paper-with-me

홈 › Papers

Sinkhorn Distance Minimization for Knowledge Distillation

2024-02-27 · Xiao Cui, Yulei Qin, Yuting Gao, Enwei Zhang, Zihan Xu, Tong Wu, Ke Li, Xing Sun, Wengang Zhou, Houqiang Li

Knowledge distillation (KD) has been widely adopted to compress large language models (LLMs). Existing KD methods investigate various divergence measures including the Kullback-Leibler (KL), reverse Kullback-Leibler (RKL), and Jensen-Shannon (JS) divergences. However, due to limitations inherent in their assumptions and definitions, these measures fail to deliver effective supervision when few distribution overlap exists between the teacher and the student. In this paper, we show that the aforementioned KL, RKL, and JS divergences respectively suffer from issues of mode-averaging, mode-collapsing, and mode-underestimation, which deteriorates logits-based KD for diverse NLP tasks. We propose the Sinkhorn Knowledge Distillation (SinKD) that exploits the Sinkhorn distance to ensure a nuanced and precise assessment of the disparity between teacher and student distributions. Besides, profit by properties of the Sinkhorn metric, we can get rid of sample-wise KD that restricts the perception of divergence in each teacher-student sample pair. Instead, we propose a batch-wise reformulation to capture geometric intricacies of distributions across samples in the high-dimensional space. Comprehensive evaluation on GLUE and SuperGLUE, in terms of comparability, validity, and generalizability, highlights our superiority over state-of-the-art methods on all kinds of LLMs with encoder-only, encoder-decoder, and decoder-only architectures.

📄 PDF Abstract BibTeX arXiv:2402.17110

Code (1)

2018cx/sinkd 공식 구현 pytorch

Tasks

DecoderKnowledge 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

Imitation Learning with Sinkhorn Distances

2020-08-20 · Georgios Papagiannis, Yunpeng Li

Imitation learning algorithms have been interpreted as variants of divergence minimization problems. The ability to compare occupancy measures between experts and learners is crucial in their effectiveness in learning fr…

Imitation LearningMuJoCo

Improving Neural Cross-Lingual Summarization via Employing Optimal Transport Distance for Knowledge Distillation

2021-12-07 · Thong Nguyen, Luu Anh Tuan

Current state-of-the-art cross-lingual summarization models employ multi-task learning paradigm, which works on a shared vocabulary module and relies on the self-attention mechanism to attend among tokens in two language…

Knowledge DistillationMulti-Task Learning

Multi-Level Optimal Transport for Universal Cross-Tokenizer Knowledge Distillation on Language Models

2024-12-19 · Xiao Cui, Mo Zhu, Yulei Qin, Liang Xie 외

Knowledge distillation (KD) has become a prevalent technique for compressing large language models (LLMs). Existing KD methods are constrained by the need for identical tokenizers (i.e., vocabularies) between teacher and…

Knowledge Distillation

Feature-aligned N-BEATS with Sinkhorn divergence

2023-05-24 · Joonhun Lee, Myeongho Jeon, Myungjoo Kang, Kyunghyun Park

We propose Feature-aligned N-BEATS as a domain-generalized time series forecasting model. It is a nontrivial extension of N-BEATS with doubly residual stacking principle (Oreshkin et al. [45]) into a representation learn…

Domain GeneralizationRepresentation LearningTime SeriesTime Series Forecasting+1

Massively scalable Sinkhorn distances via the Nyström method

2018-12-12 · NeurIPS 2019 12 · Jason Altschuler, Francis Bach, Alessandro Rudi, Jonathan Niles-Weed

The Sinkhorn "distance", a variant of the Wasserstein distance with entropic regularization, is an increasingly popular tool in machine learning and statistical inference. However, the time and memory requirements of sta…