Commit History

Auteur SHA1 Bericht Datum
  AlpinDale 00503b9fc1 feat: non-uniform quantization via `compressed-tensors` for llama 5 maanden geleden
  AlpinDale ee2c5d34da feat: add fp8 channel-wise weight quantization support 5 maanden geleden
  AlpinDale 98cb1c4cd1 feat: support fp8 via `llm-compressor` 5 maanden geleden
  AlpinDale e2dbe5f05c feat: add sparse marlin for compressed tensors 6 maanden geleden
  AlpinDale aba03b4756 feat: dynamic per-token activation quantization 6 maanden geleden