提交历史

作者 SHA1 备注 提交日期
  AlpinDale 00503b9fc1 feat: non-uniform quantization via `compressed-tensors` for llama 5 月之前
  AlpinDale ee2c5d34da feat: add fp8 channel-wise weight quantization support 5 月之前
  AlpinDale 98cb1c4cd1 feat: support fp8 via `llm-compressor` 5 月之前
  AlpinDale e2dbe5f05c feat: add sparse marlin for compressed tensors 5 月之前
  AlpinDale aba03b4756 feat: dynamic per-token activation quantization 5 月之前