A library providing k-bit optimizers and matrix multiplication routines for efficient quantization of large language models, enabling 8-bit and 4-bit inference and training on consumer GPUs with significantly reduced memory requirements.

bitsandbytes

bitsandbytes-0.50.2-2.1-cp310-cp310-manylinux_2_38_aarch64.whl

bitsandbytes-0.50.2-2.1-cp311-cp311-manylinux_2_38_aarch64.whl

bitsandbytes-0.50.2-2.1-cp312-cp312-manylinux_2_38_aarch64.whl

bitsandbytes-0.50.2-2.1-cp313-cp313-manylinux_2_38_aarch64.whl

bitsandbytes-0.50.2-2.1-cp314-cp314-manylinux_2_38_aarch64.whl

bitsandbytes-0.50.2-2.1-cp310-cp310-manylinux_2_38_x86_64.whl

bitsandbytes-0.50.2-2.1-cp311-cp311-manylinux_2_38_x86_64.whl

bitsandbytes-0.50.2-2.1-cp312-cp312-manylinux_2_38_x86_64.whl

bitsandbytes-0.50.2-2.1-cp313-cp313-manylinux_2_38_x86_64.whl

bitsandbytes-0.50.2-2.1-cp314-cp314-manylinux_2_38_x86_64.whl