Save and Load a Prompter¶

Persist optimized prompters for production use.

Save After Optimization¶

from dspydantic import Prompter

# Create and optimize
prompter = Prompter(model=MyModel, model_id="openai/gpt-4o-mini")
result = prompter.optimize(examples=examples)

# Save the optimized prompter
prompter.save("./my_prompter")

What is saved: - Optimized field descriptions - Optimized system and instruction prompts - Model configuration (field types, structure)

What is NOT saved: - API keys (you provide these at load time) - Examples - Optimization history

Load in Production¶

from dspydantic import Prompter

# Load the optimized prompter
prompter = Prompter.load(
    "./my_prompter",
    model=MyModel,
    model_id="openai/gpt-4o-mini",
    # API key can come from environment variable
)

# Use it for extraction
result = prompter.run("input text")

Configure DSPy before loading if needed:

import dspy

dspy.configure(lm=dspy.LM("openai/gpt-4o-mini", api_key="your-key"))

prompter = Prompter.load(
    "./my_prompter",
    model=MyModel,
    model_id="openai/gpt-4o-mini"
)

Versioning¶

Use versioning for managing multiple prompter versions:

from datetime import datetime

# Timestamp versioning
timestamp = datetime.now().strftime("%Y%m%d_%H%M%S")
prompter.save(f"./prompters/{timestamp}/model")

# Semantic versioning
prompter.save("./prompters/v1.0.0/model")

# Latest symlink (optional, depends on your setup)
# ln -s v1.0.0 ./prompters/latest

Model Upgrade Compatibility¶

Safe changes (don't require re-optimization): - Adding new optional fields

Incompatible changes (require re-optimization): - Changing field types - Renaming fields - Removing fields

Tips¶

Always test loaded prompters before production
Use environment variables for API keys
Version your prompters for easy rollback
Document which model version each prompter uses
Keep a record of optimization parameters used