Summarization on BookSum
0.277RewardRL (Proposed)
Evaluation results
| Method | Links | ||||
|---|---|---|---|---|---|
| RL (Proposed)Model=Granite-3.3-2B2026.03 | 0.277 | 88.44 | 30.66 | 22.06 | |
| StaticModel=Granite-3.3-2B2026.03 | 0.212 | — | — | — | |
| GreedyModel=Granite-3.3-2B2026.03 | 0.147 | — | — | — | |
| RL (Proposed)Model=Qwen-2.5-0.5B2026.03 | 0.147 | 950 | 77.11 | 57.14 | |
| StaticModel=Qwen-2.5-0.5B2026.03 | 0.083 | — | — | — | |
| GreedyModel=Qwen-2.5-0.5B2026.03 | 0.014 | — | — | — |