Conceptself-consistency / 自一致性检验

Self-Consistency Checking

A method for verifying the reliability of LLM outputs by sampling the same question multiple times and comparing the consistency of the resulting answers. Greater divergence among answers indicates lower reliability.

Timeline (last 90 days)

Sep 12

Self-consistency checking, by having the model sample the same question multiple times and comparing answer stability, consumes only inference time without increasing peak VRAM usage.

Unverified50%

All Facts (1)

Unverified

Self-consistency checking, by having the model sample the same question multiple times and comparing answer stability, consumes only inference time without increasing peak VRAM usage.

50%

Source Articles

Hallucination Detection for Local LLMs in Practice: Low-VRAM Approaches from 1.5B to 120BSep 12