Scientific papers require models to jointly reason over text, equations, figures, tables, code, and datasets while preserving the provenance of supporting evidence. Existing benchmarks typically evaluate these capabilities in isolation, leaving it unclear whether multimodal models can support realistic scientific‑reading workflows.
We introduce SciDocBench, a workflow‑centered benchmark for scientific document understanding. It includes 124 expert‑authored, difficulty‑screened questions organized into seven research‑assistant capability groups and 19 subtasks across five scientific domains. Each question is instantiated under four matched conditions—English or Chinese, and all‑images‑first or interleaved document representations—yielding 496 evaluation instances for controlled analysis.
The strongest evaluated system achieves only 62.6/100, with pronounced weaknesses in document perception, evidence grounding, verification, and cross‑document reasoning.
To translate these diagnostics into scalable training signals, we propose SciDocIR, a typed evidence‑graph representation that preserves scientific document objects, layout and cross‑reference relations, and provenance.
Building on SciDocIR, we construct SciDocDataset, containing approximately 15 K supervised fine‑tuning samples and 8 K reinforcement‑learning samples across 14 verifiable subtasks.
Together, SciDocBench, SciDocIR, and SciDocDataset form an evaluation‑to‑training framework for diagnosing and improving scientific‑document assistants. The project page is https://github.com/InternLM/SciDocBench.
Review