@inproceedings{sengupta-etal-2026-pruning,
title = "Pruning Laws for Large Language Models",
author = "Sengupta, Ayan and
Chaudhary, Siddhant and
Chakraborty, Tanmoy",
editor = "Lee, Hung-Yi and
Dev, Sunipa and
Li, Jessy and
Elliott, Desmond",
booktitle = "Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing",
month = oct,
year = "2026",
address = "Budapest, Hungary",
publisher = "Association for Computational Linguistics",
url = "https://aclanthology.org/2026.emnlp-main.2367/",
pages = "53725--53744",
ISBN = "979-8-89176-459-0",
abstract = "Scaling up model parameters and training data consistently improves the performance of large language models (LLMs), but at the cost of rapidly growing memory and compute requirements, which makes deployment on resource-limited hardware infeasible. Model pruning, a widely used compression technique, reduces inference costs by removing redundant parameters. However, its impact on downstream performance remains unpredictable and is typically assessed only through costly empirical sweeps. To address this gap, we introduce pruning laws {--} simple and interpretable scaling relations that connect a pruned LLM{'}s post-pruning performance to its unpruned performance and pruning ratio. Across ten LLMs (1.3B{--}30B parameters), three pruning strategies (unstructured, width, and depth), and eight diverse tasks, we show that pruning laws achieve strong predictive accuracy (average extrapolation error {\ensuremath{<}}7{\%}), reliably quantify performance degradation, and identify critical pruning thresholds beyond which recovery is infeasible. Moreover, we demonstrate that the same laws transfer universally across architectures, pruning methods, and even unseen models in zero-shot and one-shot setups. These results provide both researchers and practitioners with a principled framework to select pruning strategies, estimate safe pruning ratios without exhaustive tuning, and deploy LLMs efficiently under real-world compute and latency constraints."
}<?xml version="1.0" encoding="UTF-8"?>
<modsCollection xmlns="http://www.loc.gov/mods/v3">
<mods ID="sengupta-etal-2026-pruning">
<titleInfo>
<title>Pruning Laws for Large Language Models</title>
</titleInfo>
<name type="personal">
<namePart type="given">Ayan</namePart>
<namePart type="family">Sengupta</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Siddhant</namePart>
<namePart type="family">Chaudhary</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Tanmoy</namePart>
<namePart type="family">Chakraborty</namePart>
<role>
<roleTerm authority="marcrelator" type="text">author</roleTerm>
</role>
</name>
<originInfo>
<dateIssued>2026-10</dateIssued>
</originInfo>
<typeOfResource>text</typeOfResource>
<relatedItem type="host">
<titleInfo>
<title>Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing</title>
</titleInfo>
<name type="personal">
<namePart type="given">Hung-Yi</namePart>
<namePart type="family">Lee</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Sunipa</namePart>
<namePart type="family">Dev</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Jessy</namePart>
<namePart type="family">Li</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<name type="personal">
<namePart type="given">Desmond</namePart>
<namePart type="family">Elliott</namePart>
<role>
<roleTerm authority="marcrelator" type="text">editor</roleTerm>
</role>
</name>
<originInfo>
<publisher>Association for Computational Linguistics</publisher>
<place>
<placeTerm type="text">Budapest, Hungary</placeTerm>
</place>
</originInfo>
<genre authority="marcgt">conference publication</genre>
<identifier type="isbn">979-8-89176-459-0</identifier>
</relatedItem>
<abstract>Scaling up model parameters and training data consistently improves the performance of large language models (LLMs), but at the cost of rapidly growing memory and compute requirements, which makes deployment on resource-limited hardware infeasible. Model pruning, a widely used compression technique, reduces inference costs by removing redundant parameters. However, its impact on downstream performance remains unpredictable and is typically assessed only through costly empirical sweeps. To address this gap, we introduce pruning laws – simple and interpretable scaling relations that connect a pruned LLM’s post-pruning performance to its unpruned performance and pruning ratio. Across ten LLMs (1.3B–30B parameters), three pruning strategies (unstructured, width, and depth), and eight diverse tasks, we show that pruning laws achieve strong predictive accuracy (average extrapolation error \ensuremath<7%), reliably quantify performance degradation, and identify critical pruning thresholds beyond which recovery is infeasible. Moreover, we demonstrate that the same laws transfer universally across architectures, pruning methods, and even unseen models in zero-shot and one-shot setups. These results provide both researchers and practitioners with a principled framework to select pruning strategies, estimate safe pruning ratios without exhaustive tuning, and deploy LLMs efficiently under real-world compute and latency constraints.</abstract>
<identifier type="citekey">sengupta-etal-2026-pruning</identifier>
<location>
<url>https://aclanthology.org/2026.emnlp-main.2367/</url>
</location>
<part>
<date>2026-10</date>
<extent unit="page">
<start>53725</start>
<end>53744</end>
</extent>
</part>
</mods>
</modsCollection>
%0 Conference Proceedings
%T Pruning Laws for Large Language Models
%A Sengupta, Ayan
%A Chaudhary, Siddhant
%A Chakraborty, Tanmoy
%Y Lee, Hung-Yi
%Y Dev, Sunipa
%Y Li, Jessy
%Y Elliott, Desmond
%S Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing
%D 2026
%8 October
%I Association for Computational Linguistics
%C Budapest, Hungary
%@ 979-8-89176-459-0
%F sengupta-etal-2026-pruning
%X Scaling up model parameters and training data consistently improves the performance of large language models (LLMs), but at the cost of rapidly growing memory and compute requirements, which makes deployment on resource-limited hardware infeasible. Model pruning, a widely used compression technique, reduces inference costs by removing redundant parameters. However, its impact on downstream performance remains unpredictable and is typically assessed only through costly empirical sweeps. To address this gap, we introduce pruning laws – simple and interpretable scaling relations that connect a pruned LLM’s post-pruning performance to its unpruned performance and pruning ratio. Across ten LLMs (1.3B–30B parameters), three pruning strategies (unstructured, width, and depth), and eight diverse tasks, we show that pruning laws achieve strong predictive accuracy (average extrapolation error \ensuremath<7%), reliably quantify performance degradation, and identify critical pruning thresholds beyond which recovery is infeasible. Moreover, we demonstrate that the same laws transfer universally across architectures, pruning methods, and even unseen models in zero-shot and one-shot setups. These results provide both researchers and practitioners with a principled framework to select pruning strategies, estimate safe pruning ratios without exhaustive tuning, and deploy LLMs efficiently under real-world compute and latency constraints.
%U https://aclanthology.org/2026.emnlp-main.2367/
%P 53725-53744
Markdown (Informal)
[Pruning Laws for Large Language Models](https://aclanthology.org/2026.emnlp-main.2367/) (Sengupta et al., EMNLP 2026)
ACL
- Ayan Sengupta, Siddhant Chaudhary, and Tanmoy Chakraborty. 2026. Pruning Laws for Large Language Models. In Proceedings of the 2026 Conference on Empirical Methods in Natural Language Processing, pages 53725–53744, Budapest, Hungary. Association for Computational Linguistics.