paper-with-me

Papers

SLIME: Stabilized Likelihood Implicit Margin Enforcement for Preference Optimization

2026-02-02 · Maksim Afanasyev, Illarion Iov arxiv

Direct preference optimization methods have emerged as a computationally efficient alternative to Reinforcement Learning from Human Feedback (RLHF) for aligning Large Language Models (LLMs). Latest approaches have streamlined the alignment process by deriving implicit reward functions, yet they often suffer from a critical objective mismatch: optimizing the relative margin between chosen and rejected responses does not guarantee the preservation of the chosen response's absolute likelihood. This can lead to unlearning, where the model degrades the probability of high-quality outputs to satisfy margin constraints, and formatting collapse caused by the over-penalization of rejected sequences. In this work, we introduce SLIME (Stabilized Likelihood Implicit Margin Enforcement), a reference-free alignment objective designed to decouple preference learning from generation quality. SLIME incorporates a three-pronged objective: (1) an anchoring term to maximize the likelihood of preferred responses; (2) a stabilizing penalty that prevents the probabilities of rejected tokens from collapsing to zero; and (3) a dual-margin mechanism that combines hard and soft constraints for precise boundary shaping. Our results demonstrate that SLIME achieves superior performance compared to state-of-the-art baselines while maintaining higher generation stability.

📄 PDF Abstract BibTeX arXiv:2602.02383

Code (0)

등록된 구현이 없습니다.

Tasks

Reinforcement Learning

Similar Papers 제목 키워드 기반

IRLS and Slime Mold: Equivalence and Convergence

2016-01-12 · Damian Straszak, Nisheeth K. Vishnoi

In this paper we present a connection between two dynamical systems arising in entirely different contexts: one in signal processing and the other in biology. The first is the famous Iteratively Reweighted Least Squares …

compressed sensing

Stabilized Likelihood-based Imitation Learning via Denoising Continuous Normalizing Flow

2021-09-29 · Xin Zhang, Yanhua Li, Ziming Zhang, Christopher Brinton 외

State-of-the-art imitation learning (IL) approaches, e.g, GAIL, apply adversarial training to minimize the discrepancy between expert and learner behaviors, which is prone to unstable training and mode collapse. In this …

DenoisingImitation Learning

SLiMe: Segment Like Me

2023-09-06 · Aliasghar Khani, Saeid Asgari Taghanaki, Aditya Sanghi, Ali Mahdavi Amiri 외

Significant strides have been made using large vision-language models, like Stable Diffusion (SD), for a variety of downstream tasks, including image editing, image correspondence, and 3D shape generation. Inspired by th…

3D Shape GenerationSegmentation

Offline Preference Optimization via Maximum Marginal Likelihood Estimation

2025-10-27 · Saeed Najafi, Alona Fyshe arxiv

Aligning Large Language Models (LLMs) with human preferences is crucial, but standard methods like Reinforcement Learning from Human Feedback (RLHF) are often complex and unstable. In this work, we propose a new, simpler…

Reinforcement Learning

SLIMER-IT: Zero-Shot NER on Italian Language

2024-09-24 · Andrew Zamai, Leonardo Rigutini, Marco Maggini, Andrea Zugarini

Traditional approaches to Named Entity Recognition (NER) frame the task into a BIO sequence labeling problem. Although these systems often excel in the downstream task at hand, they require extensive annotated data and s…

named-entity-recognitionNamed Entity RecognitionNamed Entity Recognition (NER)NER