Overview
Stanford’s framework for programming language model pipelines with structured prompts, signatures, and optimization.
Details
DSPy is Stanford’s framework for programming language model pipelines, replacing manual prompt engineering with structured signatures, modules, and optimizers that improve prompts automatically. DSPy enables systematic optimization of LLM pipelines for accuracy and reliability. DSPy represents a programming paradigm for LLM development distinct from typical agent frameworks.
Tags
agent-framework, open-source, prompt-optimization, research, stanford