项目描述
LettuceDetect is a hallucination detection framework for Retrieval-Augmented Generation (RAG) applications, identifying unsupported parts of an answer by comparing it to the provided context. It offers token-level precision, optimized inference, long context window support, multilingual capabilities, and is MIT-licensed.
Project Information
Created on 2/5/2025
Updated on 6/30/2025
Categories
ai-content-generation
machine-learning-framework
text-processing
Tags
algorithm-model
data-processing
open-source-community
model-deployment
multimodal
Topics
python
information-extraction
nlp
hallucination-evaluation
bert
hallucination-detection
token-classification
pytorch