Pith. sign in

REVIEW 9 cited by

Writing-Zero: Bridge the Gap Between Non-verifiable Tasks and Verifiable Rewards

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2506.00103 v2 pith:NMZKXXJA submitted 2025-05-30 cs.CL

Writing-Zero: Bridge the Gap Between Non-verifiable Tasks and Verifiable Rewards

classification cs.CL
keywords rewardtasksrewardsverifiablewritingnon-verifiablepairwisetraining
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

Reinforcement learning with verifiable rewards (RLVR) has enabled large language models (LLMs) to achieve remarkable breakthroughs in reasoning tasks with objective ground-truth answers, such as mathematics and code generation. However, a significant gap remains for non-verifiable tasks, like creative writing and open-ended dialogue, where quality assessment is inherently subjective and lacks definitive references. Existing approaches for these domains often rely on scalar reward models trained with human preferences, which suffer from limited generalization and are prone to reward hacking, such as over-explanation and length bias. In this work, we propose a unified RLVR-based training paradigm that bridges the gap between non-verifiable tasks and verifiable rewards. We introduce a writing-principle-based pairwise Generative Reward Model (GenRM) and a novel Bootstrapped Relative Policy Optimization (BRPO) algorithm. The pairwise writing GenRM leverages self-principled critique to transform subjective assessments into reliable, verifiable rewards, while BRPO enables dynamic, reference-free pairwise comparison by leveraging a bootstrapped response as temporary reference from within group rollouts during RL training. Our approach empowers LLMs to develop robust writing capabilities without supervised fine-tuning, as demonstrated by Writing-Zero, which shows consistent improvement and strong resistance to reward hacking compared to scalar reward baselines. Furthermore, our method achieves competitive results on both in-house and open-source writing benchmarks. Our findings suggest the potential to unify rule-based, reference-based, and reference-free reward modeling under the RLVR framework, thus paving the way for a comprehensive and scalable RL training paradigm applicable across all language tasks.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 9 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Diagnosing and Repairing Persona Collapse in LLM Advice

    cs.CY 2026-07 conditional novelty 7.0

    LLMs collapse advice into a single supportive persona; Inverse-Process Distillation restores human-like persona diversity, yet raters still prefer the collapsed default.

  2. CapRL++: Unified Reinforcement Learning with Verifiable Rewards for Dense Image and Video Captioning

    cs.CV 2026-06 unverdicted novelty 7.0

    CapRL++ applies reinforcement learning with verifiable rewards to dense image and video captioning by scoring captions via the accuracy of a vision-free LLM answering MCQs from the caption alone.

  3. Reproducing, Analyzing, and Detecting Reward Hacking in Rubric-Based Reinforcement Learning

    cs.LG 2026-06 unverdicted novelty 7.0

    CHERRL is a new controllable testbed for reproducing, analyzing, and detecting reward hacking in rubric-based RL by injecting known biases into LLM-as-a-Judge systems.

  4. PReMISE: Policy Rubrics as Measurement Specifications for LLM Judges

    cs.AI 2026-05 unverdicted novelty 7.0

    PReMISE discovers and audits rubric sets for LLM judges, finding no existing source meets all reliability, preference-fit, and robustness criteria simultaneously while showing two repair methods improve accuracy and r...

  5. StoryAlign: Evaluating and Training Reward Models for Story Generation

    cs.CL 2026-05 unverdicted novelty 7.0

    StoryReward, trained on a new 100k story preference dataset, sets state-of-the-art performance on the introduced StoryRMB benchmark for aligning LLM stories with human preferences.

  6. From Coarse to Fine: Benchmarking and Reward Modeling for Writing-Centric Generation Tasks

    cs.CL 2026-04 unverdicted novelty 6.0

    WEval and WRL introduce fine-grained benchmarking and requirement-selective sample construction for training writing reward models, yielding substantial gains on writing benchmarks with strong generalization.

  7. Teaching Language Models to Forecast Research Success Through Comparative Idea Evaluation

    cs.LG 2026-04 unverdicted novelty 6.0

    Small LMs reach 77.1% accuracy at comparative forecasting of research idea success on benchmarks after supervised fine-tuning, with RLVR yielding interpretable reasoning at 71.35%.

  8. Conversation for Non-verifiable Learning: Self-Evolving LLMs through Meta-Evaluation

    cs.CL 2026-01 unverdicted novelty 6.0

    CoNL lets LLMs self-improve on non-verifiable tasks by rewarding critiques that produce better solutions in multi-agent conversations, jointly optimizing generation and judging without external feedback.

  9. OPERA: Aligning Open-Ended Reasoning via Objective Perplexity-based Reinforcement Learning

    cs.CL 2026-06 unverdicted novelty 5.0

    OPERA uses perplexity dynamics as intrinsic rewards for RL alignment on open-ended tasks, synthesizes a 20k trajectory dataset via guiding words and log-prob rollouts, and sets new SOTA on Qwen3-8B matching some propr...