Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models

<p dir="ltr">In the era of Large Language Models (LLMs), Knowledge Distillation (KD) enables the transfer of capabilities from proprietary LLMs to open-source models. This survey provides a detailed discussion of the basic principles, algorithms, and implementation methods of knowled...

Full description

Saved in:
Bibliographic Details
Main Author: Dingzong Zhang (23275066) (author)
Other Authors: Devi Listiyani (23275069) (author), Priyanka Singh (256412) (author), Manoranjan Mohanty (23275072) (author)
Published: 2025
Subjects:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1864513522576130048
author Dingzong Zhang (23275066)
author2 Devi Listiyani (23275069)
Priyanka Singh (256412)
Manoranjan Mohanty (23275072)
author2_role author
author
author
author_facet Dingzong Zhang (23275066)
Devi Listiyani (23275069)
Priyanka Singh (256412)
Manoranjan Mohanty (23275072)
author_role author
dc.creator.none.fl_str_mv Dingzong Zhang (23275066)
Devi Listiyani (23275069)
Priyanka Singh (256412)
Manoranjan Mohanty (23275072)
dc.date.none.fl_str_mv 2025-04-04T06:00:00Z
dc.identifier.none.fl_str_mv 10.1109/access.2025.3554586
dc.relation.none.fl_str_mv https://figshare.com/articles/journal_contribution/Distilling_Wisdom_A_Review_on_Optimizing_Learning_From_Massive_Language_Models/31443841
dc.rights.none.fl_str_mv CC BY 4.0
info:eu-repo/semantics/openAccess
dc.subject.none.fl_str_mv Information and computing sciences
Artificial intelligence
Machine learning
Artificial intelligence (AI)
large language model (LLM)
knowledge distillation (KD)
optimization
Transformers
Computational modeling
Surveys
Natural language processing
Predictive models
Technological innovation
Encoding
Context modeling
dc.title.none.fl_str_mv Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
dc.type.none.fl_str_mv Text
Journal contribution
info:eu-repo/semantics/publishedVersion
text
contribution to journal
description <p dir="ltr">In the era of Large Language Models (LLMs), Knowledge Distillation (KD) enables the transfer of capabilities from proprietary LLMs to open-source models. This survey provides a detailed discussion of the basic principles, algorithms, and implementation methods of knowledge distillation. It explores KD’s impact on LLMs, emphasizing its utility in model compression, performance enhancement, and self-improvement. Through the analysis of practical examples such as DistilBERT, TinyBERT, and MobileBERT, the paper demonstrates how knowledge distillation can markedly enhance the efficiency and applicability of large language models in real-world scenarios. The discussion encompasses the varied applications of KD across multiple domains, including industrial systems, embedded systems, Natural Language Processing (NLP), multi-modal processing, and vertical domains, such as medicine, law, science, finance, and materials science. This survey outlines current KD methodologies and future research directions, highlighting its role in advancing AI technologies and fostering innovation across different sectors.</p><h2 dir="ltr">Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2025.3554586" target="_blank">https://dx.doi.org/10.1109/access.2025.3554586</a></p>
eu_rights_str_mv openAccess
id Manara2_decd8fbd7cb40f33be2a07afc7368fe1
identifier_str_mv 10.1109/access.2025.3554586
network_acronym_str Manara2
network_name_str Manara2
oai_identifier_str oai:figshare.com:article/31443841
publishDate 2025
repository.mail.fl_str_mv
repository.name.fl_str_mv
repository_id_str
rights_invalid_str_mv CC BY 4.0
spelling Distilling Wisdom: A Review on Optimizing Learning From Massive Language ModelsDingzong Zhang (23275066)Devi Listiyani (23275069)Priyanka Singh (256412)Manoranjan Mohanty (23275072)Information and computing sciencesArtificial intelligenceMachine learningArtificial intelligence (AI)large language model (LLM)knowledge distillation (KD)optimizationTransformersComputational modelingSurveysNatural language processingPredictive modelsTechnological innovationEncodingContext modeling<p dir="ltr">In the era of Large Language Models (LLMs), Knowledge Distillation (KD) enables the transfer of capabilities from proprietary LLMs to open-source models. This survey provides a detailed discussion of the basic principles, algorithms, and implementation methods of knowledge distillation. It explores KD’s impact on LLMs, emphasizing its utility in model compression, performance enhancement, and self-improvement. Through the analysis of practical examples such as DistilBERT, TinyBERT, and MobileBERT, the paper demonstrates how knowledge distillation can markedly enhance the efficiency and applicability of large language models in real-world scenarios. The discussion encompasses the varied applications of KD across multiple domains, including industrial systems, embedded systems, Natural Language Processing (NLP), multi-modal processing, and vertical domains, such as medicine, law, science, finance, and materials science. This survey outlines current KD methodologies and future research directions, highlighting its role in advancing AI technologies and fostering innovation across different sectors.</p><h2 dir="ltr">Other Information</h2><p dir="ltr">Published in: IEEE Access<br>License: <a href="https://creativecommons.org/licenses/by/4.0/deed.en" target="_blank">https://creativecommons.org/licenses/by/4.0/</a><br>See article on publisher's website: <a href="https://dx.doi.org/10.1109/access.2025.3554586" target="_blank">https://dx.doi.org/10.1109/access.2025.3554586</a></p>2025-04-04T06:00:00ZTextJournal contributioninfo:eu-repo/semantics/publishedVersiontextcontribution to journal10.1109/access.2025.3554586https://figshare.com/articles/journal_contribution/Distilling_Wisdom_A_Review_on_Optimizing_Learning_From_Massive_Language_Models/31443841CC BY 4.0info:eu-repo/semantics/openAccessoai:figshare.com:article/314438412025-04-04T06:00:00Z
spellingShingle Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
Dingzong Zhang (23275066)
Information and computing sciences
Artificial intelligence
Machine learning
Artificial intelligence (AI)
large language model (LLM)
knowledge distillation (KD)
optimization
Transformers
Computational modeling
Surveys
Natural language processing
Predictive models
Technological innovation
Encoding
Context modeling
status_str publishedVersion
title Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
title_full Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
title_fullStr Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
title_full_unstemmed Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
title_short Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
title_sort Distilling Wisdom: A Review on Optimizing Learning From Massive Language Models
topic Information and computing sciences
Artificial intelligence
Machine learning
Artificial intelligence (AI)
large language model (LLM)
knowledge distillation (KD)
optimization
Transformers
Computational modeling
Surveys
Natural language processing
Predictive models
Technological innovation
Encoding
Context modeling