# Reward Aggregation on Average HealthBench, WritingBench, PLawBench Source: [Reward Aggregation on Average HealthBench, WritingBench, PLawBench]() Collection index: [llms.txt]() Data updated: 2026-06-04T08:56:53.970Z Evaluation of model performance on the Reward Aggregation task, calculated as the average result across the HealthBench, WritingBench, and PLawBench components, using Average Score and Average Absolute Gain metrics. A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage. | Scope | Recorded value | | --- | --- | | Task | Reward Aggregation | | Dataset (including any stated split) | Average HealthBench, WritingBench, PLawBench | | Metric | Average Score | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 18 of 18 | | Coverage | All results currently returned for this benchmark | Evaluation settings may differ between results. The original papers and recorded configurations are needed to establish comparability. A shared metric name alone does not establish an identical protocol. ## Metrics - [Average Score](): Higher is better; unit: SCALAR. - [Average Absolute Gain](): Higher is better; unit: SCALAR. ## Recorded results | Method | Average Score | Result date | Report paper | | --- | --- | --- | --- | | RuscaRL + GEAR | 75.7 | Not recorded | 6a210667c10484bac23eff5b | | RuscaRL + GEAR | 74.7 | Not recorded | 6a210667c10484bac23eff5b | | GEAR | 73.9 | Not recorded | 6a210667c10484bac23eff5b | | GEAR | 72.6 | Not recorded | 6a210667c10484bac23eff5b | | RuscaRL | 69.3 | Not recorded | 6a210667c10484bac23eff5b | | RuscaRL | 68.6 | Not recorded | 6a210667c10484bac23eff5b | | Flat aggregation | 66.6 | Not recorded | 6a210667c10484bac23eff5b | | Flat aggregation | 66.4 | Not recorded | 6a210667c10484bac23eff5b | | RaR-Explicit + GEAR | 66.1 | Not recorded | 6a210667c10484bac23eff5b | | RaR-Explicit + GEAR | 65.1 | Not recorded | 6a210667c10484bac23eff5b | | Hard gating | 64.1 | Not recorded | 6a210667c10484bac23eff5b | | Hard gating | 64 | Not recorded | 6a210667c10484bac23eff5b | | RaR-Explicit | 60 | Not recorded | 6a210667c10484bac23eff5b | | RaR-Explicit | 59.6 | Not recorded | 6a210667c10484bac23eff5b | | MeRF + GEAR | 58.9 | Not recorded | 6a210667c10484bac23eff5b | | MeRF + GEAR | 58.2 | Not recorded | 6a210667c10484bac23eff5b | | MeRF | 53.4 | Not recorded | 6a210667c10484bac23eff5b | | MeRF | 52.6 | Not recorded | 6a210667c10484bac23eff5b | ### RuscaRL + GEAR Average Score: 75.7. Result ID: 6a213dccfbc373592c1cf1ff. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### RuscaRL + GEAR Average Score: 74.7. Result ID: 6a213dccfbc373592c1cf1f6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### GEAR Average Score: 73.9. Result ID: 6a213dccfbc373592c1cf1f9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Aggregation-only"}. This record does not assert independent reproduction. ### GEAR Average Score: 72.6. Result ID: 6a213dccfbc373592c1cf1f0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Aggregation-only"}. This record does not assert independent reproduction. ### RuscaRL Average Score: 69.3. Result ID: 6a213dccfbc373592c1cf1fe. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### RuscaRL Average Score: 68.6. Result ID: 6a213dccfbc373592c1cf1f5. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### Flat aggregation Average Score: 66.6. Result ID: 6a213dccfbc373592c1cf1f7. [Reporting paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Aggregation-only"}. This record does not assert independent reproduction. ### Flat aggregation Average Score: 66.4. Result ID: 6a213dccfbc373592c1cf1ee. [Reporting paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Aggregation-only"}. This record does not assert independent reproduction. ### RaR-Explicit + GEAR Average Score: 66.1. Result ID: 6a213dccfbc373592c1cf1fb. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### RaR-Explicit + GEAR Average Score: 65.1. Result ID: 6a213dccfbc373592c1cf1f2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### Hard gating Average Score: 64.1. Result ID: 6a213dccfbc373592c1cf1f8. [Reporting paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Aggregation-only"}. This record does not assert independent reproduction. ### Hard gating Average Score: 64. Result ID: 6a213dccfbc373592c1cf1ef. [Reporting paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Aggregation-only"}. This record does not assert independent reproduction. ### RaR-Explicit Average Score: 60. Result ID: 6a213dccfbc373592c1cf1fa. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### RaR-Explicit Average Score: 59.6. Result ID: 6a213dccfbc373592c1cf1f1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### MeRF + GEAR Average Score: 58.9. Result ID: 6a213dccfbc373592c1cf1fd. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### MeRF + GEAR Average Score: 58.2. Result ID: 6a213dccfbc373592c1cf1f4. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### MeRF Average Score: 53.4. Result ID: 6a213dccfbc373592c1cf1fc. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Llama-3.1-8B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. ### MeRF Average Score: 52.6. Result ID: 6a213dccfbc373592c1cf1f3. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Policy Model":"Qwen2.5-7B-Instruct","Pipeline Category":"Plug-in replacement in rubric-guided RL pipelines"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.