# Language Similarity Grounding on SST-2 Source: [Language Similarity Grounding on SST-2]() Collection index: [llms.txt]() Data updated: 2026-05-18T08:40:54.981Z Evaluation of model performance on grounding language similarity tasks using the SST-2 dataset, typically measured by correlation 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 | Language similarity grounding | | Dataset (including any stated split) | SST-2 | | Metric | Accuracy Correlation | | Unit | SCALAR | | Direction | Higher is better | | Results loaded | 31 of 31 | | 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 Correlation](): Higher is better; unit: SCALAR. - [JSD Correlation](): Higher is better; unit: SCALAR. - [Label Random Performance](): Higher is better; unit: SCALAR. - [Shortcut Score](): Higher is better; unit: SCALAR. ## Recorded results | Method | Accuracy Correlation | Result date | Report paper | | --- | --- | --- | --- | | AD-CKA | 0.43 | Not recorded | 6a0a9d29dbc836b87d82b49b | | UnifDiff | 0.35 | Not recorded | 6a0a9d29dbc836b87d82b49b | | 2nd-Cos | 0.3 | Not recorded | 6a0a9d29dbc836b87d82b49b | | AD-DistCorr | 0.25 | Not recorded | 6a0a9d29dbc836b87d82b49b | | AD-RBF-CKA | 0.24 | Not recorded | 6a0a9d29dbc836b87d82b49b | | RSMDiff | 0.2 | Not recorded | 6a0a9d29dbc836b87d82b49b | | Jaccard | 0.17 | Not recorded | 6a0a9d29dbc836b87d82b49b | | RankSim | 0.14 | Not recorded | 6a0a9d29dbc836b87d82b49b | | ConcDiff | 0.12 | Not recorded | 6a0a9d29dbc836b87d82b49b | | ProcDist | 0.1 | Not recorded | 6a0a9d29dbc836b87d82b49b | | RTD | 0.05 | Not recorded | 6a0a9d29dbc836b87d82b49b | | AlignCos | 0.02 | Not recorded | 6a0a9d29dbc836b87d82b49b | | CKA | -0.06 | Not recorded | 6a0a9d29dbc836b87d82b49b | | IMD | -0.06 | Not recorded | 6a0a9d29dbc836b87d82b49b | | Multi-Scale CKA | -0.06 | Not recorded | 6a0a9d29dbc836b87d82b49b | | SVCCA | -0.08 | Not recorded | 6a0a9d29dbc836b87d82b49b | | PermProc | -0.09 | Not recorded | 6a0a9d29dbc836b87d82b49b | | DistCorr | -0.1 | Not recorded | 6a0a9d29dbc836b87d82b49b | | Multi-Scale DistCorr | -0.12 | Not recorded | 6a0a9d29dbc836b87d82b49b | | MagDiff | -0.13 | Not recorded | 6a0a9d29dbc836b87d82b49b | | AngShape | -0.14 | Not recorded | 6a0a9d29dbc836b87d82b49b | | OrthProc | -0.14 | Not recorded | 6a0a9d29dbc836b87d82b49b | | RBF-CKA | -0.16 | Not recorded | 6a0a9d29dbc836b87d82b49b | | Multi-Scale RBF-CKA | -0.17 | Not recorded | 6a0a9d29dbc836b87d82b49b | | RSA | -0.23 | Not recorded | 6a0a9d29dbc836b87d82b49b | | PWCCA | -0.33 | Not recorded | 6a0a9d29dbc836b87d82b49b | | HardCorr | -0.33 | Not recorded | 6a0a9d29dbc836b87d82b49b | | SoftCorr | -0.33 | Not recorded | 6a0a9d29dbc836b87d82b49b | | GULP | -0.36 | Not recorded | 6a0a9d29dbc836b87d82b49b | | LinReg | -0.38 | Not recorded | 6a0a9d29dbc836b87d82b49b | | EOS | -0.38 | Not recorded | 6a0a9d29dbc836b87d82b49b | ### AD-CKA Accuracy Correlation: 0.43. Result ID: 6a0ad08b47f34ce924f1053e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### UnifDiff Accuracy Correlation: 0.35. Result ID: 6a0ad08b47f34ce924f1053c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### 2nd-Cos Accuracy Correlation: 0.3. Result ID: 6a0ad08b47f34ce924f10535. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### AD-DistCorr Accuracy Correlation: 0.25. Result ID: 6a0ad08b47f34ce924f10542. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### AD-RBF-CKA Accuracy Correlation: 0.24. Result ID: 6a0ad08b47f34ce924f10540. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### RSMDiff Accuracy Correlation: 0.2. Result ID: 6a0ad08b47f34ce924f10534. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### Jaccard Accuracy Correlation: 0.17. Result ID: 6a0ad08b47f34ce924f10536. [Reporting paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### RankSim Accuracy Correlation: 0.14. Result ID: 6a0ad08b47f34ce924f10537. [Reporting paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### ConcDiff Accuracy Correlation: 0.12. Result ID: 6a0ad08b47f34ce924f1053a. [Reporting paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### ProcDist Accuracy Correlation: 0.1. Result ID: 6a0ad08b47f34ce924f1052c. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### RTD Accuracy Correlation: 0.05. Result ID: 6a0ad08b47f34ce924f10539. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### AlignCos Accuracy Correlation: 0.02. Result ID: 6a0ad08b47f34ce924f10526. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### CKA Accuracy Correlation: -0.06. Result ID: 6a0ad08b47f34ce924f1052e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### IMD Accuracy Correlation: -0.06. Result ID: 6a0ad08b47f34ce924f10538. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### Multi-Scale CKA Accuracy Correlation: -0.06. Result ID: 6a0ad08b47f34ce924f1053d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### SVCCA Accuracy Correlation: -0.08. Result ID: 6a0ad08b47f34ce924f10525. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### PermProc Accuracy Correlation: -0.09. Result ID: 6a0ad08b47f34ce924f1052b. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### DistCorr Accuracy Correlation: -0.1. Result ID: 6a0ad08b47f34ce924f10530. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### Multi-Scale DistCorr Accuracy Correlation: -0.12. Result ID: 6a0ad08b47f34ce924f10541. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### MagDiff Accuracy Correlation: -0.13. Result ID: 6a0ad08b47f34ce924f1053b. [Reporting paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### AngShape Accuracy Correlation: -0.14. Result ID: 6a0ad08b47f34ce924f10527. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### OrthProc Accuracy Correlation: -0.14. Result ID: 6a0ad08b47f34ce924f1052a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### RBF-CKA Accuracy Correlation: -0.16. Result ID: 6a0ad08b47f34ce924f1052f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### Multi-Scale RBF-CKA Accuracy Correlation: -0.17. Result ID: 6a0ad08b47f34ce924f1053f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### RSA Accuracy Correlation: -0.23. Result ID: 6a0ad08b47f34ce924f10533. [Reporting paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### PWCCA Accuracy Correlation: -0.33. Result ID: 6a0ad08b47f34ce924f10524. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### HardCorr Accuracy Correlation: -0.33. Result ID: 6a0ad08b47f34ce924f10528. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### SoftCorr Accuracy Correlation: -0.33. Result ID: 6a0ad08b47f34ce924f1052d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### GULP Accuracy Correlation: -0.36. Result ID: 6a0ad08b47f34ce924f10532. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### LinReg Accuracy Correlation: -0.38. Result ID: 6a0ad08b47f34ce924f10529. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. ### EOS Accuracy Correlation: -0.38. Result ID: 6a0ad08b47f34ce924f10531. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"Domain":"Language","Architecture":"BERT"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.