Agentic Task on LocalSearch
40ScoreDR-Rubric-8B (Gemini)
Evaluation Results
| Method | Links | |
|---|---|---|
| DR-Rubric-8B (Gemini)Training=SFT+RL, 1K2026.05 | 40 | |
| Qwen2.5-7B2026.05 | 37.1 | |
| Qwen3-8B2026.05 | 37.1 | |
| DR-Rubric-8B (GPT-5)Training=SFT+RL, 1K2026.05 | 37.1 | |
| WebExplorer-8BTraining=SFT+RL, 25K2026.05 | 36.8 | |
| DR-Tulu-RL-8BTraining=SFT+RL, 25K2026.05 | 36.5 | |
| DR-Tulu-SFT-8BTraining=SFT, 16K2026.05 | 36.4 | |
| DR-Rubric-8B (BS-3)Training=SFT+RL, 3K2026.05 | 36.4 | |
| Qwen3-8B-SFTTraining=SFT, 1K2026.05 | 36 | |
| Search-R1-7BTraining=RL, 90K2026.05 | 34.2 |