paper-with-me

Papers

Multi-Stage Balanced Distillation: Addressing Long-Tail Challenges in Sequence-Level Knowledge Distillation

2024-06-19 · YuHang Zhou, Jing Zhu, Paiheng Xu, Xiaoyu Liu, Xiyao Wang, Danai Koutra, Wei Ai, Furong Huang

Large language models (LLMs) have significantly advanced various natural language processing tasks, but deploying them remains computationally expensive. Knowledge distillation (KD) is a promising solution, enabling the transfer of capabilities from larger teacher LLMs to more compact student models. Particularly, sequence-level KD, which distills rationale-based reasoning processes instead of merely final outcomes, shows great potential in enhancing students' reasoning capabilities. However, current methods struggle with sequence level KD under long-tailed data distributions, adversely affecting generalization on sparsely represented domains. We introduce the Multi-Stage Balanced Distillation (BalDistill) framework, which iteratively balances training data within a fixed computational budget. By dynamically selecting representative head domain examples and synthesizing tail domain examples, BalDistill achieves state-of-the-art performance across diverse long-tailed datasets, enhancing both the efficiency and efficacy of the distilled models.

📄 PDF Abstract BibTeX arXiv:2406.13114

Code (1)

tonyzhou98/long_tail_kd 공식 구현 pytorch

Tasks

Knowledge Distillation

Methods 이 논문이 사용한 방법론

Knowledge Distillation A very simple way to improve the performance of almost any machine learning algorithm is to train many different models on the same data and then to average their predictions.…

Similar Papers 제목 키워드 기반

Class-Balanced Distillation for Long-Tailed Visual Recognition

2021-04-12 · Ahmet Iscen, André Araujo, Boqing Gong, Cordelia Schmid

Real-world imagery is often characterized by a significant imbalance of the number of images per class, leading to long-tailed distributions. An effective and simple approach to long-tailed visual recognition is to learn…

Image ClassificationKnowledge DistillationLong-tail Learning

FedNoRo: Towards Noise-Robust Federated Learning by Addressing Class Imbalance and Label Noise Heterogeneity

2023-05-09 · Nannan Wu, Li Yu, Xuefeng Jiang, Kwang-Ting Cheng 외

Federated noisy label learning (FNLL) is emerging as a promising tool for privacy-preserving multi-source decentralized learning. Existing research, relying on the assumption of class-balanced global data, might be incap…

Federated LearningKnowledge DistillationPrivacy Preserving

Self Supervision to Distillation for Long-Tailed Visual Recognition

2021-09-09 · ICCV 2021 10 · TianHao Li, LiMin Wang, Gangshan Wu

Deep learning has achieved remarkable progress for visual recognition on large-scale balanced datasets but still performs poorly on real-world long-tailed data. Previous methods often adopt class re-balanced training str…

Long-tail Learning

REMEDI: Relative Feature Enhanced Meta-Learning with Distillation for Imbalanced Prediction

2025-05-12 · Fei Liu, Huanhuan Ren, Yu Guan, Xiuxu Wang 외

Predicting future vehicle purchases among existing owners presents a critical challenge due to extreme class imbalance (<0.5% positive rate) and complex behavioral patterns. We propose REMEDI (Relative feature Enhanced M…

Meta-Learning

Long-tailed Adversarial Training with Self-Distillation

2025-03-09 · Seungju Cho, Hongsin Lee, Changick Kim

Adversarial training significantly enhances adversarial robustness, yet superior performance is predominantly achieved on balanced datasets. Addressing adversarial robustness in the context of unbalanced or long-tailed d…

Adversarial Robustness