Skip to content

OrbitQuant 0.1.5

Choose a tag to compare

@iamwavecut iamwavecut released this 09 Jul 22:01

Paper-conformant quantizer update.

  • Adds converged beta-marginal Lloyd-Max codebook version 2.
  • Matches Algorithm 1 activation normalization with s + epsilon on CPU, CUDA/Triton, and Metal/MPS.
  • Preserves explicit version 1 loading for existing packed artifacts.
  • Validates artifact codebooks and RPBH sidecars against the runtime basis.
  • Keeps auto_fused packed CUDA/MPS inference as the default runtime.