Reward Aggregation on WritingBench
85ScoreRuscaRL + GEAR
Evaluation results
| Method | Links | ||
|---|---|---|---|
| RuscaRL + GEARPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 85 | 8.3 | |
| GEARPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Aggregation-only2026.06 | 82.4 | 6.2 | |
| RuscaRL + GEARPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 81.5 | 6.1 | |
| GEARPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Aggregation-only2026.06 | 79.6 | 5.5 | |
| RuscaRLPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 76.7 | — | |
| Flat aggregationPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Aggregation-only2026.06 | 76.2 | — | |
| RuscaRLPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 75.4 | — | |
| Flat aggregationPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Aggregation-only2026.06 | 74.1 | — | |
| RaR-Explicit + GEARPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 73.6 | 5.7 | |
| Hard gatingPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Aggregation-only2026.06 | 72.7 | — | |
| Hard gatingPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Aggregation-only2026.06 | 72.5 | — | |
| RaR-Explicit + GEARPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 71.6 | 4.6 | |
| MeRF + GEARPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 69.8 | 7.5 | |
| MeRF + GEARPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 68.7 | 6.8 | |
| RaR-ExplicitPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 67.9 | — | |
| RaR-ExplicitPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 67 | — | |
| MeRFPolicy Model=Llama-3.1-8B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 62.3 | — | |
| MeRFPolicy Model=Qwen2.5-7B-Instruct, Pipeline Category=Plug-in replacement in rubric-guided RL pipelines2026.06 | 61.9 | — |