List your company

Question Answering on HotpotQA (Mean per-step regret)

0.188Mean Per-Step Regretϵ-FTRL
0.187280.192140.1970.20186Feb 23, 2026

Evaluation results

MethodLinks
ϵ-FTRLStrategy Category=Non-Contextual2026.020.188
DisambStrategy Category=Static Prompts2026.020.191
LinUCBStrategy Category=Contextual Linear2026.020.191
LinEXP3Strategy Category=Contextual Linear2026.020.192
TSStrategy Category=Contextual Linear2026.020.194
LinFTPLStrategy Category=Contextual Linear2026.020.196
EXP3Strategy Category=Non-Contextual2026.020.197
TSStrategy Category=Non-Contextual2026.020.197
LinUCB+KLStrategy Category=Contextual Linear2026.020.197
No-Rewrite (NoRw)Strategy Category=Base2026.020.198
SimplStrategy Category=Static Prompts2026.020.199
FTPLStrategy Category=Non-Contextual2026.020.199
ExpandStrategy Category=Static Prompts2026.020.201
ParaStrategy Category=Static Prompts2026.020.203
ClarifyStrategy Category=Static Prompts2026.020.206