ResearchBenchmarksSingle-Monster RTFM Task on RTFM Single-MonsterSingle-Monster RTFM Task on RTFM Single-MonsterFollow99One SL Performancetxt2π35.5652.0368.584.97Jun 21, 2026Evaluation resultsMethodMethodLinksOne SL PerformanceOne SL (Shuffle) PerformanceOne NL PerformanceOne NL (Shuffle) Performancetxt2πTraining steps=100MTraining steps=100M2026.0699685050Reader2026.0699857963HRLLI2026.0699928379ReaderArchitecture=TransformerArchitecture=Transformer2026.0697907559Oracle MCTS2026.0697979797txt2π2026.0638413942