lossless-compression

Open repos and models their makers tagged lossless-compression, most starred first. The same for everyone.

AI on my own machineRepo

NWC

Neural Weight Compression: lossless LLM weight compression decoded inside the CUDA matvec. BF16 31 % smaller and faster than cuBLAS; fp8 13 % smaller at native fp8 speed. Bit-identical weights.

  • 4
  • Cuda
  • Apache-2.0
  • 2 mo ago
parda21Take it
Search GitHub and Hugging Face for lossless-compression

Get , open it once and sign in. Then come back here and press the button again.