AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

LongWriter-Zero: Mastering Ultra-Long Text Generation via Reinforcement Learning

Yuhao Wu, Yushi Bai, Zhiqiang Hu, Roy Ka-Wei Lee, Juanzi Li

ICLR 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add a think step (Think prompt) to RL training (ablation)LongWriter-Zero (Qwen2.5-32B base) · Arena-WriteElo: 700 -> 1,200 (Base-nothink vs Base-think)Improved
add continual pretraining on 30B writing tokens before RL (ablation)LongWriter-Zero · Arena-WriteElo: 1221 -> 1447Improved
add continual pretraining before RL (ablation)LongWriter-Zero · WritingBenchaverage critic score: 8.12 -> 8.69Improved
add think step (ablation)LongWriter-Zero · WritingBenchaverage critic score: 8.04 -> 8.12Improved
add think step (ablation)LongWriter-Zero · Arena-WriteElo: 668 -> 1221Improved
use RL instead of SFT on the continual-pretrained base (ablation)Qwen2.5-32B continual-pretrained · Arena-WriteElo: 971 -> 1447Improved
add continual pretraining before SFT (ablation)Qwen2.5-32B SFT · Arena-WriteElo: 964 -> 971No clear effect
RL with full pipeline instead of LongWriter synthetic-data SFT (ablation)Qwen2.5-32B · WritingBenchaverage critic score: 8.08 -> 8.69Improved
apply the RL pipeline at 14B scaleQwen2.5-14B · WritingBenchaverage critic score: 7.60 -> 8.49Improved
apply the RL pipeline at 14B scaleQwen2.5-14B · Arena-WriteElo: 744 -> 1198Improved
train with composite length, writing and format rewards from base modelBase-nothink · Arena-WriteElo: 200 -> over 600Improved
normalize each reward component within group before averaging advantagesfinal reward · RL trainingtraining stabilityImproved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy