SamanPandey's picture
Completed Precision Irrigation RL Agent
9b50e1b
Raw History Blame Contribute Delete
3.24 kB
"""
scripts/train.py โ€” Phase 3 RL training scaffold
-------------------------------------------------
Trains PPO on PrecisionIrrigation-v0 using Stable-Baselines3.
Usage:
python scripts/train.py
python scripts/train.py --timesteps 200000 --algo dqn
Requires:
pip install stable-baselines3[extra] tensorboard
"""
import sys
import os
import argparse
sys.path.insert(0, os.path.join(os.path.dirname(__file__), ".."))
try:
import gymnasium as gym
import envs # noqa: F401 โ€” triggers registration
except ImportError:
print("โŒ gymnasium not installed. Run: pip install gymnasium")
sys.exit(1)
try:
from stable_baselines3 import PPO, DQN
from stable_baselines3.common.env_util import make_vec_env
from stable_baselines3.common.evaluation import evaluate_policy
from stable_baselines3.common.callbacks import EvalCallback, StopTrainingOnRewardThreshold
except ImportError:
print("โŒ stable-baselines3 not installed. Run: pip install stable-baselines3[extra]")
sys.exit(1)
ALGO_MAP = {"ppo": PPO, "dqn": DQN}
MODELS_DIR = os.path.join(os.path.dirname(__file__), "..", "models")
LOGS_DIR = os.path.join(os.path.dirname(__file__), "..", "logs")
os.makedirs(MODELS_DIR, exist_ok=True)
os.makedirs(LOGS_DIR, exist_ok=True)
def train(algo_name: str = "ppo", total_timesteps: int = 100_000, n_envs: int = 4):
AlgoCls = ALGO_MAP[algo_name.lower()]
print(f"\n๐Ÿšœ Training {algo_name.upper()} on PrecisionIrrigation-v0")
print(f" timesteps={total_timesteps:,} parallel_envs={n_envs}\n")
# Vectorised environments for faster training
vec_env = make_vec_env("PrecisionIrrigation-v0", n_envs=n_envs, seed=0)
eval_env = gym.make("PrecisionIrrigation-v0")
# Stop early if reward threshold reached
stop_cb = StopTrainingOnRewardThreshold(reward_threshold=25.0, verbose=1)
eval_cb = EvalCallback(
eval_env,
best_model_save_path=MODELS_DIR,
log_path=LOGS_DIR,
eval_freq=max(total_timesteps // 20, 1000),
n_eval_episodes=10,
callback_on_new_best=stop_cb,
verbose=1,
)
model_kwargs: dict = {"verbose": 1, "tensorboard_log": LOGS_DIR}
if algo_name == "ppo":
model_kwargs.update({"n_steps": 256, "batch_size": 64, "learning_rate": 3e-4})
model = AlgoCls("MultiInputPolicy", vec_env, **model_kwargs)
model.learn(total_timesteps=total_timesteps, callback=eval_cb, progress_bar=True)
save_path = os.path.join(MODELS_DIR, f"{algo_name}_irrigation_final")
model.save(save_path)
print(f"\n๐Ÿ’พ Model saved โ†’ {save_path}.zip")
# Final evaluation
mean_r, std_r = evaluate_policy(model, eval_env, n_eval_episodes=20, deterministic=True)
print(f"๐Ÿ“Š Final eval โ€” mean reward: {mean_r:.2f} ยฑ {std_r:.2f}")
vec_env.close()
eval_env.close()
return model
if __name__ == "__main__":
parser = argparse.ArgumentParser()
parser.add_argument("--algo", default="ppo", choices=["ppo", "dqn"])
parser.add_argument("--timesteps", default=100_000, type=int)
parser.add_argument("--n-envs", default=4, type=int)
args = parser.parse_args()
train(args.algo, args.timesteps, args.n_envs)