skrl

import gymnasium as gym from skrl.agents import PPO from skrl.resources.preprocessors import RunningStandardization from skrl.trainers import SequentialTrainer from skrl.memories import RandomMemory # Create environment env = gym.make("CartPole-v1") # Create memory memory = RandomMemory(memory_size=1000, num_envs=1) # Initialize agent agent = PPO( observation_space=env.observation_space, action_space=env.action_space, memory=memory, device="cpu", config={ "rollouts": 200, "learning_epochs": 10, "mini_batches": 4, "discount_factor": 0.99, "learning_rate": 3e-4, } ) # Trainer trainer = SequentialTrainer(agent=agent, environment=env) # Train trainer.train()