Long-context Summary on Summary Task (test)
63.2Multi-turn Coverage ScoreRLSTA
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| RLSTAModel=Qwen3-4B Instruct-25072026.03 | 63.2 | 75.8 | 83.5 | |
| BaseModel=Qwen3-4B Instruct-25072026.03 | 55.5 | 75 | 74 | |
| BaseModel=Qwen2.5-7B Instruct2026.03 | 44.6 | 59.4 | 75.1 | |
| RLSTAModel=Qwen2.5-7B Instruct2026.03 | 44.1 | 59 | 74.7 | |
| RLSTAModel=Llama-3.2-3B Instruct2026.03 | 43.4 | 55.5 | 78.2 | |
| RLSTAModel=Qwen2.5-3B Instruct2026.03 | 43.3 | 52 | 83.2 | |
| BaseModel=Llama-3.2-3B Instruct2026.03 | 38.4 | 54.9 | 69.8 | |
| BaseModel=Qwen2.5-3B Instruct2026.03 | 35.9 | 52.4 | 68.5 |