AI ethics evidence index Search
Source-linked daily record
Evidence record 4927 · automatically gathered

RLearner-LLM: Balancing Logical Grounding and Fluency in Large Language Models via Hybrid Direct Preference Optimization

Direct Preference Optimization (DPO), the efficient alternative to PPO-based RLHF, falls short on knowledge-intensive generation: standard preference signals from human annotators or LLM judges exhibit a systematic verbosity bias that rewards fluency over logical correctness. This blindspot leaves a logical alignment gap -- SFT models reach NLI entailment of only 0.05-0.22 despite producing fluent text. We propose RLearner-LLM with Hybrid-DPO: an automated preference pipeline that fuses a DeBERT

Record details

Published: 6 May 2026
Source: arXiv
Category: Research
Topics: Bias & fairness · Safety & alignment
Retrieved: 14 July 2026

source-onlyevidence status

These records share source-supplied organisations, an exact publisher byline, automatic topics or regions. The reason is shown on every link; related does not mean supporting, agreeing with or verifying this record.

Shared topic: Bias & fairness · Shared topic: Safety & alignment

Misaligned by Reward: Socially Undesirable Preferences in LLMs

arXiv · 6 May 2026

Shared topic: Bias & fairness · Shared topic: Safety & alignment

Brainrot: Deskilling and Addiction are Overlooked AI Risks

arXiv · 5 May 2026

Shared topic: Bias & fairness · Shared topic: Safety & alignment

Resolving the bias-precision paradox with stochastic causal representation learning for personalized medicine

arXiv · 7 May 2026

Shared topic: Bias & fairness · Shared topic: Safety & alignment

Strat-LLM: Stratified Strategy Alignment for LLM-based Stock Trading with Real-time Multi-Source Signals

arXiv · 7 May 2026

Shared topic: Bias & fairness · Shared topic: Safety & alignment

Moral Sensitivity in LLMs: A Tiered Evaluation of Contextual Bias via Behavioral Profiling and Mechanistic Interpretability

arXiv · 4 May 2026

Shared topic: Bias & fairness · Shared topic: Safety & alignment

Quality-Conditioned Agreement in Automated Short Answer Scoring: Mid-Range Degradation and the Impact of Task-Specific Adaptation

arXiv · 8 May 2026

How to cite this record

ethics.ai (6 May 2026), “RLearner-LLM: Balancing Logical Grounding and Fluency in Large Language Models via Hybrid Direct Preference Optimization,” evidence record 4927, https://ethics.ai/record/4927 (originally published by arXiv).

JSON

Use and limitations

This page is a stable index and citation surface for a source record. ethics.ai did not author the underlying report and has not independently verified every claim. Automatic topics may be imperfect. For consequential use, quote and cite the original publisher.