Efficient Multi-Task Inference with a Shared Backbone and Lightweight Task-Specific Adapters for Automatic Scoring

Ehsan Latif, Yifan Zhou, Luyan Fang, Xiaoming Zhai
Proceedings of the Innovation and Responsibility in AI-Supported Education Workshop, PMLR 273:212-220, 2025.

Abstract

The integration of Artificial Intelligence (AI) in education requires scalable and efficient frameworks that balance performance, adaptability, and cost. This paper addresses these needs by proposing a shared backbone model architecture enhanced with lightweight LoRA adapters for task-specific fine-tuning, targeting the automated scoring of student responses across 27 mutually exclusive tasks. By achieving competitive performance (average QWK of 0.848 compared to 0.888 for fully fine-tuned models) while reducing GPU memory consumption by 60% and inference latency by 40%, the framework demonstrates significant efficiency gains. This approach aligns with the workshop’s focus on improving language models for educational tasks, creating responsible innovations for cost-sensitive deployment, and supporting educators by streamlining assessment workflows. The findings underscore the potential of scalable AI to enhance learning outcomes while maintaining fairness and transparency in automated scoring systems.

Cite this Paper

BibTeX
@InProceedings{pmlr-v273-latif25a, title = {Efficient Multi-Task Inference with a Shared Backbone and Lightweight Task-Specific Adapters for Automatic Scoring}, author = {Latif, Ehsan and Zhou, Yifan and Fang, Luyan and Zhai, Xiaoming}, booktitle = {Proceedings of the Innovation and Responsibility in AI-Supported Education Workshop}, pages = {212--220}, year = {2025}, editor = {Wang, Zichao and Woodhead, Simon and Ananda, Muktha and Mallick, Debshila Basu and Sharpnack, James and Burstein, Jill}, volume = {273}, series = {Proceedings of Machine Learning Research}, month = {03 Mar}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v273/main/assets/latif25a/latif25a.pdf}, url = {https://proceedings.mlr.press/v273/latif25a.html}, abstract = {The integration of Artificial Intelligence (AI) in education requires scalable and efficient frameworks that balance performance, adaptability, and cost. This paper addresses these needs by proposing a shared backbone model architecture enhanced with lightweight LoRA adapters for task-specific fine-tuning, targeting the automated scoring of student responses across 27 mutually exclusive tasks. By achieving competitive performance (average QWK of 0.848 compared to 0.888 for fully fine-tuned models) while reducing GPU memory consumption by 60% and inference latency by 40%, the framework demonstrates significant efficiency gains. This approach aligns with the workshop’s focus on improving language models for educational tasks, creating responsible innovations for cost-sensitive deployment, and supporting educators by streamlining assessment workflows. The findings underscore the potential of scalable AI to enhance learning outcomes while maintaining fairness and transparency in automated scoring systems.} }
Endnote
%0 Conference Paper %T Efficient Multi-Task Inference with a Shared Backbone and Lightweight Task-Specific Adapters for Automatic Scoring %A Ehsan Latif %A Yifan Zhou %A Luyan Fang %A Xiaoming Zhai %B Proceedings of the Innovation and Responsibility in AI-Supported Education Workshop %C Proceedings of Machine Learning Research %D 2025 %E Zichao Wang %E Simon Woodhead %E Muktha Ananda %E Debshila Basu Mallick %E James Sharpnack %E Jill Burstein %F pmlr-v273-latif25a %I PMLR %P 212--220 %U https://proceedings.mlr.press/v273/latif25a.html %V 273 %X The integration of Artificial Intelligence (AI) in education requires scalable and efficient frameworks that balance performance, adaptability, and cost. This paper addresses these needs by proposing a shared backbone model architecture enhanced with lightweight LoRA adapters for task-specific fine-tuning, targeting the automated scoring of student responses across 27 mutually exclusive tasks. By achieving competitive performance (average QWK of 0.848 compared to 0.888 for fully fine-tuned models) while reducing GPU memory consumption by 60% and inference latency by 40%, the framework demonstrates significant efficiency gains. This approach aligns with the workshop’s focus on improving language models for educational tasks, creating responsible innovations for cost-sensitive deployment, and supporting educators by streamlining assessment workflows. The findings underscore the potential of scalable AI to enhance learning outcomes while maintaining fairness and transparency in automated scoring systems.
APA
Latif, E., Zhou, Y., Fang, L. & Zhai, X.. (2025). Efficient Multi-Task Inference with a Shared Backbone and Lightweight Task-Specific Adapters for Automatic Scoring. Proceedings of the Innovation and Responsibility in AI-Supported Education Workshop, in Proceedings of Machine Learning Research 273:212-220 Available from https://proceedings.mlr.press/v273/latif25a.html.

Related Material