Gene and Doug, RAG is not the same as "training" an LLM. Retrieval-Augmented Generation (RAG) uses external information retrieval to answer questions. It first retrieves relevant documents, then the language model generates a response using both the retrieved data and its own knowledge. This enhances accuracy without needing to retrain the model.
Apple’s MO is to be ‘proactive’ in reaction to long simmering tensions with developers and/or regulators. Great discussion… especially liked the political questions put to AI models. Shows the relevance of the old adage GIGO - garbage in, garbage out… and value of good supplemental datasets.
It make sense Doug's answer re cahtbot. Grok training data is from X and it may be more lean to the opinion of majority of the people using X. It like raising a child in certain environment and his\her behavior is influenced by parents & people around him\her. 😊
Gene and Doug, RAG is not the same as "training" an LLM. Retrieval-Augmented Generation (RAG) uses external information retrieval to answer questions. It first retrieves relevant documents, then the language model generates a response using both the retrieved data and its own knowledge. This enhances accuracy without needing to retrain the model.
Apple’s MO is to be ‘proactive’ in reaction to long simmering tensions with developers and/or regulators. Great discussion… especially liked the political questions put to AI models. Shows the relevance of the old adage GIGO - garbage in, garbage out… and value of good supplemental datasets.
It make sense Doug's answer re cahtbot. Grok training data is from X and it may be more lean to the opinion of majority of the people using X. It like raising a child in certain environment and his\her behavior is influenced by parents & people around him\her. 😊
Isn’t Gemini based on Claude?
It looks like Gene needs a vacation - take some time on the beach man