Average PPO implementation

.gitignore

@@ -13,6 +13,7 @@ cleanrl/videos/*
 benchmark/**/*.svg
 benchmark/**/*.pkl
 mjkey.txt
+.idea
 # Byte-compiled / optimized / DLL files
 __pycache__/
 *.py[cod]

benchmark/apo.sh

@@ -0,0 +1,28 @@
+export WANDB_ENTITY=openrlbenchmark
+
+poetry install -E "mujoco pybullet"
+python -c "import mujoco_py"
+
+#xvfb-run -a python -m cleanrl_utils.benchmark \
+#    --env-ids Swimmer-v3 \
+#    --command "poetry run python cleanrl/apo_continuous_action.py --track --capture-video --gae-lambda=0.99" \
+#    --num-seeds 3 \
+#    --workers 3
+
+#xvfb-run -a python -m cleanrl_utils.benchmark \
+#    --env-ids HalfCheetah-v3 \
+#    --command "poetry run python cleanrl/apo_continuous_action.py --track --capture-video --gae-lambda=0.9" \
+#    --num-seeds 3 \
+#    --workers 3
+
+#xvfb-run -a python -m cleanrl_utils.benchmark \
+#    --env-ids Ant-v3 \
+#    --command "poetry run python cleanrl/apo_continuous_action.py --track --capture-video --gae-lambda=0.9" \
+#    --num-seeds 3 \
+#    --workers 3
+
+xvfb-run -a python -m cleanrl_utils.benchmark \
+    --env-ids Hopper-v3 \
+    --command "poetry run python cleanrl/apo_continuous_action.py --track --capture-video --gae-lambda=0.99" \
+    --num-seeds 3 \
+    --workers 3

cleanrl/apo_continuous_action.py

@@ -0,0 +1,325 @@
+import argparse
+import os
+import random
+import time
+from distutils.util import strtobool
+
+import gym
+import numpy as np
+import pybullet_envs  # noqa
+import torch
+import torch.nn as nn
+import torch.optim as optim
+from torch.distributions.normal import Normal
+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="weather to capture videos of the agent performances (check out `videos` folder)")
+
+    # Algorithm specific arguments
+    parser.add_argument("--env-id", type=str, default="HalfCheetahBulletEnv-v0",
+        help="the id of the environment")
+    parser.add_argument("--total-timesteps", type=int, default=1_000_000,
+        help="total timesteps of the experiments")
+    parser.add_argument("--learning-rate", type=float, default=3e-4,
+        help="the learning rate of the optimizer")
+    parser.add_argument("--num-envs", type=int, default=1,
+        help="the number of parallel game environments")
+    parser.add_argument("--num-steps", type=int, default=2048,
+        help="the number of steps to run in each environment per policy rollout")
+    parser.add_argument("--anneal-lr", type=lambda x: bool(strtobool(x)), default=False, nargs="?", const=True,
+        help="Toggle learning rate annealing for policy and value networks")
+    parser.add_argument("--gae-lambda", type=float, default=0.95,
+        help="the lambda for the general advantage estimation")
+    parser.add_argument("--num-minibatches", type=int, default=32,
+        help="the number of mini-batches")
+    parser.add_argument("--update-epochs", type=int, default=10,
+        help="the K epochs to update the policy")
+    parser.add_argument("--norm-adv", type=lambda x: bool(strtobool(x)), default=True, nargs="?", const=True,
+        help="Toggles advantages normalization")
+    parser.add_argument("--clip-coef", type=float, default=0.2,
+        help="the surrogate clipping coefficient")
+    parser.add_argument("--ent-coef", type=float, default=0.0,
+        help="coefficient of the entropy")
+    parser.add_argument("--vf-coef", type=float, default=0.5,
+        help="coefficient of the value function")
+    parser.add_argument("--max-grad-norm", type=float, default=0.5,
+        help="the maximum norm for the gradient clipping")
+    parser.add_argument("--target-kl", type=float, default=None,
+        help="the target KL divergence threshold")
+
+    # APO specific arguments
+    parser.add_argument("--tau", type=float, default=0.1,
+        help="exponential moving average coefficient for reward and value rate estimates")
+    parser.add_argument("--value-constraint", type=float, default=1.0,
+        help="lagrangian multiplier, controls the bias in value estimation (average value constraint)")
+    args = parser.parse_args()
+    args.batch_size = int(args.num_envs * args.num_steps)
+    args.minibatch_size = int(args.batch_size // args.num_minibatches)
+    # 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 and idx == 0:
+            env = gym.wrappers.RecordVideo(env, f"videos/{run_name}")
+        env = gym.wrappers.ClipAction(env)
+        env = gym.wrappers.NormalizeObservation(env)
+        env = gym.wrappers.TransformObservation(env, lambda obs: np.clip(obs, -10, 10))
+        # IMPORTANT: in average reward setting gamma is always 1.0
+        env = gym.wrappers.NormalizeReward(env, gamma=1.0)
+        env = gym.wrappers.TransformReward(env, lambda reward: np.clip(reward, -10, 10))
+        env.seed(seed)
+        env.action_space.seed(seed)
+        env.observation_space.seed(seed)
+        return env
+
+    return thunk
+
+
+def layer_init(layer, std=np.sqrt(2), bias_const=0.0):
+    torch.nn.init.orthogonal_(layer.weight, std)
+    torch.nn.init.constant_(layer.bias, bias_const)
+    return layer
+
+
+class Agent(nn.Module):
+    def __init__(self, envs):
+        super().__init__()
+        self.critic = nn.Sequential(
+            layer_init(nn.Linear(np.array(envs.single_observation_space.shape).prod(), 64)),
+            nn.Tanh(),
+            layer_init(nn.Linear(64, 64)),
+            nn.Tanh(),
+            layer_init(nn.Linear(64, 1), std=1.0),
+        )
+        self.actor_mean = nn.Sequential(
+            layer_init(nn.Linear(np.array(envs.single_observation_space.shape).prod(), 64)),
+            nn.Tanh(),
+            layer_init(nn.Linear(64, 64)),
+            nn.Tanh(),
+            layer_init(nn.Linear(64, np.prod(envs.single_action_space.shape)), std=0.01),
+        )
+        self.actor_logstd = nn.Parameter(torch.zeros(1, np.prod(envs.single_action_space.shape)))
+
+    def get_value(self, x):
+        return self.critic(x)
+
+    def get_action_and_value(self, x, action=None):
+        action_mean = self.actor_mean(x)
+        action_logstd = self.actor_logstd.expand_as(action_mean)
+        action_std = torch.exp(action_logstd)
+        probs = Normal(action_mean, action_std)
+        if action is None:
+            action = probs.sample()
+        return action, probs.log_prob(action).sum(1), probs.entropy().sum(1), self.critic(x)
+
+
+if __name__ == "__main__":
+    args = parse_args()
+    run_name = f"{args.env_id}__{args.exp_name}__{args.seed}__{int(time.time())}"
+    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 + i, i, args.capture_video, run_name) for i in range(args.num_envs)]
+    )
+    assert isinstance(envs.single_action_space, gym.spaces.Box), "only continuous action space is supported"
+
+    agent = Agent(envs).to(device)
+    optimizer = optim.Adam(agent.parameters(), lr=args.learning_rate, eps=1e-5)
+
+    # ALGO Logic: Storage setup
+    obs = torch.zeros((args.num_steps, args.num_envs) + envs.single_observation_space.shape).to(device)
+    actions = torch.zeros((args.num_steps, args.num_envs) + envs.single_action_space.shape).to(device)
+    logprobs = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    rewards = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    dones = torch.zeros((args.num_steps, args.num_envs)).to(device)
+    values = torch.zeros((args.num_steps, args.num_envs)).to(device)
+
+    # IMPORTANT: reward and value rate estimates, they are crucial for the APO algorithm
+    reward_rate, value_rate = 0.0, 0.0
+
+    # TRY NOT TO MODIFY: start the game
+    global_step = 0
+    start_time = time.time()
+    next_obs = torch.Tensor(envs.reset()).to(device)
+    next_done = torch.zeros(args.num_envs).to(device)
+    num_updates = args.total_timesteps // args.batch_size
+
+    for update in range(1, num_updates + 1):
+        # Annealing the rate if instructed to do so.
+        if args.anneal_lr:
+            frac = 1.0 - (update - 1.0) / num_updates
+            lrnow = frac * args.learning_rate
+            optimizer.param_groups[0]["lr"] = lrnow
+
+        for step in range(0, args.num_steps):
+            global_step += 1 * args.num_envs
+            obs[step] = next_obs
+            dones[step] = next_done
+
+            # ALGO LOGIC: action logic
+            with torch.no_grad():
+                action, logprob, _, value = agent.get_action_and_value(next_obs)
+                values[step] = value.flatten()
+            actions[step] = action
+            logprobs[step] = logprob
+
+            # TRY NOT TO MODIFY: execute the game and log data.
+            next_obs, reward, done, info = envs.step(action.cpu().numpy())
+            rewards[step] = torch.tensor(reward).to(device).view(-1)
+            next_obs, next_done = torch.Tensor(next_obs).to(device), torch.Tensor(done).to(device)
+
+            for item in info:
+                if "episode" in item.keys():
+                    print(f"global_step={global_step}, episodic_return={item['episode']['r']}")
+                    writer.add_scalar("charts/episodic_return", item["episode"]["r"], global_step)
+                    writer.add_scalar("charts/episodic_length", item["episode"]["l"], global_step)
+                    writer.add_scalar("charts/episodic_mean_reward", item["episode"]["r"] / item["episode"]["l"], global_step)
+                    break
+
+        # ALGO LOGIC: update reward and value rate estimates
+        mean_batch_value = values.mean().item()
+        reward_rate = (1 - args.tau) * reward_rate + args.tau * rewards.mean().item()
+        value_rate = (1 - args.tau) * value_rate + args.tau * mean_batch_value
+
+        # bootstrap value if not done
+        with torch.no_grad():
+            next_value = agent.get_value(next_obs).reshape(1, -1)
+            advantages = torch.zeros_like(rewards).to(device)
+            lastgaelam = 0
+            for t in reversed(range(args.num_steps)):
+                if t == args.num_steps - 1:
+                    nextnonterminal = 1.0 - next_done
+                    nextvalues = next_value
+                else:
+                    nextnonterminal = 1.0 - dones[t + 1]
+                    nextvalues = values[t + 1]
+                # IMPORTANT: GAE for average reward value function, for formulas see APO paper
+                delta = rewards[t] - reward_rate + nextnonterminal * nextvalues - values[t]
+                advantages[t] = lastgaelam = delta + args.gae_lambda * nextnonterminal * lastgaelam
+            # A = R - V => R = A + V
+            returns = advantages + values
+
+        # flatten the batch
+        b_obs = obs.reshape((-1,) + envs.single_observation_space.shape)
+        b_logprobs = logprobs.reshape(-1)
+        b_actions = actions.reshape((-1,) + envs.single_action_space.shape)
+        b_advantages = advantages.reshape(-1)
+        b_returns = returns.reshape(-1)
+        b_values = values.reshape(-1)
+
+        # Optimizing the policy and value network
+        b_inds = np.arange(args.batch_size)
+        clipfracs = []
+        for epoch in range(args.update_epochs):
+            np.random.shuffle(b_inds)
+            for start in range(0, args.batch_size, args.minibatch_size):
+                end = start + args.minibatch_size
+                mb_inds = b_inds[start:end]
+
+                _, newlogprob, entropy, newvalue = agent.get_action_and_value(b_obs[mb_inds], b_actions[mb_inds])
+                logratio = newlogprob - b_logprobs[mb_inds]
+                ratio = logratio.exp()
+
+                with torch.no_grad():
+                    # calculate approx_kl http://joschu.net/blog/kl-approx.html
+                    old_approx_kl = (-logratio).mean()
+                    approx_kl = ((ratio - 1) - logratio).mean()
+                    clipfracs += [((ratio - 1.0).abs() > args.clip_coef).float().mean().item()]
+
+                mb_advantages = b_advantages[mb_inds]
+                if args.norm_adv:
+                    mb_advantages = (mb_advantages - mb_advantages.mean()) / (mb_advantages.std() + 1e-8)
+
+                # Policy loss
+                pg_loss1 = -mb_advantages * ratio
+                pg_loss2 = -mb_advantages * torch.clamp(ratio, 1 - args.clip_coef, 1 + args.clip_coef)
+                pg_loss = torch.max(pg_loss1, pg_loss2).mean()
+
+                # Value loss
+                newvalue = newvalue.view(-1)
+                # IMPORTANT: average value constraint as it is called in the paper
+                target_return = b_returns[mb_inds] - args.value_constraint * value_rate
+                v_loss = 0.5 * ((newvalue - target_return) ** 2).mean()
+
+                entropy_loss = entropy.mean()
+                loss = pg_loss - args.ent_coef * entropy_loss + v_loss * args.vf_coef
+
+                optimizer.zero_grad()
+                loss.backward()
+                nn.utils.clip_grad_norm_(agent.parameters(), args.max_grad_norm)
+                optimizer.step()
+
+            if args.target_kl is not None:
+                if approx_kl > args.target_kl:
+                    break
+
+        y_pred, y_true = b_values.cpu().numpy(), b_returns.cpu().numpy()
+        var_y = np.var(y_true)
+        explained_var = np.nan if var_y == 0 else 1 - np.var(y_true - y_pred) / var_y
+
+        # TRY NOT TO MODIFY: record rewards for plotting purposes
+        writer.add_scalar("charts/learning_rate", optimizer.param_groups[0]["lr"], global_step)
+        writer.add_scalar("losses/value_loss", v_loss.item(), global_step)
+        writer.add_scalar("losses/policy_loss", pg_loss.item(), global_step)
+        writer.add_scalar("losses/entropy", entropy_loss.item(), global_step)
+        writer.add_scalar("losses/old_approx_kl", old_approx_kl.item(), global_step)
+        writer.add_scalar("losses/approx_kl", approx_kl.item(), global_step)
+        writer.add_scalar("losses/clipfrac", np.mean(clipfracs), global_step)
+        writer.add_scalar("losses/explained_variance", explained_var, global_step)
+        print("SPS:", int(global_step / (time.time() - start_time)))
+        writer.add_scalar("charts/SPS", int(global_step / (time.time() - start_time)), global_step)
+        # APO specific metrics
+        writer.add_scalar("losses/reward_rate", reward_rate, global_step)
+        writer.add_scalar("losses/value_rate", value_rate, global_step)
+        writer.add_scalar("losses/mean_batch_value", mean_batch_value, global_step)
+
+    envs.close()
+    writer.close()