UNDO: Understanding Distillation as Optimization

• URL: http://arxiv.org/abs/2504.02521v1
• Date: Thu, 03 Apr 2025 12:18:51 GMT
• Title: UNDO: Understanding Distillation as Optimization
• Authors: Kushal Jain, Piyushi Goyal, Kumar Shridhar,
• Abstract summary: We introduce the UNDO: UNderstanding Distillation as Optimization framework.<n>Each iteration directly targets the student's learning deficiencies, motivating the teacher to provide tailored and enhanced rationales.<n> Empirical evaluations on various challenging mathematical and commonsense reasoning tasks demonstrate that our iterative distillation method, UNDO, significantly outperforms standard one-step distillation methods.
• Score: 9.100811514331498
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Knowledge distillation has emerged as an effective strategy for compressing large language models' (LLMs) knowledge into smaller, more efficient student models. However, standard one-shot distillation methods often produce suboptimal results due to a mismatch between teacher-generated rationales and the student's specific learning requirements. In this paper, we introduce the UNDO: UNderstanding Distillation as Optimization framework, designed to bridge this gap by iteratively identifying the student's errors and prompting the teacher to refine its explanations accordingly. Each iteration directly targets the student's learning deficiencies, motivating the teacher to provide tailored and enhanced rationales that specifically address these weaknesses. Empirical evaluations on various challenging mathematical and commonsense reasoning tasks demonstrate that our iterative distillation method, UNDO, significantly outperforms standard one-step distillation methods, achieving performance gains of up to 20%. Additionally, we show that teacher-generated data refined through our iterative process remains effective even when applied to different student models, underscoring the broad applicability of our approach. Our work fundamentally reframes knowledge distillation as an iterative teacher-student interaction, effectively leveraging dynamic refinement by the teacher for better knowledge distillation.

Related papers

• The Staged Knowledge Distillation in Video Classification: Harmonizing Student Progress by a Complementary Weakly Supervised Framework [21.494759678807686]
  We propose a new weakly supervised learning framework for knowledge distillation in video classification. Our approach leverages the concept of substage-based learning to distill knowledge based on the combination of student substages and the correlation of corresponding substages. Our proposed substage-based distillation approach has the potential to inform future research on label-efficient learning for video data.
  arXiv  Detail & Related papers  (2023-07-11T12:10:42Z)
• Student-friendly Knowledge Distillation [1.5469452301122173]
  We propose student-friendly knowledge distillation (SKD) to simplify teacher output into new knowledge representations. SKD contains a softening processing and a learning simplifier. The experimental results on the CIFAR-100 and ImageNet datasets show that our method achieves state-of-the-art performance.
  arXiv  Detail & Related papers  (2023-05-18T11:44:30Z)
• Tailoring Instructions to Student's Learning Levels Boosts Knowledge Distillation [52.53446712834569]
  Learning Good Teacher Matters (LGTM) is an efficient training technique for incorporating distillation influence into the teacher's learning process. Our LGTM outperforms 10 common knowledge distillation baselines on 6 text classification tasks in the GLUE benchmark.
  arXiv  Detail & Related papers  (2023-05-16T17:50:09Z)
• HomoDistil: Homotopic Task-Agnostic Distillation of Pre-trained Transformers [49.79405257763856]
  This paper focuses on task-agnostic distillation. It produces a compact pre-trained model that can be easily fine-tuned on various tasks with small computational costs and memory footprints. We propose Homotopic Distillation (HomoDistil), a novel task-agnostic distillation approach equipped with iterative pruning.
  arXiv  Detail & Related papers  (2023-02-19T17:37:24Z)
• EmbedDistill: A Geometric Knowledge Distillation for Information Retrieval [83.79667141681418]
  Large neural models (such as Transformers) achieve state-of-the-art performance for information retrieval (IR) We propose a novel distillation approach that leverages the relative geometry among queries and documents learned by the large teacher model. We show that our approach successfully distills from both dual-encoder (DE) and cross-encoder (CE) teacher models to 1/10th size asymmetric students that can retain 95-97% of the teacher performance.
  arXiv  Detail & Related papers  (2023-01-27T22:04:37Z)
• Teaching What You Should Teach: A Data-Based Distillation Method [20.595460553747163]
  We introduce the "Teaching what you Should Teach" strategy into a knowledge distillation framework. We propose a data-based distillation method named "TST" that searches for desirable augmented samples to assist in distilling more efficiently and rationally. To be specific, we design a neural network-based data augmentation module with priori bias, which assists in finding what meets the teacher's strengths but the student's weaknesses.
  arXiv  Detail & Related papers  (2022-12-11T06:22:14Z)
• On the benefits of knowledge distillation for adversarial robustness [53.41196727255314]
  We show that knowledge distillation can be used directly to boost the performance of state-of-the-art models in adversarial robustness. We present Adversarial Knowledge Distillation (AKD), a new framework to improve a model's robust performance.
  arXiv  Detail & Related papers  (2022-03-14T15:02:13Z)
• Fixing the Teacher-Student Knowledge Discrepancy in Distillation [72.4354883997316]
  We propose a novel student-dependent distillation method, knowledge consistent distillation, which makes teacher's knowledge more consistent with the student. Our method is very flexible that can be easily combined with other state-of-the-art approaches.
  arXiv  Detail & Related papers  (2021-03-31T06:52:20Z)
• Learning Student-Friendly Teacher Networks for Knowledge Distillation [50.11640959363315]
  We propose a novel knowledge distillation approach to facilitate the transfer of dark knowledge from a teacher to a student. Contrary to most of the existing methods that rely on effective training of student models given pretrained teachers, we aim to learn the teacher models that are friendly to students.
  arXiv  Detail & Related papers  (2021-02-12T07:00:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.