afnio 0.2.5

Afnio Python library and Tellurio Studio CLI tool

Afnio: Making AI System Optimization Easy for Everyone

Afnio is a framework for automatic prompt and hyperparameter optimization, particularly designed for complex AI systems where Language Models (LMs) are employed multiple times in workflows, such as in LM pipelines and agent-driven architectures. Effortlessly build and optimize AI systems for classification, information retrieval, question-answering, etc.

• Accelerated AI System Development: Ship complex AI systems faster thanks to high-level UX and easy-to-debug runtime.
• State-of-the-Art Performance: Leverage built-in optimizers to automatically refine prompts and tune model parameters for any LM task, ensuring optimal performance.
• LM Agnostic: Decouple prompts and parameters from application logic, reducing LM model selection to a single hyperparameter in Afnio’s optimizers. Seamlessly switch between models without any additional rework.
• Minimal and Flexible: Pure Python with no API calls or dependencies, ensuring seamless integration with any tools or libraries.
• Progressive Disclosure of Complexity: Leverage diverse UX workflows, from high-level abstractions to fine-grained control, designed to suit various user profiles. Start simple and customize as needed, without ever feeling like you’re falling off a complexity cliff.
• Define-by-Run Scheme: Your compound AI system is dynamically defined at runtime through forward computation, allowing for seamless handling of complex control flows like conditionals and loops, common in agent-based AI applications. With no need for precompilation, Afnio adapts on the fly to your evolving system.

Contributing Guidelines

:computer: Would love to contribute? Please follows our contribution guidelines.