A failed or refused page counts with all its tests failed, instead of multiplying the headline by page coverage. From results.parquet, unchanged: the five LLM rows and Unlimited-OCR 2 move on the card; llm_run_summaries.json and the latency and cost charts are omni_parse_bench main at 7a2849f.
22 days ago
Update headline versus latency with Opus and Sol and document timing definitions
Indra OCR: its verdicts in results.parquet, its row in the card's table, and on the parameters and weak-spots charts; shorter captions on the latency, parameters and cost charts
results.parquet: the open-weight models' verdicts; the card: one table and one weak-spots chart for every system, and the open-weight models' headline against parameters
- results.parquet: every parser's verdict on every test; card wording (4a8ab4d4376114da1970768ec9e0076a99fa5a1b)
- results.parquet: every parser's verdict on every test; card wording (part 2) (7ed5b014cf95387b130557c4ac35032d3d5601e7)