Description
Building on the recent successes of distributed training of RL agents, R2D2 is an RL approach that trains a RNN-based RL agents from distributed prioritized experience replay. Using a single network architecture and fixed set of hyperparameters, Recurrent Replay Distributed DQN quadrupled the previous state of the art on Atari-57, and matches the state of the art on DMLab-30. It was the first agent to exceed human-level performance in 52 of the 57 Atari games.
Papers Using This Method
The R2D2 Deep Neural Network Series for Scalable Non-Cartesian Magnetic Resonance Imaging2025-03-12Towards a robust R2D2 paradigm for radio-interferometric imaging: revisiting DNN training and architecture2025-03-04S-R2D2: a spherical extension of the R2D2 deep neural network series paradigm for wide-field radio-interferometric imaging2025-03-03R2D2: Remembering, Reflecting and Dynamic Decision Making for Web Agents2025-01-21A Deep-Based Approach for Multi-Descriptor Feature Extraction: Applications on SAR Image Registration2024-11-05Latent Adversarial Training Improves Robustness to Persistent Harmful Behaviors in LLMs2024-07-22Does Refusal Training in LLMs Generalize to the Past Tense?2024-07-16Simplifying Deep Temporal Difference Learning2024-07-05Jailbreaking Leading Safety-Aligned LLMs with Simple Adaptive Attacks2024-04-02Scalable Non-Cartesian Magnetic Resonance Imaging with R2D22024-03-26R2D2 image reconstruction with model uncertainty quantification in radio astronomy2024-03-26The R2D2 deep neural network series paradigm for fast precision imaging in radio astronomy2024-03-08CLEANing Cygnus A deep and fast with R2D22023-09-06Exploring the Promise and Limits of Real-Time Recurrent Learning2023-05-30HPointLoc: Point-based Indoor Place Recognition using Synthetic RGB-D Images2022-12-30Meta-Referential Games to Learn Compositional Learning Behaviours2022-07-16R2D2: Robust Data-to-Text with Replacement Detection2022-05-25CCMB: A Large-scale Chinese Cross-modal Benchmark2022-05-08Semantic Exploration from Language Abstractions and Pretrained Representations2022-04-08Fast-R2D2: A Pretrained Recursive Neural Network based on Pruned CKY for Grammar Induction and Text Representation2022-03-01