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

TL;DR

This paper introduces an efficient multi-task AI framework for automated student response scoring, using shared models and lightweight adapters to reduce resource use while maintaining high accuracy.

Contribution

It proposes a novel shared backbone architecture with lightweight adapters for multi-task scoring, significantly reducing computational costs compared to fully fine-tuned models.

Findings

Achieves an average QWK of 0.848, close to fully fine-tuned models' 0.888.

Reduces GPU memory consumption by 60%.

Lowers inference latency by 40%.

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.