List your company

Multi-step rollout prediction on 7 Environments Suite (held-out episodes)

69Token F1 (t=1)PatchWorld-Residual
55.4858.9962.566.01May 29, 2026

Evaluation results

MethodLinks
PatchWorld-Residual2026.0569646058
LLM-Direct2026.0564636059
WorldCoder2026.0563605857
PoE-World2026.0563575351
PatchWorld-Simple2026.0556504340