Compose declarative LM pipelines with optimizers like BootstrapFewShot and MIPRO.
npx skills add https://github.com/KarlinskyS/hermesSkills --skill dspy-karlinskys
Please help me install this Agent Skill. Skill: dspy Source: https://github.com/KarlinskyS/hermesSkills/tree/main/mlops/research/dspy Command: npx skills add https://github.com/KarlinskyS/hermesSkills --skill dspy-karlinskys
DSPy removes the ad-hoc, brittle process of manual prompt engineering by providing a declarative framework to compose, optimize, and maintain language-model-based systems across development and production.
Create a ChainOfThought QA module that retrieves three passages, answers concisely, and optimize it using BootstrapFewShot.
High-intent search queries and answers about installing and using this skill.
You can optimize language model prompts programmatically by applying optimizers like BootstrapFewShot, MIPRO, and BootstrapFinetune to declarative modules, automatically improving prompts and few-shot demonstrations using representative training examples.
The best way to build a modular RAG pipeline is by composing declarative modules that retrieve top passages and apply ChainOfThought reasoning, allowing you to structure multi-stage workflows for question answering and classification.
Yes, you can swap LM providers like OpenAI, Anthropic, and local runtimes in your multi-stage AI pipelines, allowing you to configure and export optimized modules for production use without altering the underlying pipeline logic.
Manual prompt engineering becomes brittle because it relies on ad-hoc adjustments that fail to scale, whereas using a declarative framework allows you to compose, automatically optimize, and maintain language-model systems across development and production.
DSPy works with local runtimes for classification and summarization tasks by defining typed inputs and outputs through Signatures and Modules, enabling multi-provider configuration and automatic optimization of the underlying language model calls.
zengbaochengBuild language-model workflows with declarative typed signatures in Python.
belokonmDeploy ML models to production with infrastructure-as-code templates and monitoring.
luokai25Develop RAG systems with chunking, embedding, vector search, and reranking.
yanochka11Optimize declarative language model programs with automated improvement workflows.
