Commit History

Autor SHA1 Mensaxe Data
  AlpinDale 9be43994fe feat: fbgemm quantization support (#601) hai 5 meses
  AlpinDale 00503b9fc1 feat: non-uniform quantization via `compressed-tensors` for llama hai 5 meses
  AlpinDale e1475fbec7 feat: MoE support with Pallas GMM kernel for TPUs hai 6 meses
  AlpinDale 4bbf66451a chore: add CustomAP interface to UnquantizedFusedMoEMethod hai 6 meses
  AlpinDale 1efd0f89b7 feat: support FP8 for DeepSeekV2 MoE hai 6 meses
  AlpinDale cf472315cc refactor: isolate FP8 from mixtral hai 6 meses