paper-with-me

Papers

Rethinking Data Distillation: Do Not Overlook Calibration

2023-07-24 · ICCV 2023 1 · Dongyao Zhu, Bowen Lei, Jie Zhang, Yanbo Fang, Ruqi Zhang, Yiqun Xie, Dongkuan Xu

Neural networks trained on distilled data often produce over-confident output and require correction by calibration methods. Existing calibration methods such as temperature scaling and mixup work well for networks trained on original large-scale data. However, we find that these methods fail to calibrate networks trained on data distilled from large source datasets. In this paper, we show that distilled data lead to networks that are not calibratable due to (i) a more concentrated distribution of the maximum logits and (ii) the loss of information that is semantically meaningful but unrelated to classification tasks. To address this problem, we propose Masked Temperature Scaling (MTS) and Masked Distillation Training (MDT) which mitigate the limitations of distilled data and achieve better calibration results while maintaining the efficiency of dataset distillation.

📄 PDF Abstract BibTeX arXiv:2307.12463

Code (1)

Guang000/Awesome-Dataset-Distillation

Tasks

Dataset Distillation

Methods 이 논문이 사용한 방법론

fail 설명 없음
Mixup Mixup is a data augmentation technique that generates a weighted combination of random image pairs from the training data. Given two images and their ground truth labels:…

Similar Papers 제목 키워드 기반

Rethinking the Knowledge Distillation From the Perspective of Model Calibration

2021-10-31 · Lehan Yang, Jincen Song

Recent years have witnessed dramatically improvements in the knowledge distillation, which can generate a compact student model for better efficiency while retaining the model effectiveness of the teacher model. Previous…

Knowledge Distillation

BD-KD: Balancing the Divergences for Online Knowledge Distillation

2022-12-25 · Ibtihel Amara, Nazanin Sepahvand, Brett H. Meyer, Warren J. Gross 외

We address the challenge of producing trustworthy and accurate compact models for edge devices. While Knowledge Distillation (KD) has improved model compression in terms of achieving high accuracy performance, calibratio…

Knowledge DistillationModel CompressionTransfer Learning

Faithful Knowledge Distillation

2023-06-07 · Tom A. Lamb, Rudy Brunel, Krishnamurthy Dj Dvijotham, M. Pawan Kumar 외

Knowledge distillation (KD) has received much attention due to its success in compressing networks to allow for their deployment in resource-constrained systems. While the problem of adversarial robustness has been studi…

Adversarial RobustnessKnowledge Distillation

Rethinking the Role of Temperature in Large Language Model Distillation

2026-05-29 · Hoang-Chau Luong, Lingwei Chen arxiv

Reverse Kullback-Leibler (RKL) divergence is widely favored over forward KL (FKL) in large language models (LLM) distillation, yet this preference is largely based on comparisons that omit the temperature $τ$, overlookin…

Rethinking Reverse KL as Adaptive Entropy Distillation

2026-08-05 · Shizhen Li, Zhiyu Shen, Yuyin Lu, Yunhe Pang 외 arxiv

Knowledge distillation (KD) is widely used to transfer the capabilities of large language models (LLMs) to smaller students, but existing objectives often struggle to balance faithful imitation and robust generation. In …

Mathematical ReasoningKnowledge Distillation