List your company

Indirect Object Identification on IOI evaluation episodes (held-out)

2.976Policy ScoreMechRL
2.82722.90162.9763.0504May 25, 2026

Evaluation results

MethodLinks
MechRLK=1, In-episode picks=50, Evaluation episodes=≥ 202026.052.976—0.028
OracleSelection=per-episode, Evaluation episodes=≥ 202026.05—2.948—