← All Recipes
Bazantic // Recipe
certscroll-blind-scenario-eval
certscroll-blind-scenario-eval
Executes a zero-leakage blind evaluation on syllabus scenario questions. Answers are masked during retrieval; candidate models articulate chain-of-thought and grade choices against official leaf citations.
Recipe inputs
Complete fields before a run.
Number of scenario questions to test in the batch.
Certification track identifier (e.g. 'aws-mla-c01', 'anthropic-claude-certified-architect', 'nvidia-nca-genl').
Cognitive difficulty tier to evaluate.