Question Answering on TriviaQA (Mean per-step regret)
0.131Mean Per-Step RegretNo-Rewrite (NoRw)
Evaluation results
| Method | Links | |
|---|---|---|
| No-Rewrite (NoRw)Strategy Category=Base2026.02 | 0.131 | |
| ParaStrategy Category=Static Prompts2026.02 | 0.145 | |
| TSStrategy Category=Non-Contextual2026.02 | 0.148 | |
| LinEXP3Strategy Category=Contextual Linear2026.02 | 0.148 | |
| FTPLStrategy Category=Non-Contextual2026.02 | 0.15 | |
| SimplStrategy Category=Static Prompts2026.02 | 0.151 | |
| EXP3Strategy Category=Non-Contextual2026.02 | 0.153 | |
| LinUCB+KLStrategy Category=Contextual Linear2026.02 | 0.153 | |
| ϵ-FTRLStrategy Category=Non-Contextual2026.02 | 0.154 | |
| LinUCBStrategy Category=Contextual Linear2026.02 | 0.155 | |
| TSStrategy Category=Contextual Linear2026.02 | 0.155 | |
| LinFTPLStrategy Category=Contextual Linear2026.02 | 0.157 | |
| ExpandStrategy Category=Static Prompts2026.02 | 0.16 | |
| DisambStrategy Category=Static Prompts2026.02 | 0.162 | |
| ClarifyStrategy Category=Static Prompts2026.02 | 0.167 |