# Semantic Textual Similarity on STS-B (Leakage, Confidence, Pearson evaluation) Source: [Semantic Textual Similarity on STS-B (Leakage, Confidence, Pearson evaluation)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:48:33.878Z Evaluation of Semantic Textual Similarity performance on the STS-B dataset, focusing on metrics related to data leakage, model confidence, and Pearson correlation. 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 | Semantic Textual Similarity | | Dataset (including any stated split) | STSB | | Metric | Leakage | | Unit | SCALAR | | Direction | Lower 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 - [Leakage](): Lower is better; unit: SCALAR. - [Confidence](): Higher is better; unit: SCALAR. - [Pearson Correlation (Downstream)](): Higher is better; unit: SCALAR. ## Recorded results | Method | Leakage | Result date | Report paper | | --- | --- | --- | --- | | SPARSE | 2.68 | Not recorded | 698abbaec97a0b3e3b5de035 | | PurMech | 19.03 | Not recorded | 698abbaec97a0b3e3b5de035 | | LapMech | 20.75 | Not recorded | 698abbaec97a0b3e3b5de035 | | SPARSE | 32.39 | Not recorded | 698abbaec97a0b3e3b5de035 | | PurMech | 49.95 | Not recorded | 698abbaec97a0b3e3b5de035 | | LapMech | 53.79 | Not recorded | 698abbaec97a0b3e3b5de035 | | SPARSE | 64.79 | Not recorded | 698abbaec97a0b3e3b5de035 | | PurMech | 69.15 | Not recorded | 698abbaec97a0b3e3b5de035 | | LapMech | 71.82 | Not recorded | 698abbaec97a0b3e3b5de035 | | SPARSE | 73.1 | Not recorded | 698abbaec97a0b3e3b5de035 | | PurMech | 74.98 | Not recorded | 698abbaec97a0b3e3b5de035 | | LapMech | 76.15 | Not recorded | 698abbaec97a0b3e3b5de035 | | SPARSE | 76.42 | Not recorded | 698abbaec97a0b3e3b5de035 | | PurMech | 77.4 | Not recorded | 698abbaec97a0b3e3b5de035 | | LapMech | 78.94 | Not recorded | 698abbaec97a0b3e3b5de035 | | Non-protected | 86.75 | Not recorded | 698abbaec97a0b3e3b5de035 | ### SPARSE Leakage: 2.68. Result ID: 6996dcfd5698665ec4297149. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"5"}. This record does not assert independent reproduction. ### PurMech Leakage: 19.03. Result ID: 6996dcfd5698665ec4297148. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"5"}. This record does not assert independent reproduction. ### LapMech Leakage: 20.75. Result ID: 6996dcfd5698665ec4297147. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"5"}. This record does not assert independent reproduction. ### SPARSE Leakage: 32.39. Result ID: 6996dcfd5698665ec429714c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"10"}. This record does not assert independent reproduction. ### PurMech Leakage: 49.95. Result ID: 6996dcfd5698665ec429714b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"10"}. This record does not assert independent reproduction. ### LapMech Leakage: 53.79. Result ID: 6996dcfd5698665ec429714a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"10"}. This record does not assert independent reproduction. ### SPARSE Leakage: 64.79. Result ID: 6996dcfd5698665ec429714f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"20"}. This record does not assert independent reproduction. ### PurMech Leakage: 69.15. Result ID: 6996dcfd5698665ec429714e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"20"}. This record does not assert independent reproduction. ### LapMech Leakage: 71.82. Result ID: 6996dcfd5698665ec429714d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"20"}. This record does not assert independent reproduction. ### SPARSE Leakage: 73.1. Result ID: 6996dcfd5698665ec4297152. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"30"}. This record does not assert independent reproduction. ### PurMech Leakage: 74.98. Result ID: 6996dcfd5698665ec4297151. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"30"}. This record does not assert independent reproduction. ### LapMech Leakage: 76.15. Result ID: 6996dcfd5698665ec4297150. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"30"}. This record does not assert independent reproduction. ### SPARSE Leakage: 76.42. Result ID: 6996dcfd5698665ec4297155. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"40"}. This record does not assert independent reproduction. ### PurMech Leakage: 77.4. Result ID: 6996dcfd5698665ec4297154. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"40"}. This record does not assert independent reproduction. ### LapMech Leakage: 78.94. Result ID: 6996dcfd5698665ec4297153. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"40"}. This record does not assert independent reproduction. ### Non-protected Leakage: 86.75. Result ID: 6996dcfd5698665ec4297156. [Reporting paper]() Recorded evaluation configuration: {"epsilon (ϵ)":"infinity"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.