BETA A Hallucination Firewall: Real-Time Detection and Correction of LLM Hallucinations using Sparse Autoencoder Interpretability and Semantic Entropy
