Dspy on AI Knowledge Base

Adapters

Mon, 01 Jan 0001 00:00:00 +0000

Advanced Tool Use

Mon, 01 Jan 0001 00:00:00 +0000

Async

Mon, 01 Jan 0001 00:00:00 +0000

Audio

Mon, 01 Jan 0001 00:00:00 +0000

Building a Creative Text-Based AI Game

Mon, 01 Jan 0001 00:00:00 +0000

Building AI Agents with DSPy

Mon, 01 Jan 0001 00:00:00 +0000

Building AI Applications by Customizing DSPy Modules

Mon, 01 Jan 0001 00:00:00 +0000

Building RAG as Agent

Mon, 01 Jan 0001 00:00:00 +0000

Cache

Mon, 01 Jan 0001 00:00:00 +0000

Cheatsheet

Mon, 01 Jan 0001 00:00:00 +0000

DSPy cheatsheet for quick reference

Classification

Mon, 01 Jan 0001 00:00:00 +0000

Classification Finetuning

Mon, 01 Jan 0001 00:00:00 +0000

Code Generation for Unfamiliar Libraries

Mon, 01 Jan 0001 00:00:00 +0000

Community Ports

Mon, 01 Jan 0001 00:00:00 +0000

Community Resources

Mon, 01 Jan 0001 00:00:00 +0000

Contributing

Mon, 01 Jan 0001 00:00:00 +0000

Data Handling

Mon, 01 Jan 0001 00:00:00 +0000

Debugging & Observability

Mon, 01 Jan 0001 00:00:00 +0000

Deployment

Mon, 01 Jan 0001 00:00:00 +0000

DSPy Assertions

Mon, 01 Jan 0001 00:00:00 +0000

Email Information Extraction

Mon, 01 Jan 0001 00:00:00 +0000

Entity Extraction

Mon, 01 Jan 0001 00:00:00 +0000

Evaluation Overview

Mon, 01 Jan 0001 00:00:00 +0000

This is the foundational concept page for DSPy’s evaluate-then-optimize workflow, and it is essential reading before you touch any teleprompter. The key insight is that DSPy treats evaluation as a first-class input to compilation rather than an afterthought — your dev set and metric become the signal the optimizer uses to rewrite prompts. Start here, then move to the metrics page to define what good actually means for your task. Watch out for evaluating on the same examples you optimize against, which inflates scores and hides overfitting.

FAQ

Mon, 01 Jan 0001 00:00:00 +0000

Common questions on DSPy usage, comparisons, optimization, and debugging

Financial Analysis with Yahoo Finance

Mon, 01 Jan 0001 00:00:00 +0000

Finetuning Agents

Mon, 01 Jan 0001 00:00:00 +0000

Generating llms.txt

Mon, 01 Jan 0001 00:00:00 +0000

GEPA for AIME (Math)

Mon, 01 Jan 0001 00:00:00 +0000

GEPA for Code Backdoor Classification (AI control)

Mon, 01 Jan 0001 00:00:00 +0000

GEPA for Privacy-Conscious Delegation

Mon, 01 Jan 0001 00:00:00 +0000

GEPA for Structured Information Extraction for Enterprise Tasks

Mon, 01 Jan 0001 00:00:00 +0000

Get Started

Mon, 01 Jan 0001 00:00:00 +0000

DSPy overview and quick start guide

Image Generation Prompt iteration

Mon, 01 Jan 0001 00:00:00 +0000

Language Models

Mon, 01 Jan 0001 00:00:00 +0000

Learning DSPy

Mon, 01 Jan 0001 00:00:00 +0000

Three stages of building AI systems - programming, evaluation, and optimization

Managing Conversation History

Mon, 01 Jan 0001 00:00:00 +0000

Math Reasoning

Mon, 01 Jan 0001 00:00:00 +0000

MCP

Mon, 01 Jan 0001 00:00:00 +0000

Memory-Enabled ReAct Agents

Mon, 01 Jan 0001 00:00:00 +0000

Metrics

Mon, 01 Jan 0001 00:00:00 +0000

In DSPy a metric is the objective function that drives both evaluation and optimization, so this page matters more than a typical reference — your metric definition directly shapes how teleprompters compile and improve a program. Pay close attention to the difference between simple answer-matching metrics and metrics that themselves call an LM to judge quality, since the latter adds cost and variance you have to control. A common pitfall is returning a bare boolean where an optimizer expects a float score. Read the evaluation overview first, then pair this with the optimizers documentation.

Modules

Mon, 01 Jan 0001 00:00:00 +0000

Multi-Hop RAG

Mon, 01 Jan 0001 00:00:00 +0000

Optimization Overview

Mon, 01 Jan 0001 00:00:00 +0000

Optimizers

Mon, 01 Jan 0001 00:00:00 +0000

Output Refinement

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Overview

Mon, 01 Jan 0001 00:00:00 +0000

Privacy-Conscious Delegation

Mon, 01 Jan 0001 00:00:00 +0000

Program Of Thought

Mon, 01 Jan 0001 00:00:00 +0000

Programming Overview

Mon, 01 Jan 0001 00:00:00 +0000

DSPy takes a fundamentally different approach to working with LLMs: instead of manually writing and tweaking prompts, you define typed signatures (input/output schemas) and let DSPy’s optimizers compile them into effective prompts automatically. This is a paradigm shift from traditional prompt engineering, and it means the skills you build here are more about program design than wordsmithing. Focus on understanding Modules and Signatures first, as they are the core abstractions everything else builds on. A common gotcha is expecting DSPy to work well without a good set of training examples — the optimizers need representative input-output pairs to produce high-quality compiled prompts, so invest in your dataset before tuning the optimizer.

Retrieval-Augmented Generation (RAG)

Mon, 01 Jan 0001 00:00:00 +0000

RL for Multi-Hop Research

Mon, 01 Jan 0001 00:00:00 +0000

RL for Privacy-Conscious Delegation

Mon, 01 Jan 0001 00:00:00 +0000

Saving and Loading

Mon, 01 Jan 0001 00:00:00 +0000

Signatures

Mon, 01 Jan 0001 00:00:00 +0000

Streaming

Mon, 01 Jan 0001 00:00:00 +0000

Tools

Mon, 01 Jan 0001 00:00:00 +0000

Tracking DSPy Optimizers

Mon, 01 Jan 0001 00:00:00 +0000

Tutorials Overview

Mon, 01 Jan 0001 00:00:00 +0000

Use Cases

Mon, 01 Jan 0001 00:00:00 +0000

Use MCP in DSPy

Mon, 01 Jan 0001 00:00:00 +0000