Bias Detection
Bias detection is the task of detecting and measuring racism, sexism and otherwise discriminatory behavior in a model (Source: https://stereoset.mit.edu/)
Papers
Showing 1–10 of 199 papers
Benchmark Results
| # | Model | Metric | Claimed | Verified | Status |
|---|---|---|---|---|---|
| 1 | GPT-2 (small) | ICAT Score | 72.97 | — | Unverified |
| 2 | XLNet (large) | ICAT Score | 72.03 | — | Unverified |
| 3 | GPT-2 (medium) | ICAT Score | 71.73 | — | Unverified |
| 4 | BERT (base) | ICAT Score | 71.21 | — | Unverified |
| 5 | GPT-2 (large) | ICAT Score | 70.54 | — | Unverified |
| 6 | BERT (large) | ICAT Score | 69.89 | — | Unverified |
| 7 | RoBERTa (base) | ICAT Score | 67.5 | — | Unverified |
| 8 | GAL 120B | ICAT Score | 65.6 | — | Unverified |
| 9 | XLNet (base) | ICAT Score | 62.1 | — | Unverified |
| 10 | GPT-3 (text-davinci-002) | ICAT Score | 60.8 | — | Unverified |