Four-class classification on CB (evaluation set)
59.08PrecisionIssue discovery agent
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| Issue discovery agentTools=Tools 1 & 2, Base Model=REDA 7B2026.01 | 59.08 | 70.49 | 92.85 | 62.91 | |
| Issue discovery agentTools=Tools 1 & 3, Base Model=REDA 7B2026.01 | 55.2 | 62.03 | 92.58 | 57.8 | |
| Issue discovery agentTools=Tool 1 (governance models), Base Model=REDA 7B2026.01 | 51.95 | 59.98 | 92.33 | 54.64 | |
| Issue discovery agentTools=Tool 2 (embedding models), Base Model=REDA 7B2026.01 | 46.52 | 47.95 | 89.06 | 37.38 | |
| Issue discovery agentTools=Tool 3 (long-term memory), Base Model=REDA 7B2026.01 | 41.52 | 47.95 | 89.04 | 37.26 | |
| Issue discovery agentTools=No tools, Base Model=REDA 7B2026.01 | 39.85 | 47.7 | 88.96 | 36.83 | |
| Supervised fine-tuningBase Model=REDA 7B, Tuning=LoRA2026.01 | 32.99 | 45.15 | 89.83 | 36.88 | |
| Supervised fine-tuningBase Model=Qwen2.5-VL 7B, Tuning=LoRA2026.01 | 32.95 | 45.14 | 89.8 | 36.26 |