surfaceclassificationcredibility_llm
Classificationdimension · returns varchar

CREDIBILITY_LLM

LLM-backed credibility assessment (escape hatch for CREDIBILITY)

Per-row classifier — stable across GROUP BY.

classificationllmllm-escape-hatchtext

Arguments

nametypedescription
textVARCHAR
strictnessVARCHAR
num_levelsINTEGER

About

LLM-backed escape hatch for credibility assessment. Use when CREDIBILITY (zero-shot NLI) can't apply nuanced rules — for example, domain-specific strictness criteria or assessments that require weighing multiple conflicting signals. For routine credibility bucketing, prefer CREDIBILITY — batch-NLI is much faster and handles most cases.

Examples

LLM escape hatch assesses credibility

SELECT
  credibility_llm ('Peer-reviewed study shows...')

Nearby rabbit holes

same domain
Climb back to The Looking Glass