Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents Mike Taylor

(ebook) (audiobook) (audiobook)

Nowość
Promocja Przejdź

Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents Mike Taylor - okladka książki

Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents Mike Taylor - audiobook MP3

Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents Mike Taylor - audiobook CD

Autor:: Mike Taylor
Wydawnictwo:: O'Reilly Media (Z chęcią przeczytam książkę w języku polskim)
Ocena:: Bądź pierwszym, który oceni tę książkę
Stron:: 346
Dostępne formaty:: ePub

Mobi

Ebook

245,65 zł ~~279,00 zł~~ (-12%)

228,65 zł najniższa cena z 30 dni

Dodaj do koszyka Dostępny natychmiast po opłaceniu zakupu lub Kup na prezent Kup 1-kliknięciem

Przenieś na półkę

Do przechowalni

Kup w zestawie z dodatkowym rabatem

Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents Mike Taylor

Prompt Engineering for Generative AI James Phoenix, Mike Taylor

Java in a Nutshell. A Desktop Quick Reference. 9th Edition Benjamin J. Evans, Jason R. Clark, David Flanagan

Cena zestawu: 653.74 zł

Oszczędzasz: 153,26 zł (19%)

Dodaj do koszyka

AI agents need the right context at the right time to do a good job. Too much input increases cost and harms accuracy, while too little causes instability and hallucinations. Context Engineering with DSPy introduces a practical, evaluation-driven way to design AI systems that remain reliable, predictable, and easy to maintain as they grow.

AI engineer and educator Mike Taylor explains DSPy in a clear, approachable style, showing how its modular structure, portable programs, and built-in optimizers help teams move beyond guesswork. Through real examples and step-by-step guidance, you'll learn how DSPy's signatures, modules, datasets, and metrics work together to solve context engineering problems that evolve as models change and workloads scale.

This book supports AI engineers, data scientists, machine learning practitioners, and software developers building AI agents, retrieval-augmented generation (RAG) systems, and multistep reasoning workflows that hold up in production.

Understand the core ideas behind context engineering and why they matter
Structure LLM pipelines with DSPy's maintainable, reusable components
Apply evaluation-driven optimizers like GEPA and MIPROv2 for measurable improvements
Create reproducible RAG and agentic workflows with clear metrics
Develop AI systems that stay robust across providers, model updates, and real-world constraints

Wybrane bestsellery

O autorze książki

Mike Taylor współtworzył agencję marketingu wzrostowego Ladder. Prowadził niezwykle popularne kursy marketingu i AI w serwisach LinkedIn Learning, Udemy i Vexpower.

Ebooka "Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents" przeczytasz na:

czytnikach Inkbook, Kindle, Pocketbook, Onyx Booxs i innych
systemach Windows, MacOS i innych

systemach Windows, Android, iOS, HarmonyOS
na dowolnych urządzeniach i aplikacjach obsługujących formaty: PDF, EPub, Mobi

Masz pytania? Zajrzyj do zakładki Pomoc »

Oceny i opinie klientów: Context Engineering with DSPy. Self-Optimizing Prompt Pipelines for Building Reliable AI Agents Mike Taylor

(0)

Szczegóły książki

ISBN Ebooka:: 979-83-416-7124-9, 9798341671249
Data wydania ebooka :: 2026-10-06 Data wydania ebooka często jest dniem wprowadzenia tytułu do sprzedaży i może nie być równoznaczna z datą wydania książki papierowej. Dodatkowe informacje możesz znaleźć w darmowym fragmencie. Jeśli masz wątpliwości skontaktuj się z nami sklep@helion.pl.
Język publikacji:: angielski
Rozmiar pliku ePub:: 3.3MB
Rozmiar pliku Mobi:: 3.3MB

Zgłoś erratę
Kategorie:
Sztuczna inteligencja » Prompt Engineering
Sztuczna inteligencja » Inżynieria AI
Sztuczna inteligencja » Budowanie aplikacji AI

Spis treści książki

Preface
- Software Requirements for This Book
  - Installing the Project with uv
  - Installing Without uv
  - Configuring Environment Variables
  - Additional Software for Later Chapters
    - Deno
    - Docker and Qdrant
    - Other optional local tools
  - Additional Tips
- Conventions Used in This Book
- Using Code Examples
- OReilly Online Learning
- How to Contact Us
- Acknowledgments
1. Introduction to Context Engineering
- Defining Context Engineering
- Context Engineering Versus Prompt Engineering
- Introducing DSPy
- Failure Modes for Long Contexts
  - Context Overflow
  - Token Costs
  - Context Distraction
  - Lost in the Middle
  - Prompt Injection
  - Context Confusion
  - Context Fragmentation
  - Context Clash
  - Context Drift
  - Context Poisoning
  - Context Underspecification
- Context Engineering Techniques
  - Context Editing
  - Context Summarization
  - Memory Offloading
  - Prompt Optimization
  - Few-Shot Learning
  - Retrieval-Augmented Generation
  - Tool Loadout
  - Guardrails
  - Control Flow
  - Subagents
- Conclusion
2. Introduction to DSPy
- Language Models
  - Setting Up Language Models
  - Changing LM Provider
- Signatures
  - Defining a Signature
  - Get Over the Unfamiliarity
- Modules
  - Predict Module
  - Built-In Modules
  - ReAct Agents
  - Stacking Signatures Together
  - Recursive Language Models
- Evaluation
  - Example Datasets
  - Evaluation Metrics
  - Running Evaluations
  - Train-Test Split
- Optimizers
- DSPys Philosophy
- How I Use DSPy
- Conclusion
3. DSPy in Eight Steps
- Configuring Our Program
- Specify Your Signatures
  - Humanize AI Text
  - Detect AI Text
- Build Your Modules
  - TextTransformer
  - AIDetector
- Explore a Few Examples
  - Run Your Program
  - Inspect History
  - MLflow Traces
- Collect Your Dataset
  - Example Dataset
  - Load from CSV
  - Convert to DSPy Example Format
  - Train-Test Split
- Define Your Metrics
  - Exact Match
  - LLM Judge
  - Testing Your Metric
- Establish a Baseline
  - Evaluate Your Judge
  - Evaluate Your Task
- Optimize Your Program
  - Running the GEPA Optimizer
  - LLM Judge Metric
  - Optimizing Your Task Program
  - Evaluating Your Optimized Program
- Test and Iterate
- Conclusion
4. Strategies for Collecting Datasets
- Error Analysis
  - Manually Review and Annotate Data
  - Categorize Errors
  - Define a Dataset
- Domain Experts
  - Qualitative Interview
  - Example Correction
  - Gold Standard or Fools Gold?
- Third-Party Datasets
  - Built-In DSPy Datasets
  - Hugging Face Datasets
  - Kaggle Datasets
  - Practical Example: Combining Third-Party and Custom Data
- Synthetic Bootstrapping
  - Model Distillation
  - Data Enrichment
  - Synthetic de-identification
- Conclusion
5. Evaluation Metrics
- String Comparison
  - Regex Match
    - Example: Verifying output structure
    - Example: Checking for key information
  - Edit Distance
- Semantic Similarity
  - Embedder
  - Embedding Distance
    - Using embeddings as a metric
    - Multi-field similarity
  - NLP Classifiers
    - BLEU (Bilingual Evaluation Understudy)
    - ROUGE (Recall-Oriented Understudy for Gisting Evaluation)
    - Token F1
- LLM-as-a-Judge
  - Human-in-the-Loop
  - Training a Classifier
    - Step 1: Create judge Signature
    - Step 2: Convert human labels to training data
    - Step 3: Train the judge
    - Step 4: Validate judge accuracy
  - Evaluator-Optimizer Pattern
  - Cost Management
- Panel of Judges
  - Criteria Rubrics
  - Weighted Scoring
  - Multi-Predictor Feedback
    - Providing per-predictor feedback
    - Real-world example: Multi-hop QA
- Conclusion
6. Deep Dive into Prompt Optimizers
- Setting Up the AI Detection Module
- Automatic Few-Shot Learning Optimizers
  - LabeledFewShot
  - BootstrapFewShot
  - BootstrapFewShotWithRandomSearch (BootstrapRS)
  - KNNFewShot
- Automatic Instruction Optimizers
  - COPRO
  - MIPROv2
  - GEPA
    - Custom instruction proposers
  - SIMBA
- Automatic Fine-Tuning Optimizers
  - BootstrapFinetune
  - BetterTogether
- Program Transformations with Ensemble
- Choosing the Right Optimizer
- Conclusion
7. Custom Modules, Types, and Adapters
- Built-In Modules
  - Predict
  - ChainOfThought
  - BestOfN
  - Refine
  - MultiChainComparison
  - ReAct
  - ProgramOfThought
  - CodeAct
  - Recursive Language Models
  - Flex
  - Built-In Module Summary
- Custom DSPy Programs
  - Patterns for Building Programs with Multiple Modules
    - Pattern 1: Sequential pipeline
    - Pattern 2: Conditional branching
    - Pattern 3: Wrapping modules
    - Pattern 4: Per-module model assignment
    - Pattern 5: Self-refinement loop
  - Building a Custom Module: Verbalized Sampling
- Other Components
  - Majority
  - Parallel
- Multimodal Types
  - Images
  - Audio
  - Subclassing Type for Videos
- Chat Adapters
  - ChatAdapter
  - JSONAdapter
  - XMLAdapter
  - TwoStepAdapter
- Building Custom Adapters
- Conclusion
8. Building AI Agents
- Agents Versus Workflows
  - Definition of an Agent
  - Agentic Design Patterns
  - When to Use Agents Versus Workflows
- DSPys ReAct Agent
  - Defining Tools
  - ReAct Module
  - Subagents as Tools
  - Framework Comparison
    - DSPy
    - LangChain
    - Pydantic AI
    - OpenAI Agents SDK
    - CrewAI
  - Tracing Trajectories
  - MCP Integrations
- Adding Retrieval to Agents
  - Retrieval Clients
  - Web Search Tools
  - Multi-Hop Search
    - ReAct with retrieval tools
    - Custom multi-hop module
- Giving Agents Memory
  - Conversation History
  - Memory Editing
  - Editing Tool Results
  - Memory with Mem0
  - Context Management with RLMs
- Conclusion
9. Real-World Use Cases
- Social Media Sentiment Classifier
  - Capturing Your Teams Preferences as Examples
  - Optimizing with BootstrapFewShotWithRandomSearch
- Invoice Entity Extraction
  - Few-Shot Examples
  - Majority Voting
  - GEPA Instruction Learning
- Marketing Content Generator
  - Generating Custom Styles
  - Style Transfer
  - Semantic Distance
- News Research Summarizer
  - Web Search
  - Chunking Strategy
  - Recursive Summarization
- Fashion Video Generator
  - Generate Text-to-Image Tool
  - Generate Image-to-Video Tool
  - Evaluator-Optimizer Pattern
- Customer Service RAG Agent
  - RAG Retrieval Client
  - Prompt Injection Guardrails
  - Privacy Guardrails
- Financial Analyst Agent
  - Document Parsing
  - RLM for Financial Analysis
  - Trajectory Evaluation
- Conclusion
10. Putting DSPy into Production
- Monitoring and Observability with MLflow
  - Setting Up MLflow
  - Viewing Logs
  - Tracing Optimizers
  - Tracking Experiments
  - Querying Traces with the MLflow MCP Server
- Deployment with FastAPI
  - FastAPI Route
  - Saving and Loading Programs
  - Exporting Prompts in OpenAI Format
  - Avoiding DSPy-Flavored Markers in Exported Prompts
  - Guardrails
  - Async
  - Streaming
  - Error Handling and Rate Limits
  - Caching in Production
- Building a Gradio User Interface (DSPyUI)
  - Signatures
  - Modules
  - Datasets
  - Optimization
  - Inference
- DSPy in Other Languages
- Conclusion
11. Optimizing Coding Agents with DSPy
- Optimizing a SKILL.md End-to-End
  - Turning Failed Sessions into Skills
  - The optimize_anything Library
- Building CLIs with DSPy
  - Task Decomposition
  - Synthetic Task Generation
  - Testing Cheaper Models
- Self-Evolving Agents
  - Automated Skill Discovery with RLMs
  - Testing AGENTS.md Instructions
  - Clawsona: Optimizing Agent Personas
- The Optimizer Is a Commodity
- What You Built
- Closing Thoughts
Index