Official Implementation for the paper "SR-AIF: Solving Sparse-Reward Robotic Tasks from Pixels with Active Inference and World Models"

Predictive Coding for Decision Transformer (IROS 2024)

Official PyTorch Implementation for Metric Residual Networks for Sample Efficient Goal-Conditioned Reinforcement Learning

Sparse environment for MuJoCo suite (v2 and v3)

腾讯开悟人工智能全球公开赛（智能体算法博弈赛道）四川省三等奖项目 | 强化学习迷宫导航 | 奖励塑形 + Target DQN

ICM+RND 分层新颖信号融合好奇心驱动 PPO 智能体 · 稀疏奖励探索管线（Crafter / Atari Montezuma / MiniGrid）· RTX3060 6GB 显存优化至 ~2.2GB · 144 单元测试 · MIT

This project empirically investigates that Weight Agnostic Neural Networks (WANN) are a promising alternative to traditional Reinforcement Learning in sparse settings, using the capabilities of evolutionary search.