Center for Human-Compatible AI
CHAI seeks to develop the conceptual and technical wherewithal to reorient the general thrust of AI research towards provably beneficial systems.
Pinned Repositories
adversarial-policies
Find best-response to a fixed policy in multi-agent RL
eirli
An Empirical Investigation of Representation Learning for Imitation (EIRLI), NeurIPS'21
evaluating-rewards
Library to compare and evaluate reward functions
human_aware_rl
Code for "On the Utility of Learning about Humans for Human-AI Coordination"
imitation
Clean PyTorch implementations of imitation and reward learning algorithms
overcooked-demo
Web application where humans can play Overcooked with AI agents.
overcooked_ai
A benchmark environment for fully cooperative human-AI performance.
rlsp
Reward Learning by Simulating the Past
seals
Benchmark environments for reward modelling and imitation learning algorithms.
tensor-trust
A prompt injection game to collect data for robust ML research
Center for Human-Compatible AI's Repositories
HumanCompatibleAI/rlsp
Reward Learning by Simulating the Past
HumanCompatibleAI/atari-irl
HumanCompatibleAI/deep-rlsp
Code accompanying "Learning What To Do by Simulating the Past", ICLR 2021.
HumanCompatibleAI/population-irl
(Experimental) Inverse reinforcement learning from trajectories generated by multiple agents with different (but correlated) rewards
HumanCompatibleAI/learning_biases
Infer how suboptimal agents are suboptimal while planning, for example if they are hyperbolic time discounters.
HumanCompatibleAI/human_ai_robustness
HumanCompatibleAI/better-adversarial-defenses
Training in bursts for defending against adversarial policies
HumanCompatibleAI/interpreting-rewards
Experiments in applying interpretability techniques to learned reward functions.
HumanCompatibleAI/derail
Supporting code for diagnostic seals paper
HumanCompatibleAI/minerl
MineRL Competition for Sample Efficient Reinforcement Learning - Python Package
HumanCompatibleAI/multi-agent
HumanCompatibleAI/cs294-149-fa18-notes
LaTeX Notes from the Fall 2018 version of CS294-149: AGI Safety and Control
HumanCompatibleAI/ilqr
Iterative Linear Quadratic Regulator with auto-differentiatiable dynamics models
HumanCompatibleAI/logical-active-classification
Use active learning to classify data represented as boundaries of regions in parameter space where a parametrised logical formula holds.
HumanCompatibleAI/simulation-awareness
(experimental) RL agents should be more aligned if they do not know whether they are in simulation or in the real world
HumanCompatibleAI/interactive-behaviour-design
HumanCompatibleAI/baselines
OpenAI Baselines: high-quality implementations of reinforcement learning algorithms
HumanCompatibleAI/carla-autoware
Integration of AutoWare AV software with the CARLA simulator
HumanCompatibleAI/coiltraine
Training framework for conditional imitation learning
HumanCompatibleAI/gym
A toolkit for developing and comparing reinforcement learning algorithms.
HumanCompatibleAI/interactive-behaviour-design-baselines
HumanCompatibleAI/interactive-behaviour-design-basicfetch
HumanCompatibleAI/interactive-behaviour-design-gym
HumanCompatibleAI/malmo
Project Malmo is a platform for Artificial Intelligence experimentation and research built on top of Minecraft. We aim to inspire a new generation of research into challenging new problems presented by this unique environment. --- For installation instructions, scroll down to *Getting Started* below, or visit the project page for more information:
HumanCompatibleAI/scenario_runner
Traffic scenario definition and execution engine