Agent=GPT-5-mini2026.05 | 44.46 | 3.82 | 3.67 | |
| DIVER-4B-1020Evaluation Agent=GPT-5-mini2026.05 | 44.28 | 3.9 | 3.77 | |
| RTriever-4BEvaluation Agent=GPT-5-mini2026.05 | 44.22 | 3.98 | 3.85 | |
| Qwen3-8BEvaluation Agent=GPT-5-mini2026.05 | 44.11 | 3.81 | 3.74 | |
| GritLM (7B)Evaluation Agent=GPT-5-mini2026.05 | 40.28 | 3.89 | 3.74 | |
| ReasonIR-8BEvaluation Agent=GPT-5-mini2026.05 | 37.14 | 3.81 | 3.65 | |
| INSTRUCTOR-XLEvaluation Agent=GPT-5-mini2026.05 | 35.24 | 3.54 | 3.5 |