Added better error message for debugging on CUDA not detected failures.

Fixed missing code generation for older archs

CUDA script CMAKE fixes. some Cpuonly fixes.

Added version 0.37.0.

Fixed CUDA SETUP for tag

Updated docs (bitsandbytes-foundation#32) and changelog.

Fixed url in setup.py (bitsandbytes-foundation#38), updated changelog.

Added patch notes.

Fixed bug in Linear8bitLt, when the bias is None.

LLM.int8() release.