Training Data
Data used to train or fine-tune a machine learning model -- distinct from data a model retrieves at answer time.
Training data is the dataset a machine learning model learns from during training or fine-tuning -- it shapes the model's weights permanently. This is different from data used at inference time (like a RAG-ready knowledge base), which the model retrieves and reads on the fly without changing its underlying weights.
Example
A labeled dataset of search queries and intent categories used to fine-tune a classifier is training data. The same dataset retrieved live to answer a single question is being used as RAG context, not training data.