dqn_atari_e.py

# docs and experiment results can be found at https://docs.cleanrl.dev/rl-algorithms/dqn/#dqn_ataripy
import argparse
import os
import random
import time
from distutils.util import strtobool

import gym
import numpy as np
import torch
import torch.nn as nn
import torch.nn.functional as F
import torch.optim as optim
from stable_baselines3.common.atari_wrappers import (
    ClipRewardEnv,
    EpisodicLifeEnv,
    FireResetEnv,
    MaxAndSkipEnv,
    NoopResetEnv,
)
from stable_baselines3.common.buffers import ReplayBuffer
from torch.utils.tensorboard import SummaryWriter


def parse_args():
    # fmt: off
    parser = argparse.ArgumentParser()
    parser.add_argument("--exp-name", type=str, default=os.path.basename(__file__).rstrip(".py"),
        help="the name of this experiment")
    parser.add_argument("--seed", type=int, default=1,
        help="seed of the experiment")
    parser.add_argument("--torch-deterministic", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
        help="if toggled, `torch.backends.cudnn.deterministic=False`")
    parser.add_argument("--cuda", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
        help="if toggled, cuda will be enabled by default")
    parser.add_argument("--track", type=lambda x: bool(strtobool(x)), default=False, nargs="?", const=True,
        help="if toggled, this experiment will be tracked with Weights and Biases")
    parser.add_argument("--wandb-project-name", type=str, default="cleanRL",
        help="the wandb's project name")
    parser.add_argument("--wandb-entity", type=str, default=None,
        help="the entity (team) of wandb's project")
    parser.add_argument("--capture-video", type=lambda x: bool(strtobool(x)), default=False, nargs="?", const=True,
        help="whether to capture videos of the agent performances (check out `videos` folder)")
    parser.add_argument("--save-model", type=lambda x: bool(strtobool(x)), default=False, nargs="?", const=True,
        help="whether to save model into the `runs/{run_name}` folder")
    parser.add_argument("--upload-model", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
        help="whether to upload the saved model to huggingface")
    parser.add_argument("--hf-entity", type=str, default="mallycrip",
        help="the user or org name of the model repository from the Hugging Face Hub")

    # Algorithm specific arguments
    parser.add_argument("--env-id", type=str, default="BreakoutNoFrameskip-v4",
        help="the id of the environment")
    parser.add_argument("--total-timesteps", type=int, default=10000000,
        help="total timesteps of the experiments")
    parser.add_argument("--learning-rate", type=float, default=1e-4,
        help="the learning rate of the optimizer")
    parser.add_argument("--buffer-size", type=int, default=100000,
        help="the replay memory buffer size")
    parser.add_argument("--gamma", type=float, default=0.99,
        help="the discount factor gamma")
    parser.add_argument("--tau", type=float, default=1.,
        help="the target network update rate")
    parser.add_argument("--target-network-frequency", type=int, default=1000,
        help="the timesteps it takes to update the target network")
    parser.add_argument("--batch-size", type=int, default=32,
        help="the batch size of sample from the reply memory")
    parser.add_argument("--start-e", type=float, default=1,
        help="the starting epsilon for exploration")
    parser.add_argument("--end-e", type=float, default=0.01,
        help="the ending epsilon for exploration")
    parser.add_argument("--exploration-fraction", type=float, default=0.10,
        help="the fraction of `total-timesteps` it takes from start-e to go end-e")
    parser.add_argument("--learning-starts", type=int, default=80000,
        help="timestep to start learning")
    parser.add_argument("--train-frequency", type=int, default=4,
        help="the frequency of training")
    args = parser.parse_args()
    # fmt: on
    return args


def make_env(env_id, seed, idx, capture_video, run_name):
    def thunk():
        env = gym.make(env_id)
        env = gym.wrappers.RecordEpisodeStatistics(env)
        if capture_video:
            if idx == 0:
                env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
        env = NoopResetEnv(env, noop_max=30)
        env = MaxAndSkipEnv(env, skip=4)
        env = EpisodicLifeEnv(env)
        if "FIRE" in env.unwrapped.get_action_meanings():
            env = FireResetEnv(env)
        env = ClipRewardEnv(env)
        env = gym.wrappers.ResizeObservation(env, (84, 84))
        env = gym.wrappers.GrayScaleObservation(env)
        env = gym.wrappers.FrameStack(env, 4)
        env.seed(seed)
        env.action_space.seed(seed)
        env.observation_space.seed(seed)
        return env

    return thunk


# ALGO LOGIC: initialize agent here:
class QNetwork(nn.Module):
    def __init__(self, env):
        super().__init__()
        self.network = nn.Sequential(
            nn.Conv2d(4, 32, 8, stride=4),
            nn.ReLU(),
            nn.Conv2d(32, 64, 4, stride=2),
            nn.ReLU(),
            nn.Conv2d(64, 64, 3, stride=1),
            nn.ReLU(),
            nn.Flatten(),
            nn.Linear(3136, 512),
            nn.ReLU(),
            nn.Linear(512, env.single_action_space.n),
        )

    def forward(self, x):
        return self.network(x / 255.0)


def linear_schedule(start_e: float, end_e: float, duration: int, t: int):
    slope = (end_e - start_e) / duration
    return max(slope * t + start_e, end_e)


if __name__ == "__main__":
    args = parse_args()
    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"

    run_name = f"SpaceInvadersNoFrameskip-v4__dqn_atari__1__1675730632"
    if args.track:
        import wandb

        wandb.init(
            project=args.wandb_project_name,
            entity=args.wandb_entity,
            sync_tensorboard=True,
            config=vars(args),
            name=run_name,
            monitor_gym=True,
            save_code=True,
        )
    writer = SummaryWriter(f"runs/{run_name}")
    writer.add_text(
        "hyperparameters",
        "|param|value|\n|-|-|\n%s" % ("\n".join([f"|{key}|{value}|" for key, value in vars(args).items()])),
    )

    # TRY NOT TO MODIFY: seeding
    random.seed(args.seed)
    np.random.seed(args.seed)
    torch.manual_seed(args.seed)
    torch.backends.cudnn.deterministic = args.torch_deterministic

    device = torch.device("cuda" if torch.cuda.is_available() and args.cuda else "cpu")

    # env setup
    envs = gym.vector.SyncVectorEnv([make_env(args.env_id, args.seed, 0, args.capture_video, run_name)])
    assert isinstance(envs.single_action_space, gym.spaces.Discrete), "only discrete action space is supported"

    q_network = QNetwork(envs).to(device)
    optimizer = optim.Adam(q_network.parameters(), lr=args.learning_rate)
    target_network = QNetwork(envs).to(device)
    target_network.load_state_dict(q_network.state_dict())

    rb = ReplayBuffer(
        args.buffer_size,
        envs.single_observation_space,
        envs.single_action_space,
        device,
        optimize_memory_usage=True,
        handle_timeout_termination=True,
    )

    model_path = f"../runs/{run_name}/dqn_atari.cleanrl_model"
    print(f"model saved to {model_path}")
    from evals.dqn_eval import evaluate

    episodic_returns = evaluate(
        model_path,
        make_env,
        args.env_id,
        eval_episodes=10,
        run_name=f"{run_name}-eval",
        Model=QNetwork,
        device=device,
        epsilon=0.05,
    )
    for idx, episodic_return in enumerate(episodic_returns):
        writer.add_scalar("eval/episodic_return", episodic_return, idx)

    if args.upload_model:
        from huggingface import push_to_hub

        repo_name = f"{args.env_id}-{args.exp_name}-2"
        repo_id = f"{args.hf_entity}/{repo_name}" if args.hf_entity else repo_name
        push_to_hub(args, episodic_returns, repo_id, "DQN", f"runs/{run_name}", f"videos/{run_name}-eval")

    envs.close()
    writer.close()