dspy

Use DSPy optimizers like BootstrapFewShot or MIPRO with a metric function and training examples. Call optimizer.compile on your module with the trainset, and DSPy generates few-shot demonstrations or improved instructions automatically.

Predict makes direct LM calls without reasoning steps, making it faster for simple tasks. ChainOfThought generates a rationale before the answer, improving accuracy on reasoning tasks at roughly twice the latency.

DSPy optimizes prompts automatically from data and offers typed signatures with high modularity, while LangChain relies on manual prompt engineering with prebuilt chains. Choose DSPy when you have training data and need systematic improvement.

Yes, DSPy supports local models through dspy.OllamaLocal by pointing to a local base URL such as http://localhost:11434. It also supports OpenAI, Anthropic, and other providers through dedicated LM classes.

BootstrapFewShot works with 10-50 examples, while MIPRO benefits from 50-200 examples plus a separate validation set. BootstrapFinetune for weight fine-tuning recommends 100 or more examples.

Avoid DSPy for quick one-off prototypes where manual prompting suffices, or when you lack any training data for optimization. Simple chains with existing tools may be faster to build with lighter frameworks.