skrl.memories.random.rst

Random memory

Basic usage

import torch
from skrl.memories.torch import RandomMemory

# create a random memory object
memory = RandomMemory(memory_size=1000, num_envs=1, replacement=False)

# create tensors in memory
memory.create_tensor(name="states", size=(64, 64, 3), dtype=torch.float32)
memory.create_tensor(name="actions", size=(4,1), dtype=torch.float32)
memory.create_tensor(name="rewards", size=1, dtype=torch.float32)
memory.create_tensor(name="next_states", size=(64, 64, 3), dtype=torch.float32)
memory.create_tensor(name="dones", size=1, dtype=torch.bool)

# add data to the memory
for i in range(500):
    memory.add_samples(states=torch.rand(64, 64, 3).view(-1),
                       actions=torch.rand(4,1).view(-1),
                       rewards=torch.rand(1),
                       next_states=torch.rand(64, 64, 3).view(-1),
                       dones=torch.randint(2, size=(1,)).bool())

# sample a batch of data from the memory
batch = memory.sample(batch_size=32, names=["states", "actions", "rewards", "next_states", "dones"])

API

skrl.memories.torch.random.RandomMemory

__init__

__len__