DocXEval benchmarks explainability methods for multimodal document models
Occlusion ranked most faithful and interpretable across 13 methods, while some metrics were unreliable.
Stories per day · 30 days
2 stories, newest first · one card per story
Stories tagged #document-understanding
Retrieval heads stay sparse and shared across MLLM modalities
Ranking training on those heads raises Qwen3-VL-8B page Recall@1 on MMDocIR by 6.03 points.