Chain of Evidence: Pixel-Level Visual Attribution for Iterative Retrieval-Augmented Generation

Iterative Retrieval-Augmented Generation (iRAG) has emerged as a powerful paradigm for answering complex multi-hop questions by progressively retrieving and reasoning over external documents. However, current systems predominantly operate on parsed text, which creates two critical bottlenecks: (1) $\mathit{\text{Coarse-grained}}\mathit{\text{attribution}}$, where users are burdened with manually locating evidence within lengthy documents based on vague text-level citations; and (2) $\mathit{\text{Visual}}\mathit{\text{semantic}}\mathit{\text{loss}}$, where the conversion of visually rich documents (e.g., slides, PDFs with charts) into text discards spatial logic and layout cues essential for reasoning. To bridge this gap, we present $\mathbf{\text{Chain}}\mathbf{\text{of}}\mathbf{\text{Evidence}}\mathbf{\text{(CoE)}}$, a retriever-agnostic visual attribution framework that leverages Vision-Language Models to reason directly over screenshots of retrieved document candidates. CoE eliminates format-specific parsing and outputs precise bounding boxes, visualizing the complete reasoning chain within the retrieved candidate set. We evaluate CoE on two distinct benchmarks: $\mathbf{\text{Wiki-CoE}}$, a large-scale dataset of structured web pages derived from 2WikiMultiHopQA, and $\mathbf{\text{SlideVQA}}$, a challenging dataset of presentation slides featuring complex diagrams and free-form layouts. Experiments demonstrate that fine-tuned Qwen3-VL-8B-Instruct achieves robust performance, significantly outperforming text-based baselines in scenarios requiring visual layout understanding, while establishing a retriever-agnostic solution for pixel-level interpretable iRAG. Our code is available at https://github.com/PeiYangLiu/CoE.git.