Per-table card variant · Experimental
Table-only card
Just the Markdown table + 1-line caption. No frontmatter, no metadata.
In one paragraph
The most stripped-down card: a 1-line caption and the rendered Markdown table. <500 tokens typical. Stress test of whether structured table data alone is enough.
How the inputs are generated
Generation · 01Generator script
evaluation_runs/generate_card_variants.py:render_table_onlyInput sources
- • pipeline-v0.6.1 cards
- • Docling-extracted Markdown tables
AI use
No — pure deterministic transformation
OCR / re-OCR
Inherits from the upstream pipeline variant
Approximate processing time
~2 seconds for all 407 cards.
Resource intensity
Low — CPU-only post-processing, runs in seconds
Determinism
Deterministic (same input → same output, byte-identical)
Output location
card_sets/pipeline-v0.7-table-only/Cards produced
407 cards
Introduced
v0.7 variant family, 2026-05-22.
Evaluation results
Diagnostic · 02Avg open-tier pass rate
52%
Typical card size
~1 KB per card (median ~600 bytes)
Evaluation cycle
Cycle 16
Relative to v0.6.1 baseline
+25 percentage points open-tier versus v0.6.1 baseline