Meaning
A vector that captures the meaning and relationships of words in a continuous space for LLMs.
Definition
Embeddings are vectors that represent words or phrases in a continuous space, capturing their meanings and relationships based on the context in which they appear.
These numerical representations allow LLMs to understand the semantic connections between words, improving their ability to perform tasks like analogy completion or text classification.
Example
The words “king” and “queen” might have similar embeddings in an LLM, reflecting their related meanings and roles in language, which the model can use to generate contextually appropriate text.
