Overview
DSPy represents LLM applications as modular programs with typed signatures and predictors, then provides optimizers that can tune prompts or parameters against examples and metrics. It can be used for extraction, classification, RAG, and agent workflows.
Key features
- Composable LLM modules
- Typed signatures and predictors
- Prompt and parameter optimizers
- Evaluation-driven iteration
Requirements, installation and quick start
Install the Python package with `pip install dspy`; configure a supported language-model provider as described in the docs.
Usage
Define task signatures and modules, supply a dataset and metric, then evaluate or optimize the pipeline for the target task.
Model compatibility and use cases
Works with multiple providers and local model backends; consult current docs for the integration matrix.
License and risk notes
The repository declares the MIT License. Connected model providers and model weights are licensed separately.
Release and maintenance
See the official repository for current releases and changes.