ParentFull threadbugglebeetle·Many machine learning tasks (e.g. NER) are far less efficient if one uses an LLM compared to, say, BERT models, so I’m not sure this is the case.View on HN