Module `ilpyt.algos`

Algorithms are the main entrypoint for the ilpyt library. An algorithm's role during learning is to perform the high-level coordinatination between agent and environment during train and test time. An algorithm's key functions include a train and test function.

To create a custom algorithm, see BaseAlgorithm for more details.

Expand source code

"""
Algorithms are the main entrypoint for the ilpyt library. An algorithm's role 
during learning is to perform the high-level coordinatination between agent and 
environment during train and test time. An algorithm's key functions include a 
`train` and `test` function. 

To create a custom algorithm, see `BaseAlgorithm` for more details. 
"""

Sub-modules

ilpyt.algos.apprentice: An implementation of the Apprenticeship Learning (AppL) algorithm. This algorithm was described in the paper "Apprenticeship Learning via Inverse …
ilpyt.algos.base_algo: BaseAlgorithm is the abstract class for an algorithm. An algorithm's role during learning is to coordinate the agent and environment during train …
ilpyt.algos.bc: An implementation of a behavioral cloning (BC) algorithm, as in An Autonomous Land Vehicle in a Neural Network (ALVINN). The BC algorithm was …
ilpyt.algos.dagger: An implementation of the Dataset Aggregation (DAgger) algorithm. The DAgger algorithm was described in the paper "A Reduction of Imitation Learning …
ilpyt.algos.gail: An implementation of the Generative Adversarial Imitation Learning (GAIL) algorithm. This algorithm was described in the paper "Generative …
ilpyt.algos.gcl: An implementation of the Guided Cost Learning (GCL) algorithm. This algorithm was described in the paper "Guided Cost Learning: Deep Inverse Optimal …
ilpyt.algos.rl: A generic trainer for reinforcement learning (RL) algorithms. Compatible with the Advantage Actor Critic (A2C), Deep Q-Network (DQN), and Proximal …