Introducing Cohere Sandbox: Open-Source Libraries to Help Developers Experiment with Language AI

Introducing Cohere Sandbox: Open-Source Libraries to Help Developers Experiment with Language AI

Explore, test, and experiment with LLMs via open-source: Cohere Sandbox is now available! We invite developers excited about large language models to use our new Sandbox to build and deploy advanced natural language processing (NLP) projects, fast.

Recent advances in large language models (LLMs) have fueled state-of-the-art performance for NLP applications, such as copy generation, tooling for conversational agents, semantic search, and more. At Cohere, we are fans of empowering our developer community to experiment and work with language AI. With Cohere’s Sandbox open-source libraries, we want to inspire and empower you to quickly build novel language applications powered by Cohere’s LLMs—all while tackling your specific use case.

For the past few months, we've been hard at work building Sandbox, a place where we release experimental and "earlybird" versions of open-source language AI modules, libraries, and demos to empower and inspire developers to build and deploy novel language applications fast with Cohere's LLMs.

What is Sandbox?

Sandbox is a collection of experimental, open-source GitHub repositories (links available below) that make building applications using large language models fast and easy with Cohere, and/or showcase how such applications are built. Sandbox repositories aim to enable all developers, regardless of ML experience, to:

  • Build and deploy applications, products, services, or even entire companies with large language models
  • Fork and modify Sandbox to experiment with large language models
  • Solve their use case quickly with large language models
  • Feel empowered with the tools you need to launch fast

Why use Sandbox?

Sandbox aims to help build and strengthen language AI communities while enabling contributors to build more robust applications and services faster than ever. At Cohere, we want to help keep that pace of innovation going and empower developers to gain access, knowledge, and exposure to LLMs. Developers can contribute to Sandbox, regardless of their skill level. Just fork it, read our contribution guidelines (located under each repository's README), and support our open-source efforts.

How to Access Sandbox Open-Source Repositories

Today, we want to share those development efforts with our developer community via a series of open-source GitHub repositories included in our first release.

In this initial release, we are introducing:

  • Conversant: A framework for building conversational agents on top of the Cohere API, with a hands-on demo on how to use generative language models in conversational settings and build those interactions.
  • Route Generation: Build a functional chatbot that recognizes users' intent from descriptions, maps incoming user messages, and accelerates its training by leveraging Cohere's models to enable zero-shot learning.
  • Grounded QA:  A powerful, contextualized, factual question-answering Discord bot that uses embeddings, text generation, and web search.
  • Topically: A suite of tools that help you use the best of topic modeling to make sense of text collections (messages, articles, emails, news headlines, etc.) using large language models.
  • Toy Semantic Search: A simple semantic search engine built with the Cohere API. The search algorithm here is fairly straightforward; it uses embeddings to find the paragraph that matches the question's representation. In text sources, a concrete paragraph containing the answer is most likely to produce the best results.

What’s Next?

While LLMs are relatively new and incredibly powerful, not everyone has first-hand access to them. By sharing experimental code as part of Cohere's Sandbox, we want to empower you with the right open-source tools to discover what's possible with language AI and help you build and deploy, fast.

This is just the beginning for Sandbox, as we plan to continue releasing new hands-on repositories to help with your use cases and deliver know-how and inspiration, over time. If you're feeling creative, consider supporting our open-source efforts by forking and experimenting with the projects, and feel free to share them with your friends and colleagues.

We welcome all feedback, issues, feature requests, contributions, documentation improvements, and bug reports. For more info, visit our guide on how to contribute (located under each project's repo). Don't forget to visit our Discord community and submit any ideas or feedback. For updates on our OSS efforts, follow us on Twitter and Linkedin.