One-step next-observation prediction on Wordle (test)
0.72Token F1PatchWorld-Residual
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| PatchWorld-ResidualBackbone=Qwen3-Coder-480B, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.72 | 0.44 | |
| PatchWorld-SimpleBackbone=DeepSeek-V4-Flash, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.69 | 0.4 | |
| PatchWorld-ResidualBackbone=Mimo-v2.5, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.63 | 0.35 | |
| Word2WorldBackbone=Qwen3.5-4B, LLM Usage=SFT, Infer. LLM?=Yes2026.05 | 0.6 | 0.37 | |
| LLM-DirectBackbone=Qwen3-Coder-480B, LLM Usage=ICL, Infer. LLM?=Yes2026.05 | 0.6 | 0.21 | |
| PatchWorld-ResidualBackbone=DeepSeek-V4-Flash, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.6 | 0.34 | |
| PatchWorld-SimpleBackbone=Qwen3-Coder-480B, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.58 | 0.2 | |
| WorldCoderBackbone=DeepSeek-V4-Flash, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.55 | 0.22 | |
| PoE-WorldBackbone=DeepSeek-V4-Flash, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.55 | 0.22 | |
| LLM-DirectBackbone=DeepSeek-V4-Flash, LLM Usage=ICL, Infer. LLM?=Yes2026.05 | 0.53 | 0.19 | |
| LLM-DirectBackbone=Mimo-v2.5, LLM Usage=ICL, Infer. LLM?=Yes2026.05 | 0.51 | 0.24 | |
| PatchWorld-SimpleBackbone=Mimo-v2.5, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.47 | 0.16 | |
| PoE-WorldBackbone=Qwen3-Coder-480B, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.43 | 0.17 | |
| PoE-WorldBackbone=Mimo-v2.5, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.4 | 0.16 | |
| WorldCoderBackbone=Mimo-v2.5, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.35 | 0.13 | |
| WorldCoderBackbone=Qwen3-Coder-480B, LLM Usage=CodeGen, Infer. LLM?=No2026.05 | 0.31 | 0.11 |