# Privacy-related decision quality evaluation on MemPrivacy-Bench (test) Source: [Privacy-related decision quality evaluation on MemPrivacy-Bench (test)]() Collection index: [llms.txt]() Data updated: 2026-05-12T12:24:57.812Z Evaluation of model performance on the test split of the MemPrivacy-Bench dataset, focusing on privacy-related decision quality. 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 | Privacy-related decision quality evaluation | | Dataset (including any stated split) | MemPrivacy-Bench (test) | | Metric | GLM-5.1 Score | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 11 of 11 | | 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 - [GLM-5.1 Score](): Higher is better; unit: PERCENT. - [DeepSeek-R1-0528 Score](): Higher is better; unit: PERCENT. - [Qwen3-235B-A22B Score](): Higher is better; unit: PERCENT. ## Recorded results | Method | GLM-5.1 Score | Result date | Report paper | | --- | --- | --- | --- | | MemPrivacy-4B-RL | 83.85 | Not recorded | 6a02b46cdbc836b87d824b78 | | MemPrivacy-4B-SFT | 82.98 | Not recorded | 6a02b46cdbc836b87d824b78 | | MemPrivacy-1.7B-RL | 82.02 | Not recorded | 6a02b46cdbc836b87d824b78 | | MemPrivacy-0.6B-RL | 81.74 | Not recorded | 6a02b46cdbc836b87d824b78 | | MemPrivacy-1.7B-SFT | 81.02 | Not recorded | 6a02b46cdbc836b87d824b78 | | MemPrivacy-0.6B-SFT | 79.82 | Not recorded | 6a02b46cdbc836b87d824b78 | | Gemini-3.1-Pro | 76.15 | Not recorded | 6a02b46cdbc836b87d824b78 | | DeepSeek-V3.2-Think | 70 | Not recorded | 6a02b46cdbc836b87d824b78 | | GPT-5.2 | 66.54 | Not recorded | 6a02b46cdbc836b87d824b78 | | GLM-5 | 65.38 | Not recorded | 6a02b46cdbc836b87d824b78 | | Kimi-K2 | 63.01 | Not recorded | 6a02b46cdbc836b87d824b78 | ### MemPrivacy-4B-RL GLM-5.1 Score: 83.85. Result ID: 6a031bfc47f34ce924e74153. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Privacy Model","Model Scale":"4B","Training Protocol":"RL"}. This record does not assert independent reproduction. ### MemPrivacy-4B-SFT GLM-5.1 Score: 82.98. Result ID: 6a031bfc47f34ce924e74150. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Privacy Model","Model Scale":"4B","Training Protocol":"SFT"}. This record does not assert independent reproduction. ### MemPrivacy-1.7B-RL GLM-5.1 Score: 82.02. Result ID: 6a031bfc47f34ce924e74152. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Privacy Model","Model Scale":"1.7B","Training Protocol":"RL"}. This record does not assert independent reproduction. ### MemPrivacy-0.6B-RL GLM-5.1 Score: 81.74. Result ID: 6a031bfc47f34ce924e74151. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Privacy Model","Model Scale":"0.6B","Training Protocol":"RL"}. This record does not assert independent reproduction. ### MemPrivacy-1.7B-SFT GLM-5.1 Score: 81.02. Result ID: 6a031bfc47f34ce924e7414f. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Privacy Model","Model Scale":"1.7B","Training Protocol":"SFT"}. This record does not assert independent reproduction. ### MemPrivacy-0.6B-SFT GLM-5.1 Score: 79.82. Result ID: 6a031bfc47f34ce924e7414e. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Model Type":"Privacy Model","Model Scale":"0.6B","Training Protocol":"SFT"}. This record does not assert independent reproduction. ### Gemini-3.1-Pro GLM-5.1 Score: 76.15. Result ID: 6a031bfc47f34ce924e7414d. [Reporting paper]() Recorded evaluation configuration: {"Model Type":"General Model"}. This record does not assert independent reproduction. ### DeepSeek-V3.2-Think GLM-5.1 Score: 70. Result ID: 6a031bfc47f34ce924e7414c. [Reporting paper]() Recorded evaluation configuration: {"Model Type":"General Model"}. This record does not assert independent reproduction. ### GPT-5.2 GLM-5.1 Score: 66.54. Result ID: 6a031bfc47f34ce924e74149. [Reporting paper]() Recorded evaluation configuration: {"Model Type":"General Model"}. This record does not assert independent reproduction. ### GLM-5 GLM-5.1 Score: 65.38. Result ID: 6a031bfc47f34ce924e7414b. [Reporting paper]() Recorded evaluation configuration: {"Model Type":"General Model"}. This record does not assert independent reproduction. ### Kimi-K2 GLM-5.1 Score: 63.01. Result ID: 6a031bfc47f34ce924e7414a. [Reporting paper]() Recorded evaluation configuration: {"Model Type":"General Model"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.