# Instance Unlearning on Tiny-ImageNet 1% forgetting (test) Source: [Instance Unlearning on Tiny-ImageNet 1% forgetting (test)]() Collection index: [llms.txt]() Data updated: 2026-02-26T05:26:52.632Z Evaluation of Instance Unlearning performance on the test split of Tiny-ImageNet, specifically focusing on a scenario involving 1% data forgetting, measured by residual accuracy, forgetting accuracy, MIA, and Avg.G. 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 | Instance Unlearning | | Dataset (including any stated split) | Tiny-ImageNet 1% forgetting (test) | | Metric | Acc (Train Residual) | | Unit | PERCENT | | Direction | Higher is better | | Results loaded | 10 of 10 | | 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 - [Acc (Train Residual)](): Higher is better; unit: PERCENT. - [Acc (Train Forgetting)](): Higher is better; unit: PERCENT. - [Acc (Test Residual)](): Higher is better; unit: PERCENT. - [MIA](): Lower is better; unit: PERCENT. - [Avg.G.](): Higher is better; unit: SCALAR. ## Recorded results | Method | Acc (Train Residual) | Result date | Report paper | | --- | --- | --- | --- | | retrained | 99.67 | Not recorded | 69802fbfa8b01e76cc8e1752 | | pretrained | 99.63 | Not recorded | 69802fbfa8b01e76cc8e1752 | | SalUn | 98.47 | Not recorded | 69802fbfa8b01e76cc8e1752 | | GA | 98.37 | Not recorded | 69802fbfa8b01e76cc8e1752 | | COUN+RL | 98.3 | Not recorded | 69802fbfa8b01e76cc8e1752 | | RL | 97.56 | Not recorded | 69802fbfa8b01e76cc8e1752 | | BT | 96.46 | Not recorded | 69802fbfa8b01e76cc8e1752 | | L2UL | 94.77 | Not recorded | 69802fbfa8b01e76cc8e1752 | | LOFT | 92.65 | Not recorded | 69802fbfa8b01e76cc8e1752 | | DELETE | 91.84 | Not recorded | 69802fbfa8b01e76cc8e1752 | ### retrained Acc (Train Residual): 99.67. Result ID: 6998efcd5698665ec4547b27. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":true,"D\_fg":true,"RTE":857.21,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### pretrained Acc (Train Residual): 99.63. Result ID: 6998efcd5698665ec4547b26. [Reporting paper]() Recorded evaluation configuration: {"D\_rm":true,"D\_fg":true,"RTE":882.05,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### SalUn Acc (Train Residual): 98.47. Result ID: 6998efcd5698665ec4547b2a. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":12.79,"# param. (%)":"13,836,577 (50%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### GA Acc (Train Residual): 98.37. Result ID: 6998efcd5698665ec4547b28. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":7.92,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### COUN+RL Acc (Train Residual): 98.3. Result ID: 6998efcd5698665ec4547b2d. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":11.45,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### RL Acc (Train Residual): 97.56. Result ID: 6998efcd5698665ec4547b29. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":7.61,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### BT Acc (Train Residual): 96.46. Result ID: 6998efcd5698665ec4547b2b. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":8.23,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### L2UL Acc (Train Residual): 94.77. Result ID: 6998efcd5698665ec4547b2c. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":7919.11,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### LOFT Acc (Train Residual): 92.65. Result ID: 6998efcd5698665ec4547b2f. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":"only Sigma\_pre\_rm","D\_fg":"only Sigma\_pre\_fg","RTE":0.5,"# param. (%)":"230,400 (0.83%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. ### DELETE Acc (Train Residual): 91.84. Result ID: 6998efcd5698665ec4547b2e. [Reporting paper]() [Method origin paper]() Recorded evaluation configuration: {"D\_rm":false,"D\_fg":true,"RTE":8.42,"# param. (%)":"27,673,154 (100%)","Backbone":"Swin-T"}. This record does not assert independent reproduction. Page 1 of 1. This is an index page, not the full collection.