# Mathematical Reasoning on Polaris (test) Source: [Mathematical Reasoning on Polaris (test)]() Collection index: [llms.txt]() Data updated: 2026-05-27T08:40:39.732Z Evaluation of model performance on the test split of the Polaris dataset focusing on reinforcement learning efficiency metrics, such as cumulative rollouts and efficiency ratios. 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 | Mathematical Reasoning | | Dataset (including any stated split) | Polaris (test) | | Metric | Cumulative Rollouts (M) | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 16 of 16 | | 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 - [Cumulative Rollouts (M)](): Higher is better; unit: SCALAR. - [Efficiency Ratio (GRPO/PC)](): Higher is better; unit: SCALAR. - [Efficiency Ratio (DAPO/PC)](): Higher is better; unit: SCALAR. ## Recorded results | Method | Cumulative Rollouts (M) | Result date | Report paper | | --- | --- | --- | --- | | DAPO | 4.92 | Not recorded | 6a167a79dbd7150506f7fed0 | | GRPO | 4.1 | Not recorded | 6a167a79dbd7150506f7fed0 | | DAPO | 2.7 | Not recorded | 6a167a79dbd7150506f7fed0 | | GRPO | 2.5 | Not recorded | 6a167a79dbd7150506f7fed0 | | Pilot-Commit | 2.15 | Not recorded | 6a167a79dbd7150506f7fed0 | | DAPO | 1.69 | Not recorded | 6a167a79dbd7150506f7fed0 | | Pilot-Commit | 1.66 | Not recorded | 6a167a79dbd7150506f7fed0 | | GRPO | 1.56 | Not recorded | 6a167a79dbd7150506f7fed0 | | DAPO | 1.26 | Not recorded | 6a167a79dbd7150506f7fed0 | | Pilot-Commit | 1.04 | Not recorded | 6a167a79dbd7150506f7fed0 | | GRPO | 0.73 | Not recorded | 6a167a79dbd7150506f7fed0 | | GRPO | 0.63 | Not recorded | 6a167a79dbd7150506f7fed0 | | GRPO | 0.52 | Not recorded | 6a167a79dbd7150506f7fed0 | | Pilot-Commit | 0.51 | Not recorded | 6a167a79dbd7150506f7fed0 | | Pilot-Commit | 0.51 | Not recorded | 6a167a79dbd7150506f7fed0 | | Pilot-Commit | 0.47 | Not recorded | 6a167a79dbd7150506f7fed0 | ### DAPO Cumulative Rollouts (M): 4.92. Result ID: 6a16ace4fbc373592c0fd970. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"14B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"57.0%\*"}. This record does not assert independent reproduction. ### GRPO Cumulative Rollouts (M): 4.1. Result ID: 6a16ace4fbc373592c0fd96f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"14B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"57.0%\*"}. This record does not assert independent reproduction. ### DAPO Cumulative Rollouts (M): 2.7. Result ID: 6a16ace4fbc373592c0fd96d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"8B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"47.3%"}. This record does not assert independent reproduction. ### GRPO Cumulative Rollouts (M): 2.5. Result ID: 6a16ace4fbc373592c0fd96a. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"4B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"47.3%\*"}. This record does not assert independent reproduction. ### Pilot-Commit Cumulative Rollouts (M): 2.15. Result ID: 6a16ace4fbc373592c0fd96e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"14B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"57.0%\*"}. This record does not assert independent reproduction. ### DAPO Cumulative Rollouts (M): 1.69. Result ID: 6a16ace4fbc373592c0fd973. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"4B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"45.0%"}. This record does not assert independent reproduction. ### Pilot-Commit Cumulative Rollouts (M): 1.66. Result ID: 6a16ace4fbc373592c0fd969. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"4B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"47.3%\*"}. This record does not assert independent reproduction. ### GRPO Cumulative Rollouts (M): 1.56. Result ID: 6a16ace4fbc373592c0fd96c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"8B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"47.3%"}. This record does not assert independent reproduction. ### DAPO Cumulative Rollouts (M): 1.26. Result ID: 6a16ace4fbc373592c0fd976. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"8B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"47.1%"}. This record does not assert independent reproduction. ### Pilot-Commit Cumulative Rollouts (M): 1.04. Result ID: 6a16ace4fbc373592c0fd96b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"8B","Budget Type":"ample","Rollout count per prompt (n)":"64","Target Accuracy":"47.3%"}. This record does not assert independent reproduction. ### GRPO Cumulative Rollouts (M): 0.73. Result ID: 6a16ace4fbc373592c0fd972. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"4B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"45.0%"}. This record does not assert independent reproduction. ### GRPO Cumulative Rollouts (M): 0.63. Result ID: 6a16ace4fbc373592c0fd975. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"8B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"47.1%"}. This record does not assert independent reproduction. ### GRPO Cumulative Rollouts (M): 0.52. Result ID: 6a16ace4fbc373592c0fd978. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Scale":"14B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"54.3%"}. This record does not assert independent reproduction. ### Pilot-Commit Cumulative Rollouts (M): 0.51. Result ID: 6a16ace4fbc373592c0fd971. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"4B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"45.0%"}. This record does not assert independent reproduction. ### Pilot-Commit Cumulative Rollouts (M): 0.51. Result ID: 6a16ace4fbc373592c0fd977. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"14B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"54.3%"}. This record does not assert independent reproduction. ### Pilot-Commit Cumulative Rollouts (M): 0.47. Result ID: 6a16ace4fbc373592c0fd974. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Model Scale":"8B","Budget Type":"limited","Rollout count per prompt (n)":"16","Target Accuracy":"47.1%"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.