We’re super joyful to announce that GGML, creators of Llama.cpp, are joining HF in an effort to keep future AI open. 🔥
Georgi Gerganov and team are joining HF with the goal of scaling and supporting the community behind ggml and llama.cpp as Local AI continues to make exponential progress in the approaching years.
We have been working with Georgi and team for quite a while (we even have awesome core contributors to llama.cpp like Son and Alek within the team already) so this has been a really natural process.
llama.cpp is the elemental constructing block for local inference, and transformers is the elemental constructing block for model definition, so this is largely a match made in heaven. ❤️
What’s going to change for llama.cpp, the open source project and the community?
Not much – Georgi and team still dedicate 100% of their time maintaining llama.cpp and have full autonomy and leadership on the technical directions and the community.
HF is providing the project with long-term sustainable resources, improving the probabilities of the project to grow and thrive. The project will proceed to be 100% open-source and community driven because it is now.
Technical focus
llama.cpp is the elemental constructing block for local inference, and transformers is the elemental constructing block for definition of models and architectures, so we’ll work on ensuring it’s as seamless as possible in the long run (almost “single-click”) to ship recent models in llama.cpp from the transformers library ‘source of truth’ for model definitions.
Moreover, we’ll improve packaging and user experience of ggml-based software. As we enter the phase by which local inference becomes a meaningful and competitive alternative to cloud inference, it’s crucial to enhance and simplify the best way by which casual users deploy and access local models. We’ll work towards making llama.cpp ubiquitous and available in all places.
Our long run vision
Our shared goal is to offer the community with the constructing blocks to make open-source superintelligence accessible to the world over the approaching years.
We’ll achieve this along with the growing Local AI community, as we proceed to construct the last word inference stack that runs as efficiently as possible on our devices.
