Multi-branch Story Generation on ReedsyPrompts
0.493BLEU ScoreOursglobal
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| OursglobalBase Model=Llada-8B2026.04 | 0.493 | 0.552 | 0.565 | 0.778 | 0.212 | 0.157 | |
| UAG (global)Base Model=Llama-3B, Scope=Global2026.04 | 0.254 | 0.3119 | 0.319 | 0.657 | 0.4415 | 0.3875 | |
| NaiveBase Model=Llada-8B2026.04 | 0.18 | 0.344 | 0.356 | 0.517 | 0.316 | 0.564 | |
| UAG (local)Base Model=Llama-3B, Scope=Local2026.04 | 0.1014 | 0.212 | 0.2943 | 0.61 | 0.3905 | 0 | |
| NaiveBase Model=Llama-3B2026.04 | 0.084 | 0.241 | 0.257 | 0.47 | 0.293 | 0.005 | |
| OurslocalBase Model=Llada-8B2026.04 | 0.082 | 0.213 | 0.269 | 0.625 | 0.278 | 0.1 | |
| OursBase Model=Llada-8B2026.04 | 0.067 | 0.181 | 0.227 | 0.524 | 0.434 | 0.323 | |
| Top-pBase Model=Llama-3B2026.04 | 0.0573 | 0.1978 | 0.2026 | 0.4041 | 0.4755 | 0.193 | |
| UAGBase Model=Llama-3B2026.04 | 0.0165 | 0.093 | 0.136 | 0.43 | 0.6795 | 0.5645 | |
| Avoidance DecodingBase Model=Llama-3B2026.04 | 0.0112 | 0.1265 | 0.1483 | 0.4136 | 0.82 | 0.25 | |
| Top-k†Base Model=Llama-3B2026.04 | 0.0036 | 0.0059 | 0.018 | 0.2457 | 0.93 | 0.98 |