NADS (Navigator‑Guided Data Selection) is a fine‑tuning framework for large language models that mitigates catastrophic forgetting when the original pretraining data are unavailable. It first trains a navigator model without preservation regularization to detect capability drift, then scores synthetic data candidates using the predictive divergence between the navigator and the pretrained model. Finally, it selects a diverse, utility‑aware subset of synthetic data and fine‑tunes while distilling pretrained behavior, achieving a better balance between new‑task performance and general capability preservation with reduced selection overhead.
By Jianhao Zhang, Ou Wu, Yi Du
arXiv:2609.27355v1 Announce Type: cross
Abstract: Unlearning ensures LLM compliance by removing the influence of private or copyrighted training data. However, since LLM models typically undergo post...
By Jialu Wang, Jianing Deng, Shuqing Luo, Yuanzhe Li, Dongwei Wang, Jingtong Hu, Huanrui Yang, Song Wang, Tianlong Chen
Dataset distillation (DD) condenses large corpora into compact, information-rich subsets for efficient training and reuse. However, under noisy supervision, DD risks condensing corrupted associations together with useful signals, degrading robustness.
arXiv:2605. 12765v3 Announce Type: replace Abstract: Large Language Models memorize vast amounts of training data, raising concerns regarding privacy, copyright infringement, and safety.
By Vin\'icius Conte Turani, Ot\'avio Parraga, Jo\~ao Vitor Boer Abitante, Kristen K. Arguello, Joana Pasquali, Ramiro N. Barros, Flavio du Pin Calmon, Christian Mattjie, Rodrigo C. Barros, Lucas S. Kupssinsk\"u
arXiv:2607. 12112v1 Announce Type: cross Abstract: Federated fine-tuning of Multimodal Large Language Models (MLLMs) across distributed networks enables privacy-sensitive adaptation to evolving data streams, yet a fundamental obstacle prevents robust deployment in dynamic environments: catastrophic forgetting, wherein sequential task updates erase previously acquired knowledge across visual, linguistic, and cross-modal representations.
By Jing Liu, Chenxuanyin Zou, Jiayang Ren, Gaoyun Fang, Chengfang Li, Yan Wang, Zhenchao Ma, Bo Hu
The paper introduces HiDeR, a High Information Density Replay framework for Lifelong Person Re-Identification that replaces discrete sample selection with information compression. It uses a complexity‑aware memory allocation based on intra‑class variance and a metric‑guided condensation objective to preserve essential identity topologies, while a cross‑modality adaptation strategy bridges synthetic and real styles to improve training. Experiments show HiDeR outperforms state‑of‑the‑art methods in knowledge retention and generalization, and reduces cumulative replay cost.
By Mingyu Wang, Wei Jiang, Haojie Liu, Zhiyong Li, Weijie Mao
arXiv:2609.05966v1 Announce Type: new
Abstract: Existing unlearning approaches typically rely on post hoc weight adaptation or distillation, leading to duplicated memory costs, degraded generalizatio...
By Kushal Chakrabarti, Mayank Baranwal
The paper introduces Dr. Post-Training, a framework that treats general training data as a data‑induced regularizer to prevent overfitting to scarce target objectives, rather than merely a pool for selection. It constructs a feasible set of model update directions from general data and projects target‑driven updates onto this set, unifying standard training and existing data‑selection methods as special cases along a bias‑variance spectrum. Experiments on SFT, RLHF, and RLVR demonstrate that these methods outperform both standard training and state‑of‑the‑art data‑selection baselines, especially when general and target data are highly misaligned, while system optimizations keep overhead minimal.
By Pingbang Hu, Xueshen Liu, Z. Morley Mao, Jiaqi W. Ma
The paper introduces a framework for aligning the inductive bias of linear time‑invariant State Space Models (SSMs) with task‑specific spectral characteristics. By formalizing the bias through an SSM‑induced kernel and showing its spectrum is governed by the model’s frequency response, the authors propose Task‑Dependent Initialization (TDI), a fast power‑spectrum matching method. Experiments on synthetic data, one‑layer SSMs, and deep SSMs across real‑world benchmarks demonstrate that TDI improves data‑efficient generalization when the task’s spectral structure differs from the default SSM bias.
By Qiyu Chen, Guozhang Chen
The paper introduces LARE, a method that adjusts domain weights during large language model pretraining by reweighting losses based on running loss statistics, rather than altering data exposure. Experiments on a 1.5B model trained with 240B tokens show that LARE improves performance across five benchmarks, achieving significant gains on GSM8K (+17.06 pp) and MATH (+9.00 pp). The approach demonstrates that balancing loss scales can enhance joint capabilities without increasing domain exposure.
By Jiazheng Li, Huaqing Zhang, Jing Xu, Tianle Cai, Jingzhao Zhang
arXiv:2610.02126v1 Announce Type: cross
Abstract: We explore catastrophic forgetting in the context of large pre-trained models. By considering forgetting as a geometric problem in the input space of...
By Assaf Ben-Kish, Akarsh Kumar, James Glass, Raja Giryes
arXiv:2603. 11201v3 Announce Type: replace-cross Abstract: The world is inherently dynamic, and continual learning aims to enable models to adapt to ever-evolving data streams.
By Haihua Luo, Xuming Ran, Tommi K\"arkk\"ainen, Huiyan Xue, Zhonghua Chen, Qi Xu, Fengyu Cong