Historique des commits

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