AIbase
EN
Home
Decision Transformer Gym Halfcheetah Expert
D

Decision Transformer Gym Halfcheetah Expert

Developed by edbeeching
This is a trained decision transformer model, with training data sampled from expert trajectories in the Gym HalfCheetah environment.
Physics Model
Transformers
#Reinforcement Learning Control#Expert Trajectory Imitation#Continuous Action Decision Making
Downloads 98
Release Time : 3/16/2022

Model Overview

This model uses the Decision Transformer architecture, specifically designed for continuous control tasks in the Gym HalfCheetah environment, capable of predicting optimal actions based on environmental states.

Model Features

Expert Trajectory Training
The model is trained using expert trajectories sampled from the Gym HalfCheetah environment, enabling it to learn expert strategies.
State Normalization
Provides detailed normalization coefficients (mean and standard deviation) to ensure proper processing of input states.
Sequential Decision Making
Based on the Transformer architecture, it can handle sequential decision-making problems, taking into account historical state information.

Model Capabilities

Continuous Action Space Control
Reinforcement Learning Strategy Prediction
Robot Control Simulation

Use Cases

Robot Control
HalfCheetah Simulation Control
Achieves efficient motion control in the Gym HalfCheetah environment
Capable of learning expert strategies to achieve control effects similar to experts
Reinforcement Learning Research
Decision Transformer Application Research
Serves as a case study for the application of Decision Transformer in continuous control tasks
Validates the effectiveness of Decision Transformer in Mujoco environments
Featured Recommended AI Models
AIbase
Empowering the Future, Your AI Solution Knowledge Base
English简体中文繁體中文にほんご
© 2025AIbase