Benchmarks¶

RAGTruth (English)¶

Evaluated on the RAGTruth test set. This benchmark measures how well models detect hallucinations in LLM-generated text across QA, summarization, and data-to-text tasks.

Example-Level Detection¶

Binary classification: does the answer contain any hallucination?

Model	Type	Overall F1
GPT-4	LLM (zero-shot)	63.4%
Luna	Encoder	65.4%
lettucedetect-base-v1	Encoder (149M)	76.8%
Llama-2-13B (fine-tuned)	LLM	78.7%
lettucedetect-large-v1	Encoder (395M)	79.2%
RAG-HAT (Llama-3-8B)	LLM	83.9%

Span-Level Detection¶

LettuceDetect achieves state-of-the-art span-level results among models that report this metric, outperforming fine-tuned Llama-2-13B. Span-level evaluation measures how precisely the model can locate the exact hallucinated text within an answer.

What These Numbers Mean¶

Example F1 — Can the model tell if an answer has any hallucination? Higher is better.
Span F1 — Can the model point to exactly which parts are hallucinated? This is the harder task and where LettuceDetect excels relative to its size.

LettuceDetect models are 50-500x smaller than LLM-based detectors while achieving competitive or better accuracy.

Citation¶

@misc{Kovacs:2025,
      title={LettuceDetect: A Hallucination Detection Framework for RAG Applications},
      author={Ádám Kovács and Gábor Recski},
      year={2025},
      eprint={2502.17125},
      archivePrefix={arXiv},
      primaryClass={cs.CL},
      url={https://arxiv.org/abs/2502.17125},
}