Emotional Support Conversation on SAGE (test)
85.07SentiencePEER_RUR
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| PEER_RURReward=Redundancy-based UnifiReward2025.08 | 85.07 | 43.5 | 4.5 | |
| PEER_URReward=UnifiReward2025.08 | 73.5 | 42.5 | 13.5 | |
| Qwen3-VLStructured three-stage prompt=false2025.08 | 72.47 | 41.5 | 16.5 | |
| Qwen3-VLStructured three-stage prompt=true2025.08 | 68.39 | 28.5 | 19 | |
| Qwen3Structured three-stage prompt=true2025.08 | 52.31 | 14.5 | 28 | |
| InternVL3.5Structured three-stage prompt=true2025.08 | 48.74 | 13.5 | 26 | |
| InternLM3Structured three-stage prompt=true2025.08 | 48.41 | 15.5 | 35.5 | |
| Qwen3Structured three-stage prompt=false2025.08 | 44.84 | 9.5 | 37 | |
| soulChat22025.08 | 44.38 | 4 | 22.5 | |
| InternLM3Structured three-stage prompt=false2025.08 | 43.02 | 11.5 | 41.5 | |
| InternVL3.5Structured three-stage prompt=false2025.08 | 33.96 | 9.5 | 47.5 | |
| EmoLLMV32025.08 | 31.48 | 3.5 | 46.5 | |
| PEER_SFTFine-tuning=Supervised2025.08 | 25.96 | 4.5 | 47.5 | |
| CPsyCounX2025.08 | 20.35 | 1.5 | 53 |