DistilDoc: Knowledge Distillation for Visually-Rich Document Applications

TL;DR

This paper investigates knowledge distillation techniques to create efficient, high-performing models for visually-rich document understanding tasks, emphasizing the importance of model compression and robustness in downstream applications.

Contribution

It introduces a KD experimentation methodology for document understanding, compares various strategies across architectures, and evaluates their robustness in downstream tasks like DocVQA.

Findings

Some KD methods outperform supervised training in knowledge transfer.

Distilled models show a significant knowledge gap affecting downstream robustness.

Robustness in downstream tasks does not directly correlate with knowledge gap size.

Abstract

This work explores knowledge distillation (KD) for visually-rich document (VRD) applications such as document layout analysis (DLA) and document image classification (DIC). While VRD research is dependent on increasingly sophisticated and cumbersome models, the field has neglected to study efficiency via model compression. Here, we design a KD experimentation methodology for more lean, performant models on document understanding (DU) tasks that are integral within larger task pipelines. We carefully selected KD strategies (response-based, feature-based) for distilling knowledge to and from backbones with different architectures (ResNet, ViT, DiT) and capacities (base, small, tiny). We study what affects the teacher-student knowledge gap and find that some methods (tuned vanilla KD, MSE, SimKD with an apt projector) can consistently outperform supervised student training. Furthermore, we design downstream task setups to evaluate covariate shift and the robustness of distilled DLA models on zero-shot layout-aware document visual question answering (DocVQA). DLA-KD experiments result in a large mAP knowledge gap, which unpredictably translates to downstream robustness, accentuating the need to further explore how to efficiently obtain more semantic document layout awareness.