dmx.compressor.sparse.BlockTopK
===============================

.. currentmodule:: dmx.compressor.sparse

.. autoclass:: BlockTopK

   
   .. automethod:: __init__

   
   .. rubric:: Methods

   .. autosummary::
   
      ~BlockTopK.__init__
      ~BlockTopK.apply
      ~BlockTopK.backward
      ~BlockTopK.forward
      ~BlockTopK.from_shorthand
      ~BlockTopK.get_mask
      ~BlockTopK.jvp
      ~BlockTopK.mark_dirty
      ~BlockTopK.mark_non_differentiable
      ~BlockTopK.mark_shared_storage
      ~BlockTopK.maybe_clear_saved_tensors
      ~BlockTopK.name
      ~BlockTopK.register_hook
      ~BlockTopK.register_prehook
      ~BlockTopK.save_for_backward
      ~BlockTopK.save_for_forward
      ~BlockTopK.set_materialize_grads
      ~BlockTopK.setup_context
      ~BlockTopK.vjp
      ~BlockTopK.vmap
   
   

   
   
   .. rubric:: Attributes

   .. autosummary::
   
      ~BlockTopK.blocked
      ~BlockTopK.dirty_tensors
      ~BlockTopK.generate_vmap_rule
      ~BlockTopK.materialize_grads
      ~BlockTopK.metadata
      ~BlockTopK.needs_input_grad
      ~BlockTopK.next_functions
      ~BlockTopK.non_differentiable
      ~BlockTopK.requires_grad
      ~BlockTopK.saved_for_forward
      ~BlockTopK.saved_tensors
      ~BlockTopK.saved_variables
      ~BlockTopK.to_save