Efficiently steer language models with structured output and reduced latency using Guidance.

Guidance is an efficient programming paradigm for steering language models. With Guidance, you can control how output is structured and get high-quality output for your use case—while reducing latency and cost compared to conventional prompting or fine-tuning. It allows users to constrain generation (e.g., with regex and CFGs) and interleave control (conditionals, loops, tool use) and generation seamlessly.
Features include:
Guidance supports various backends like Transformers, llama.cpp, and OpenAI. Install it via PyPI and start controlling your language models with precision.
+4 more