Evaluation mode · Experimental
M2c — Docling Markdown mode
Model receives Docling's full linearized Markdown for the entire document.
In one paragraph
Instead of one targeted card, the model gets `docling.md` — Docling's full linearized text version of the whole document. NOAA = ~260 KB; V27 = 4.1 MB; V35 = 2.8 MB. Tests whether the linearized full-document format works when the model has sufficient context. Frontier-tier reference handles V27/V35 fine; smaller open models overflow on the larger documents.
How the inputs are generated
Generation · 01Generator script
Docling library — produced during the initial pipeline conversionInput sources
- • Docling docling.md export (whole-document linearization)
AI use
No — pure deterministic transformation
OCR / re-OCR
Inherits from Docling's extraction step
Tool:
Docling --force-reocrApproximate processing time
Docling conversion: ~25 min for V27 (with OCR); model inference: ~10 seconds per cell when prompt fits.
Resource intensity
High — Docling extraction with OCR, multi-minute
Determinism
Deterministic (same input → same output, byte-identical)
Introduced
Cycle 8, 2026-05-21.
Related variants
Cross-reference · 06- Evaluation modeM3-L4 — Oracle retrieval modeModel receives exactly one pre-selected card per question. Isolates 'can the model answer given perfect retrieval?'
- Evaluation modeM2a — Raw Docling JSON modeModel receives the raw decompressed docling.json.gz. Demonstrates why specialized evidence packaging is needed at all.