Caiche: Unified Language Model Cache

Welcome to Caiche – your gateway to efficient and cached language model interactions! 🚀

What is Caiche?

Caiche is a Unified Language Model Cache designed to streamline your language model queries. Imagine a world where you can ask a question and instantly get a response, thanks to a global vector database that stores and retrieves prompt-response pairs. That's Caiche for you – making your language model interactions faster and cheaper.

Installation

Installing Caiche is a breeze. Open your terminal and run:

pip install llm-caiche

Getting Started

Now that you have Caiche installed, let's dive into some quick code snippets to get you started:

from caiche import Caiche

# Replace <OPENAI_API_KEY> with your actual OpenAI API key
cai = Caiche(api_key="<OPENAI_API_KEY>")

# Ask a question and get a response
result = cai.query("What is the capital of Bulgaria?")

And just like that, you're ready to unlock the power of cached language model responses!

Why Caiche?

1. Cheaper Responses

Enjoy cost savings as you won't be charged for inference costs when a response is cached.

2. Faster Queries

Caiche eliminates the need to wait for language models to generate responses by storing and retrieving previous interactions. Say goodbye to unnecessary delays!

3. Global Database

Your prompts and responses are stored globally, allowing you to benefit from a vast database of interactions. Join the Caiche community and contribute to a shared knowledge space.

What's Next?

At Caiche, we're not just stopping at efficient cached responses. Our vision extends beyond, aiming to redefine the landscape of language model interactions. Here's a glimpse of what's on the horizon:

Approximate Cache for Semantically Similar Prompts

We're actively working on building an approximate cache that goes beyond storing exact prompt-response pairs. Our goal is to leverage the collected data to lower costs not only for existing prompts but also for prompts that are semantically similar. Imagine getting cost-efficient responses for a range of related queries, making your interactions even more budget-friendly and responsive.

Community and Contributions

We believe in the power of community collaboration. If you have ideas, improvements, or want to report issues, head over to our GitHub repository and let your voice be heard!

Example Use Cases

1. Conversational Agents Enhance the user experience of your chatbots and conversational agents by tapping into the Caiche global database.

2. Research and Development Expedite your language model experimentation process by utilizing stored interactions for quicker testing.

Contributing

We will be adding contributing guidelines soon.

License

Caiche is licensed under the MIT License. Feel free to explore, modify, and share!

Ready to revolutionize your language model interactions? Install Caiche, explore the possibilities, and join us on this exciting journey!

Happy querying! 🚀✨