Software
Supported Products
REVIVE is the new generation of general intelligence decision-making system for industry experts created by Polixir which transforms the complicated decision-making process into an easy-to-use workflow, to provide users with complete decision-making services and make decisions greater and easier.
GFSEncoder is a pre-trained model that efficiently and stably codes feedback control systems so that they can be represented by a vector encoding. It is a pre-trained model obtained by training on two optimization targets (discriminative target and consistency target) on 60 million interactive data collected from more than 70,000 feedback control tasks.
Codes
The implementation of ICLR-2023 paper "Discovering Generalizable Multi-agent Coordination Skills from Multi-task Offline Data".
Author's PyTorch implementation of ICML'23 paper "Policy Regularization with Dataset Constraint for Offline Reinforcement Learning" for D4RL gym and AntMaze tasks.
The implementation of AAAI'24 paper "Generalizable Task Representation Learning for Offline Meta-Reinforcement Learning with Data Limitations".
Official code for ACT: Empowering Decision Transformer with Dynamic Programming via Advantage Conditioning (AAAI'24)
A collection of offline reinforcement learning algorithms.
python package of Zeroth-Order Optimization (ZOOpt)
RLA is a tool for managing your RL experiments automatically
D3PE (Deep Data-Driven Policy Evaluation) aims to evaluation a large set of candidate policies from a fixed dataset to select best ones.
Benchmarks
Python interface for accessing the near real-world offline reinforcement learning (NeoRL) benchmark datasets