High-performance safetensors model loader supporting Linux/CUDA, ROCm, Windows, and unified-memory systems like DGX Spark. Enhances initialization speed in projects like vLLM and SGLang with command-line options.

fastsafetensors

fastsafetensors-0.3.2-cp310-cp310-manylinux_2_38_aarch64.whl

fastsafetensors-0.3.2-cp311-cp311-manylinux_2_38_aarch64.whl

fastsafetensors-0.3.2-cp312-cp312-manylinux_2_38_aarch64.whl

fastsafetensors-0.3.2-cp313-cp313-manylinux_2_38_aarch64.whl

fastsafetensors-0.3.2-cp314-cp314-manylinux_2_38_aarch64.whl

fastsafetensors-0.3.2-cp310-cp310-manylinux_2_38_x86_64.whl

fastsafetensors-0.3.2-cp311-cp311-manylinux_2_38_x86_64.whl

fastsafetensors-0.3.2-cp312-cp312-manylinux_2_38_x86_64.whl

fastsafetensors-0.3.2-cp313-cp313-manylinux_2_38_x86_64.whl

fastsafetensors-0.3.2-cp314-cp314-manylinux_2_38_x86_64.whl