Craftax

Craftax is a benchmark dataset used to evaluate goal-conditioned reinforcement learning agents by enabling them to learn from all transitions in an environment with respect to every possible goal.

Papers using Craftax (10)

Goal-Conditioned Agents that Learn Everything All at Once2026SCALAR: Learning and Composing Skills through LLM Guided Symbolic Planning and Deep RL Grounding2026Unsupervised Hierarchical Skill Discovery2026Learning Instruction-Following Policies through Open-Ended Instruction Relabeling with Large Language Models2025Nicewebrl: A Python Library For Human Subject Experiments With Reinforcement Learning Environments2025Simplifying Deep Temporal Difference Learning2024 · 1 citesAdam on Local Time: Addressing Nonstationarity in RL with Relative Adam Timesteps2024NiceWebRL: a Python library for human subject experiments with reinforcement learning environments2025CODE-SHARP: Continuous Open-ended Discovery and Evolution of Skills as Hierarchical Reward Programs2026Overcoming Valid Action Suppression in Unmasked Policy Gradient Algorithms2026