ray.rllib.core.learner.learner.Learner.apply_gradients
ray.rllib.core.learner.learner.Learner.apply_gradients#
- abstract Learner.apply_gradients(gradients_dict: Dict[Hashable, Union[torch.Tensor, tensorflow.python.ops.variables.Variable]]) None[source]#
Applies the gradients to the MultiAgentRLModule parameters.
- Parameters
gradients_dict – A dictionary of gradients in the same (flat) format as self._params. Note that top-level structures, such as module IDs, will not be present anymore in this dict. It will merely map gradient tensor references to gradient tensors.