Hugging Face has acquired the GGML.AI team, the creators of ggml and llama.cpp, while ensuring that both projects remain governed independently under ggml-org. This acquisition aims to enhance full-time maintenance and improve the accessibility of local inference for non-experts, ultimately speeding up the integration of new multimodal models and decreasing implementation delays for emerging architectures. As a result, tighter compatibility with Hugging Face’s Transformers library is expected, which will reduce discrepancies between local and hosted model runs.
Hugging Face acquires GGML.AI team, enhances llama.cpp compatibility
