Real-world Reasoning on RLPR
44.75AccuracyConductor
Evaluation Results
| Method | Links | |
|---|---|---|
| Conductorparameters=7B2025.12 | 44.75 | |
| GPT 52025.12 | 42.2 | |
| Gemini 2.5 Pro2025.12 | 40.55 | |
| Qwen3-32B (thinking)mode=thinking2025.12 | 37.25 | |
| Claude Sonnet 42025.12 | 36.7 | |
| R1-Distill-Qwen-32B2025.12 | 33.5 | |
| Qwen3-32B2025.12 | 31 | |
| gemma-3-27b-itinstruct=true2025.12 | 16.67 |