Confidence

Overview

confidence is a looper algorithm that escalates across candidate models until confidence is high enough.

It aligns to config/algorithm/looper/confidence.yaml.

Key Advantages

• Supports small-to-large escalation instead of a fixed winner.
• Makes stopping conditions explicit.
• Lets one route trade extra latency for higher confidence only when needed.

What Problem Does It Solve?

Some routes should try a cheaper candidate first and only escalate when the answer is not confident enough. confidence gives that escalation policy a dedicated algorithm instead of embedding it in application code.

When to Use

• a route should escalate across several candidate models
• confidence should decide whether to continue
• the route should stop as soon as one response is good enough

Configuration

Use this fragment inside routing.decisions[].algorithm:

algorithm:
  type: confidence
  confidence:
    confidence_method: hybrid
    threshold: 0.72
    escalation_order: small_to_large
    on_error: skip