Settings

Configure models, thresholds, and manage the alias cache

Models

Embedding Model

LLM Model (Ollama)

Thresholds

Fuzzy Accept

85%

Embedding Accept

0.65

Header Min Score

0.35

Row Accept Score

0.60

Verdict Accept Min

0.80

Verdict Review Min

0.50

Alias Cache

The alias cache stores learned column mappings for Tier-1 instant lookups.

Cache Entries

Backend offline — start the API to inspect the cache.