AgentSkillsCN

dspy-configuration

DSPy 配置、LM 设置、缓存管理与版本控制。当您需要配置语言模型、启用缓存、管理依赖关系，或搭建多供应商配置时，可使用此技能。

SKILL.md
--- frontmatter
name: dspy-configuration
description: DSPy configuration, LM setup, caching, and version management. Use when configuring language models, enabling caching, managing dependencies, or setting up multi-provider configurations.

DSPy Configuration

DSPy configuration, LM setup, caching, and version management.

Quick Start

Configure LM Globally

python
import dspy

# Configure LM globally
dspy.configure(lm=dspy.LM('gemini/gemini-3-flash'))
response = qa(question="How many floors are in the castle?")
print('Gemini-3-flash:', response.answer)

Switch LM Locally

python
# Change LM within a context block
with dspy.context(lm=dspy.LM('gemini/gemini-3-flash')):
    response = qa(question="How many floors are in the castle?")
    print('Gemini-3-flash:', response.answer)

Enable Caching

python
# Caching is enabled by default in DSPy

# Method 1: Configure cache globally
dspy.configure_cache(
    enable_disk_cache=True,
    enable_memory_cache=True,
    disk_size_limit_bytes=30_000_000_000,  # 30 GB
    memory_max_entries=1000000,
)

# Method 2: Disable caching for a specific call
with dspy.context(cache=False):
    response = qa(question="Uncached question")

# Method 3: Configure cache at LM level
lm = dspy.LM(
    "gemini/gemini-3-flash",
    cache=True,  # Enable caching (default)
)
dspy.configure(lm=lm)

Configure LM Parameters

python
dspy.configure(
    lm=dspy.LM(
        "gemini/gemini-3-flash",
        temperature=1.0,
        max_tokens=16000,
        cache=True,
    ),
)

When to Use This Skill

Use this skill when:

  • •Configuring language models for DSPy
  • •Setting up LM switching (global/local)
  • •Enabling and customizing caching
  • •Configuring multi-provider LMs
  • •Managing version compatibility
  • •Setting up Responses API for advanced models

Core Concepts

LM Configuration

Configure which language models to use and how to switch between them.

Key features:

  • •Global configuration: dspy.configure(lm=...) for entire session
  • •Local overrides: dspy.context(lm=...) for code blocks
  • •Multi-provider: Support for OpenAI, Anthropic, Together AI, and more
  • •Responses API: Enable advanced model features

See: references/lm-config.md for:

  • •LM configuration patterns
  • •Multi-provider setup
  • •Responses API configuration
  • •Best practices

Caching

Manage cache behavior to improve performance and control costs.

Cache layers:

  • •In-memory cache: Fast access using cachetools.LRUCache
  • •On-disk cache: Persistent storage using diskcache.FanoutCache
  • •Server-side cache: Managed by LLM provider (OpenAI, Anthropic)

See: references/caching.md for:

  • •Cache architecture details
  • •Custom cache key implementation
  • •Cache debugging techniques
  • •Configuration options

Version Management

Track and manage dependency versions to ensure compatibility.

Key features:

  • •Automatic versioning: save() captures dependency versions
  • •Version checking: Alerts on version mismatches
  • •Compatibility: Prevents issues from outdated dependencies

See: references/versioning.md for:

  • •Dependency versioning patterns
  • •Version mismatch detection
  • •Compatibility best practices

Scripts

The scripts/ directory provides reusable tools:

  • •compile-dspy.py: Compile DSPy modules with proper caching
  • •clear-cache.py: Clear DSPy cache safely

Progressive Disclosure

This skill uses progressive disclosure:

  1. •SKILL.md (this file): Quick reference and navigation
  2. •references/: Detailed technical docs loaded as needed

Load reference files only when you need detailed information on a specific topic.

Related Skills

  • •dspy-basics: Signature design, basic modules, program composition
  • •dspy-optimization: Teleprompters, metrics, optimization workflows
  • •dspy-advanced: ReAct agents, tool calling, output refinement