magnitudedev/magnitude
Open source inference engine that tunes its kernels on your hardware to find the fastest way to run each model. 2x faster than Ollama. Works on Apple Silicon, NVIDIA, AMD, Intel, or just a CPU.
7.6K
Stars
541
Forks
7.6K
Watchers
48
Issues
Loading additional information…