maxent\_grpo.training.optim
===========================

.. automodule:: maxent_grpo.training.optim

   
   .. rubric:: Functions

   .. autosummary::
   
      _filter_optimizer_kwargs
      apply_learning_rate
      build_optimization_handles
      clip_grad_norm_local
      configure_accumulation_steps
      detect_deepspeed_state
      epoch_progress
      optimizer_step
      require_accumulation_context
      scheduled_learning_rate
      sync_gradients_enabled
   
   .. rubric:: Classes

   .. autosummary::
   
      DeepspeedState
      DistributedType