Iterative Layer Pruning for Efficient Translation Inference

Yasmin Moslem, Muhammad Hazim Al Farouq, John D. Kelleher
10/26/2025
cs.CLcs.PF

No official code found. Best implementation: ymoslem/Model-Compression (1 star, last commit 2025-12-11).

Abstract

Large language models (LLMs) have transformed many areas of natural language processing, including machine translation. However, efficient deployment of LLMs remains challenging due to their intensive computational requirements. In this paper, we address this challenge and present our submissions to the Model Compression track at the Conference on Machine Translation (WMT 2025). In our experiments, we investigate iterative layer pruning guided by layer importance analysis. We evaluate this method using the Aya-Expanse-8B model for translation from Czech to German, and from English to Egyptian Arabic. Our approach achieves substantial reductions in model size and inference time, while maintaining the translation quality of the baseline models.

View on arXivView PDF

Code Implementations(1)

ymoslem/Model-CompressionCommunity67%

Code for the papers: "Efficient Speech Translation through Model Compression and Knowledge Distillation" and "Iterative Layer Pruning for Efficient Translation Inference"

10Python, Jupyter NotebookMay 23, 202510 months ago
layer-pruningmachine-translationmodel-compressionspeech-translation

Similar Papers

CULL-MT: Compression Using Language and Layer pruning for Machine Translation

87% match2024

Efficient Speech Translation through Model Compression and Knowledge Distillation

85% match1 repo12025

Adaptive Pruning for Large Language Models with Structural Importance Awareness

83% match2024

Frustratingly Easy Task-aware Pruning for Large Language Models

82% match2025

Streamlining Redundant Layers to Compress Large Language Models

82% match2 repos2702024

Iterative Layer-wise Distillation for Efficient Compression of Large Language Models

82% match2025

Cite this paper

@article{moslem2025iterative,
  title  = {Iterative Layer Pruning for Efficient Translation Inference},
  author = {Yasmin Moslem and Muhammad Hazim Al Farouq and John D. Kelleher},
  year   = {2025},
  eprint = {2510.22763},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2510.22763v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2510.22763.svg)](https://codeofpaper.com/paper/2510.22763)

Discussion