Member-only story
The model, positioned at the foundational layer of the LangChain framework, serves as the pivotal element in applications built on language models. Essentially, developing LangChain applications involves utilizing LangChain as a framework to solve specific problems by calling large models through APIs.
It’s fair to say that the entire logic of the LangChain framework is powered by the LLM (Large Language Model) engine. Without the model, the LangChain framework would lose its purpose. In this article, we’ll delve into the details of the model.
LangChain Model I/O
The process of utilizing the model can be broken down into three distinct stages: input prompts (corresponding to ‘Format’ in the diagram), invoking the model (‘Predict’), and interpreting the output (‘Parse’). These three components work together as a cohesive unit, and within the LangChain framework, this entire process is collectively referred to as Model I/O (Input/Output). Like the following diagram shows:
