Zero-shot Question Answering on OpenBookQA
28.4AccuracyYat (sb+α)
Evaluation results
| Method | Links | |
|---|---|---|
| Yat (sb+α)Parameterization=shared (b, ε), Scaling=learnable per-channel, Evaluation mode=zero-shot2026.05 | 28.4 | |
| Yat (pn+α)Parameterization=per-neuron, Scaling=learnable per-channel, Evaluation mode=zero-shot2026.05 | 28.2 | |
| GELUArchitecture=GELU MLP, Evaluation mode=zero-shot2026.05 | 27.87 | |
| Yat (pn+ca)Parameterization=per-neuron, Scaling=constant per-channel, Evaluation mode=zero-shot2026.05 | 27.47 | |
| Yat (sb+ca)Parameterization=shared (b, ε), Scaling=constant per-channel, Evaluation mode=zero-shot2026.05 | 27.2 |