HumanCompatibleAI/imitation

Health v2 · maintenance only

Adoption not part of the score

1779 stars · 303 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

A Python library providing clean PyTorch implementations of imitation and reward learning algorithms such as Behavioral Cloning, DAgger, GAIL, AIRL, and inverse reinforcement learning. It integrates with Gymnasium environments and Stable-Baselines3 for training agents from expert demonstrations.

Use cases

• train an agent from expert demonstrations
• implement behavioral cloning for a gymnasium environment
• run GAIL or AIRL imitation learning experiments
• perform inverse reinforcement learning to recover reward functions
• compare imitation learning algorithm baselines in research
• learn a reward model from human preferences

When to choose

• you need well-tested PyTorch implementations of standard imitation learning algorithms
• you want to train policies from demonstrations in Gymnasium environments
• you are doing research on reward learning or IRL and need reliable baselines

When to avoid

• you need general-purpose RL algorithms without demonstrations
• you want TensorFlow-based implementations
• you need production deployment of trained policies at scale

Facets

library · maturity active

machine-learning reinforcement-learning machine-learning reinforcement-learning artificial-intelligence python cross-platform imitation-learning reward-learning inverse-reinforcement-learning pytorch gymnasium reinforcement-learning-algorithms behavioral-cloning gail airl dagger

Member repositories

For agents

markdown · JSON · MCP: product_card(name="HumanCompatibleAI/imitation")