RAGulator: Lightweight Out-of-Context Detectors for Grounded Text Generation
This addresses the need for safe enterprise adoption of RAG by providing a fast, resource-efficient solution for real-time out-of-context detection.
The paper tackles the problem of detecting when LLM-generated text is semantically out-of-context from retrieved documents in RAG applications, finding that a DeBERTa-based lightweight model achieves the best performance without requiring additional preprocessing or feature engineering.
Real-time detection of out-of-context LLM outputs is crucial for enterprises looking to safely adopt RAG applications. In this work, we train lightweight models to discriminate LLM-generated text that is semantically out-of-context from retrieved text documents. We preprocess a combination of summarisation and semantic textual similarity datasets to construct training data using minimal resources. We find that DeBERTa is not only the best-performing model under this pipeline, but it is also fast and does not require additional text preprocessing or feature engineering. While emerging work demonstrates that generative LLMs can also be fine-tuned and used in complex data pipelines to achieve state-of-the-art performance, we note that speed and resource limits are important considerations for on-premise deployment.