Tag
This paper proposes an explainable LLM agent layer placed downstream of an open-world learning pipeline for oil well anomaly detection, using the Qwen3.5-397B-A17B model to provide natural-language justifications and novelty naming on the 3W dataset.