Skip to content

vllm.models.muse_spark.nvidia.ops

Modules:

  • cute_dsl –
  • gemm_rs –

    Automatic Muse Spark attention GEMM + reduce-scatter dispatch.

  • mnnvl_fusions –

    Workspaces for FlashInfer MNNVL collectives fused into Muse Spark boundaries.

  • prepare_megamoe_bf16 –

    Stage BF16 tokens and routing into MegaMoE's symmetric buffer.

  • prepare_megamoe_nvfp4 –

    Triton input staging for MuseSpark1x on the NVFP4 MegaMoE kernel.

  • prepare_trtllm_nvfp4 –

    Pre-expert RMSNorm fused with the TRT-LLM MoE's NVFP4 input quantization.