Welcome Falcon Mamba: The first strong attention-free 7B model

In three linesHugging Face introduces Falcon Mamba, a 7B attention-free model based on Mamba architecture. It matches standard attention-based models in performance while delivering faster inference and reduced memory consumption.

Summary generated by Claude — human-verified