# Machine Unlearning on ToxiGen (train) Source: [Machine Unlearning on ToxiGen (train)]() Collection index: [llms.txt]() Data updated: 2026-03-30T07:59:48.679Z Evaluation of Machine Unlearning performance on the training split of the ToxiGen dataset, focusing on accuracy metrics for the forget set ($D\_f$) and adjacent/remote retain sets ($D\_r^{adj}, D\_r^{rem}$). 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 | Machine Unlearning | | Dataset (including any stated split) | ToxiGen (train) | | Metric | Accuracy ($D\_f$) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 13 of 13 | | 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 ($D\_f$)](): Higher is better; unit: PERCENT. - [Accuracy ($D\_r^{adj}$)](): Higher is better; unit: PERCENT. - [Accuracy ($D\_r^{rem}$)](): Higher is better; unit: PERCENT. ## Recorded results | Method | Accuracy ($D\_f$) | Result date | Report paper | | --- | --- | --- | --- | | Original | 85.06 | Not recorded | 69ca03a80636e5bf4afc949f | | SSD | 67.78 | Not recorded | 69ca03a80636e5bf4afc949f | | SCRUB | 56.15 | Not recorded | 69ca03a80636e5bf4afc949f | | Munba | 51.09 | Not recorded | 69ca03a80636e5bf4afc949f | | FT | 50.04 | Not recorded | 69ca03a80636e5bf4afc949f | | GA | 46.26 | Not recorded | 69ca03a80636e5bf4afc949f | | ℓ1-sparse | 45.64 | Not recorded | 69ca03a80636e5bf4afc949f | | DELETE | 42.86 | Not recorded | 69ca03a80636e5bf4afc949f | | GDR | 20.54 | Not recorded | 69ca03a80636e5bf4afc949f | | SalUn | 13.66 | Not recorded | 69ca03a80636e5bf4afc949f | | Our method | 11.95 | Not recorded | 69ca03a80636e5bf4afc949f | | Two-Stage Machine Unlearning | 11.95 | Not recorded | 69ca03a80636e5bf4afc949f | | Retrain | 8.58 | Not recorded | 69ca03a80636e5bf4afc949f | ### Original Accuracy ($D\_f$): 85.06. Result ID: 69ca2d5847f34ce924b2a007. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SSD Accuracy ($D\_f$): 67.78. Result ID: 69ca2d5847f34ce924b2a00c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SCRUB Accuracy ($D\_f$): 56.15. Result ID: 69ca2d5847f34ce924b2a00d. [Reporting paper]() [Method origin paper]() [Implementation]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Munba Accuracy ($D\_f$): 51.09. Result ID: 69ca2d5847f34ce924b2a00b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### FT Accuracy ($D\_f$): 50.04. Result ID: 69ca2d5847f34ce924b2a008. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GA Accuracy ($D\_f$): 46.26. Result ID: 69ca2d5847f34ce924b2a009. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### ℓ1-sparse Accuracy ($D\_f$): 45.64. Result ID: 69ca2d5847f34ce924b2a00a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### DELETE Accuracy ($D\_f$): 42.86. Result ID: 69ca2d5847f34ce924b2a00f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### GDR Accuracy ($D\_f$): 20.54. Result ID: 69ca2d5847f34ce924b2a010. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### SalUn Accuracy ($D\_f$): 13.66. Result ID: 69ca2d5847f34ce924b2a00e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Our method Accuracy ($D\_f$): 11.95. Result ID: 69ca2d5847f34ce924b2a011. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Two-Stage Machine Unlearning Accuracy ($D\_f$): 11.95. Result ID: 69ca2d5847f34ce924b2a036. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. ### Retrain Accuracy ($D\_f$): 8.58. Result ID: 69ca2d5847f34ce924b2a035. [Reporting paper]() Recorded evaluation configuration: Not recorded. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.