# Mathematics on MATH 500 (avg@32 accuracy) Source: [Mathematics on MATH 500 (avg@32 accuracy)]() Collection index: [llms.txt]() Data updated: 2026-05-28T08:54:10.722Z Evaluation of model performance on the MATH 500 dataset for mathematics problems, specifically measured using the average accuracy at 32 samples (avg@32 accuracy). 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 | Mathematics | | Dataset (including any stated split) | MATH 500 | | Metric | Accuracy (avg@32) | | Unit | PERCENT | | 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 - [Accuracy (avg@32)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy (avg@32) | Result date | Report paper | | --- | --- | --- | --- | | IBTPO | 83.3 | Not recorded | 6a17cc00dbd7150506f81233 | | TreeRL | 82.5 | Not recorded | 6a17cc00dbd7150506f81233 | | TreePO | 82.2 | Not recorded | 6a17cc00dbd7150506f81233 | | IBRO | 82 | Not recorded | 6a17cc00dbd7150506f81233 | | GRPO w/ Entropy Reg† | 81.8 | Not recorded | 6a17cc00dbd7150506f81233 | | Vanilla GRPO | 81.5 | Not recorded | 6a17cc00dbd7150506f81233 | | GRPO w/ Clip-higher | 81.2 | Not recorded | 6a17cc00dbd7150506f81233 | | IBTPO | 70.1 | Not recorded | 6a17cc00dbd7150506f81233 | | TreeRL | 67.2 | Not recorded | 6a17cc00dbd7150506f81233 | | Vanilla GRPO | 66.8 | Not recorded | 6a17cc00dbd7150506f81233 | | TreePO | 66.6 | Not recorded | 6a17cc00dbd7150506f81233 | | IBRO | 66.4 | Not recorded | 6a17cc00dbd7150506f81233 | | GRPO w/ Clip-higher | 65.8 | Not recorded | 6a17cc00dbd7150506f81233 | | Initial Model | 65.7 | Not recorded | 6a17cc00dbd7150506f81233 | | GRPO w/ Entropy Reg† | 64.8 | Not recorded | 6a17cc00dbd7150506f81233 | | Initial Model | 51.2 | Not recorded | 6a17cc00dbd7150506f81233 | ### IBTPO Accuracy (avg@32): 83.3. Result ID: 6a180298fbc373592c11dbe1. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### TreeRL Accuracy (avg@32): 82.5. Result ID: 6a180298fbc373592c11dbdf. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### TreePO Accuracy (avg@32): 82.2. Result ID: 6a180298fbc373592c11dbe0. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### IBRO Accuracy (avg@32): 82. Result ID: 6a180298fbc373592c11dbde. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GRPO w/ Entropy Reg† Accuracy (avg@32): 81.8. Result ID: 6a180298fbc373592c11dbdd. [Reporting paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### Vanilla GRPO Accuracy (avg@32): 81.5. Result ID: 6a180298fbc373592c11dbdb. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GRPO w/ Clip-higher Accuracy (avg@32): 81.2. Result ID: 6a180298fbc373592c11dbdc. [Reporting paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### IBTPO Accuracy (avg@32): 70.1. Result ID: 6a180298fbc373592c11dbd9. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### TreeRL Accuracy (avg@32): 67.2. Result ID: 6a180298fbc373592c11dbd7. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### Vanilla GRPO Accuracy (avg@32): 66.8. Result ID: 6a180298fbc373592c11dbd3. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### TreePO Accuracy (avg@32): 66.6. Result ID: 6a180298fbc373592c11dbd8. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### IBRO Accuracy (avg@32): 66.4. Result ID: 6a180298fbc373592c11dbd6. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### GRPO w/ Clip-higher Accuracy (avg@32): 65.8. Result ID: 6a180298fbc373592c11dbd4. [Reporting paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### Initial Model Accuracy (avg@32): 65.7. Result ID: 6a180298fbc373592c11dbda. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-8B-Base"}. This record does not assert independent reproduction. ### GRPO w/ Entropy Reg† Accuracy (avg@32): 64.8. Result ID: 6a180298fbc373592c11dbd5. [Reporting paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. ### Initial Model Accuracy (avg@32): 51.2. Result ID: 6a180298fbc373592c11dbd2. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Base Model":"Qwen3-1.7B-Base"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.