flashinfer.fused\_moe.CudnnGroupedGemmFp8PerTensorRunner
========================================================

.. currentmodule:: flashinfer.fused_moe

.. autoclass:: CudnnGroupedGemmFp8PerTensorRunner

   
   .. automethod:: __init__

   
   .. rubric:: Methods

   .. autosummary::
   
      ~CudnnGroupedGemmFp8PerTensorRunner.__init__
      ~CudnnGroupedGemmFp8PerTensorRunner.accepts
      ~CudnnGroupedGemmFp8PerTensorRunner.build
      ~CudnnGroupedGemmFp8PerTensorRunner.check_support
      ~CudnnGroupedGemmFp8PerTensorRunner.forward
      ~CudnnGroupedGemmFp8PerTensorRunner.get_cache_key_extras
      ~CudnnGroupedGemmFp8PerTensorRunner.get_valid_tactics
      ~CudnnGroupedGemmFp8PerTensorRunner.launch_kwargs_for
      ~CudnnGroupedGemmFp8PerTensorRunner.launch_state_for
      ~CudnnGroupedGemmFp8PerTensorRunner.pack_inputs
      ~CudnnGroupedGemmFp8PerTensorRunner.precompile_tactics
      ~CudnnGroupedGemmFp8PerTensorRunner.supports_quant
      ~CudnnGroupedGemmFp8PerTensorRunner.tuning_config_for
   
   

   
   
   .. rubric:: Attributes

   .. autosummary::
   
      ~CudnnGroupedGemmFp8PerTensorRunner.backend_key
      ~CudnnGroupedGemmFp8PerTensorRunner.supported_activation_classes
      ~CudnnGroupedGemmFp8PerTensorRunner.supported_activation_classes_by_quant
      ~CudnnGroupedGemmFp8PerTensorRunner.supported_output_formats
      ~CudnnGroupedGemmFp8PerTensorRunner.supported_quant_variants
      ~CudnnGroupedGemmFp8PerTensorRunner.supported_routing_modes
      ~CudnnGroupedGemmFp8PerTensorRunner.supports_expert_parallelism
      ~CudnnGroupedGemmFp8PerTensorRunner.supports_fused_shared_experts
      ~CudnnGroupedGemmFp8PerTensorRunner.supports_nvfp4_4over6
      ~CudnnGroupedGemmFp8PerTensorRunner.supports_per_token_scale
      ~CudnnGroupedGemmFp8PerTensorRunner.weight_view_aliases
      ~CudnnGroupedGemmFp8PerTensorRunner.config