TOFU
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Machine Unlearning | TOFU Forget 10% | Aggregation Score65 | 81 | |
| Model Unlearning | TOFU Forget 5% 1.0 | Model Utility8.519 | 60 | |
| Machine Unlearning | TOFU (5%) | Forget Quality86.6 | 59 | |
| Language Model Unlearning | TOFU (Forget10) | Forget Quality (FQ)100 | 54 | |
| Machine Unlearning | TOFU Forget 1% | Aggregation Score66 | 54 | |
| Machine Unlearning | TOFU 1.0 (Forget10) | Model Utility (MU)52 | 53 | |
| Machine Unlearning | TOFU forget05 1.0 | Model Utility (MU)75.85 | 53 | |
| Machine Unlearning | TOFU 1.0 (forget01) | Average Score80.1 | 53 | |
| Machine Unlearning | TOFU Forget01 (1% authors) | Forget Quality (Rouge-L)0.99 | 48 | |
| Machine Unlearning | TOFU 1.0 (Retain Set) | ROUGE-L100 | 48 | |
| Machine Unlearning | TOFU 1.0 (Real Author) | ROUGE-L93 | 45 | |
| Machine Unlearning | TOFU | Forget Quality (FQ)1 | 43 | |
| Machine Unlearning | TOFU Forget10 (10% authors split) | Forget Quality - Rouge-L0.99 | 42 | |
| Machine Unlearning | TOFU Forget05 (5% authors) | Forget Quality (ROUGE-L)0.99 | 42 | |
| Knowledge Recovery | TOFU 10% (400 samples) 1.0 (forget) | ASR60 | 42 | |
| Machine Unlearning | TOFU (10%) | Forget Quality (FQ)1 | 37 | |
| Machine Unlearning | TOFU (1%) | Forget Quality (FQ)0.0002 | 36 | |
| Machine Unlearning | TOFU World Fact 1.0 | ROUGE-L0.884 | 34 | |
| Machine Unlearning | TOFU Forget05 Phi-1.5B model (5%) | Model Utility (MU)51.52 | 32 | |
| Machine Unlearning | TOFU Avg. | Log Forget Quality (-log(FQ))14.32 | 30 | |
| Machine Unlearning | TOFU 1% forget | -log(FQ)2.896 | 30 | |
| Machine Unlearning | TOFU 5% forget split | -log(FQ)13.955 | 30 | |
| Machine Unlearning | TOFU 10% forget | -log(FQ)27.159 | 30 | |
| Machine Unlearning | TOFU (Split99) | Forget Quality1 | 28 | |
| Machine Unlearning | TOFU Forget 5% | Aggregation Metric64 | 27 |
Showing 25 of 117 rows