Re3: Generating Longer Stories With Recursive Reprompting and Revision
Source: https://arxiv.org/abs/2210.06774
Structured fields
{
"checked_at": "2026-10-17T00:00:00.000Z",
"usefulness": "Plan, draft, rewrite, edit pipeline for 2,000+ word stories; a baseline architecture for agents that must keep plot and character premises consistent over long output.",
"source_type": "paper",
"availability": "available",
"source_owner": "Yang, Tian, Peng, Klein",
"next_check_at": "2026-10-24T00:00:00.000Z",
"known_limitations": [
"Evaluated on GPT-3-era models; results may not transfer to current models."
]
}Evidence claims
- [sourced_fact] Abstract: framework (a) prompts an LLM to construct a structured plan, (b) generates passages by injecting plan and story state into prompts, (c) reranks continuations for plot coherence and premise relevance and (d) edits the best for factual consistency; reports +14% (coherent plot) and +20% (premise relevance) absolute human-judged gains versus direct generation. · locator: arXiv abstract page source
- [upstream_claim] Research review verdict at export: reviewed (as of 2026-10-17). Recorded by the research pipeline, not a hub review outcome; evidence.review stays unreviewed until an independent reviewer records one.
Rights
| component | license | hosting | redistribution | modification | training | Evidence and notes |
|---|---|---|---|---|---|---|
| documentation | unknown | unknown | unknown | unknown | unknown | arXiv non-exclusive distribution license; reuse beyond reading not granted by that license. |