Imagen de Portada

Retaining by Doing: The Role of On-Policy Data in Mitigating Forgetting

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Chen, Howard, Razin, Noam, Narasimhan, Karthik, Chen, Danqi
Formato:	Preprint
Publicado:	2025
Materias:	Machine Learning Computation and Language
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Unintentional Unalignment: Likelihood Displacement in Direct Preference Optimization
por: Razin, Noam, et al.
Publicado: (2024)

RULE: Reinforcement UnLEarning Achieves Forget-Retain Pareto Optimality
por: Zhang, Chenlong, et al.
Publicado: (2025)

How to Train Long-Context Language Models (Effectively)
por: Gao, Tianyu, et al.
Publicado: (2024)

Why is Your Language Model a Poor Implicit Reward Model?
por: Razin, Noam, et al.
Publicado: (2025)

Understanding Deep Learning via Notions of Rank
por: Razin, Noam
Publicado: (2024)

QuRating: Selecting High-Quality Data for Training Language Models
por: Wettig, Alexander, et al.
Publicado: (2024)

Sharpness-Aware Pretraining Mitigates Catastrophic Forgetting
por: Watts, Ishaan, et al.
Publicado: (2026)

The Heuristic Core: Understanding Subnetwork Generalization in Pretrained Language Models
por: Bhaskar, Adithya, et al.
Publicado: (2024)

SimPO: Simple Preference Optimization with a Reference-Free Reward
por: Meng, Yu, et al.
Publicado: (2024)

What Makes a Reward Model a Good Teacher? An Optimization Perspective
por: Razin, Noam, et al.
Publicado: (2025)

Forgetting to Forget: Attention Sink as A Gateway for Backdooring LLM Unlearning
por: Shang, Bingqi, et al.
Publicado: (2025)

Extracting Rule-based Descriptions of Attention Features in Transformers
por: Friedman, Dan, et al.
Publicado: (2025)

Lory: Fully Differentiable Mixture-of-Experts for Autoregressive Language Model Pre-training
por: Zhong, Zexuan, et al.
Publicado: (2024)

Mitigating Catastrophic Forgetting in Mathematical Reasoning Finetuning through Mixed Training
por: Reynolds, John Graham
Publicado: (2025)

LESS: Selecting Influential Data for Targeted Instruction Tuning
por: Xia, Mengzhou, et al.
Publicado: (2024)

Representing Rule-based Chatbots with Transformers
por: Friedman, Dan, et al.
Publicado: (2024)

LoRA Soups: Merging LoRAs for Practical Skill Composition Tasks
por: Prabhakar, Akshara, et al.
Publicado: (2024)

Odysseus: Scaling VLMs to 100+ Turn Decision-Making in Games via Reinforcement Learning
por: Shi, Chengshuai, et al.
Publicado: (2026)

Do LLMs Really Forget? Evaluating Unlearning with Knowledge Correlation and Confidence Awareness
por: Wei, Rongzhe, et al.
Publicado: (2025)

Interpretability Illusions in the Generalization of Simplified Models
por: Friedman, Dan, et al.
Publicado: (2023)

Language-Guided World Models: A Model-Based Approach to AI Control
por: Zhang, Alex, et al.
Publicado: (2024)

Vanishing Gradients in Reinforcement Finetuning of Language Models
por: Razin, Noam, et al.
Publicado: (2023)

The Surprising Effectiveness of Negative Reinforcement in LLM Reasoning
por: Zhu, Xinyu, et al.
Publicado: (2025)

Cognitive Architectures for Language Agents
por: Sumers, Theodore R., et al.
Publicado: (2023)

Contextual Experience Replay for Self-Improvement of Language Agents
por: Liu, Yitao, et al.
Publicado: (2025)

Scaling Laws for Forgetting during Finetuning with Pretraining Data Injection
por: Bethune, Louis, et al.
Publicado: (2025)

Sheared LLaMA: Accelerating Language Model Pre-training via Structured Pruning
por: Xia, Mengzhou, et al.
Publicado: (2023)

Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training
por: Lai, Song, et al.
Publicado: (2025)

Entropy-Adaptive Fine-Tuning: Resolving Confident Conflicts to Mitigate Forgetting
por: Diao, Muxi, et al.
Publicado: (2026)

Evaluating Large Language Models at Evaluating Instruction Following
por: Zeng, Zhiyuan, et al.
Publicado: (2023)

Elephants Never Forget: Testing Language Models for Memorization of Tabular Data
por: Bordt, Sebastian, et al.
Publicado: (2024)

Improving Language Understanding from Screenshots
por: Gao, Tianyu, et al.
Publicado: (2024)

CURLoRA: Stable LLM Continual Fine-Tuning and Catastrophic Forgetting Mitigation
por: Fawi, Muhammad
Publicado: (2024)

Recover-to-Forget: Gradient Reconstruction from LoRA for Efficient LLM Unlearning
por: Liu, Yezi, et al.
Publicado: (2025)

Faster Transformer Decoding: N-gram Masked Self-Attention
por: Chelba, Ciprian, et al.
Publicado: (2020)

Can Small Language Models Learn, Unlearn, and Retain Noise Patterns?
por: Scaria, Nicy, et al.
Publicado: (2024)

Detecting Pretraining Data from Large Language Models
por: Shi, Weijia, et al.
Publicado: (2023)

Lecture Notes on Linear Neural Networks: A Tale of Optimization and Generalization in Deep Learning
por: Cohen, Nadav, et al.
Publicado: (2024)

PrivUn: Unveiling Latent Ripple Effects and Shallow Forgetting in Privacy Unlearning
por: Chen, Xiaoyi, et al.
Publicado: (2026)

QualEval: Qualitative Evaluation for Model Improvement
por: Murahari, Vishvak, et al.
Publicado: (2023)