API Reference

• Algorithms
  • Continuous control algorithms
  • Discrete control algorithms
• Q Functions
  • d3rlpy.models.q_functions.MeanQFunctionFactory
  • d3rlpy.models.q_functions.QRQFunctionFactory
  • d3rlpy.models.q_functions.IQNQFunctionFactory
  • d3rlpy.models.q_functions.FQFQFunctionFactory
• MDPDataset
  • d3rlpy.dataset.MDPDataset
  • d3rlpy.dataset.Episode
  • d3rlpy.dataset.Transition
  • d3rlpy.dataset.TransitionMiniBatch
• Datasets
  • d3rlpy.datasets.get_cartpole
  • d3rlpy.datasets.get_pendulum
  • d3rlpy.datasets.get_pybullet
  • d3rlpy.datasets.get_atari
  • d3rlpy.datasets.get_d4rl
  • d3rlpy.datasets.get_dataset
• Preprocessing
  • Observation
  • Action
• Optimizers
  • d3rlpy.models.optimizers.OptimizerFactory
  • d3rlpy.models.optimizers.SGDFactory
  • d3rlpy.models.optimizers.AdamFactory
  • d3rlpy.models.optimizers.RMSpropFactory
• Network Architectures
  • d3rlpy.models.encoders.DefaultEncoderFactory
  • d3rlpy.models.encoders.PixelEncoderFactory
  • d3rlpy.models.encoders.VectorEncoderFactory
  • d3rlpy.models.encoders.DenseEncoderFactory
• Metrics
  • Algorithms
  • Dynamics
• Off-Policy Evaluation
  • For continuous control algorithms
  • For discrete control algorithms
• Save and Load
  • save_model and load_model
  • from_json
  • save_policy
• Logging
  • TensorBoard
• scikit-learn compatibility
  • train_test_split
  • cross_validate
  • GridSearchCV
  • parallel execution with multiple GPUs
• Online Training
  • Standard Training
  • Batch Concurrent Training
• Model-based Algorithms
  • Dynamics Model
• Stable-Baselines3 Wrapper
  • Convert SB3 replay buffer to d3rlpy dataset
  • Convert d3rlpy to use SB3 helpers