Atari 100K

Atari 100K is a benchmark dataset used to evaluate the performance of reinforcement learning algorithms on a selection of Atari 2600 games, focusing on efficiency and control capabilities.

Papers using Atari 100K (21)

GLAM: Global-Local Variation Awareness in Mamba-based World Model2025 · 1 citesJEDI: Joint Embedding Diffusion World Model for Online Model-Based Reinforcement Learning2026Mixture-of-World Models: Scaling Multi-Task Reinforcement Learning with Modular Latent Dynamics2026Horizon Imagination: Efficient On-Policy Rollout in Diffusion World Models2026From Observations to Events: Event-Aware World Model for Reinforcement Learning2026Learning to Focus: Prioritizing Informative Histories with Structured Attention Mechanisms in Partially Observable Reinforcement Learning2025Guardian: Decoupling Exploration from Safety in Reinforcement Learning2025STORI: A Benchmark and Taxonomy for Stochastic Environments2025DyMoDreamer: World Modeling with Dynamic Modulation2025Performance Asymmetry in Model-Based Reinforcement Learning2025EDELINE: Enhancing Memory in Diffusion-based World Models via Linear-Time Sequence Modeling2025Object-Centric World Models from Few-Shot Annotations for Sample-Efficient Reinforcement Learning2025Decorrelated Soft Actor-Critic for Efficient Deep Reinforcement Learning2025Diffusion for World Modeling: Visual Details Matter in Atari2024 · 15 citesEfficientZero V2: Mastering Discrete and Continuous Control with Limited Data2024Learning to Play Atari in a World of Tokens2024SiT: Symmetry-Invariant Transformers for Generalisation in Reinforcement Learning2024Enhancing Reinforcement Learning Through Guided Search2024Spatial-Aware Decision-Making with Ring Attractors in Reinforcement Learning Systems2024Masked Generative Priors Improve World Models Sequence Modelling Capabilities2024Drama: Mamba-Enabled Model-Based Reinforcement Learning Is Sample and Parameter Efficient2024