SOURCE-LINKED INTELLIGENCE

Collapse, Not Complexity: Failure-Conditioned Decomposition Repair for End-to-End Document Parsing

arXiv · AI, language, vision and robotics · article · Sep 20, 2026 · UTC

End-to-end document parsers increasingly offer an optional reasoning mode for complex pages. On a 180-page entropy-stratified discovery sample with one frozen 4B checkpoint, complexity is the wrong decision variable. Reasoning lowers mean quality by 2.21 Overall at 1.54x tokens; a preregistered input-only model cannot predict its signed benefit (held-out AUROC 0.47, indistinguishable from chance). The benefit concentrates on pages whose ordinary pass has already collapsed, and they do not look complex: shared collapses have lower layout entropy than healthy ones yet consume 19x the tokens as d

Read original source ↗ Open in workspace

recordType

paper

region

Global