File size: 1,953 Bytes
0a5b12e | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 | """
Training script — Train the RL policy for yield optimization.
"""
import logging
import os
import sys
import json
sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
from agent.main import YieldRouterAgent, setup_logging
from agent.rl_optimizer import PPOYieldOptimizer, RWAYieldEnv, Backtester
def main():
setup_logging()
logger = logging.getLogger("train")
logger.info("=" * 60)
logger.info("Dynamic RWA Yield Router — RL Training")
logger.info("=" * 60)
# Initialize optimizer
total_timesteps = int(os.getenv("TRAIN_STEPS", "100000"))
optimizer = PPOYieldOptimizer(
total_timesteps=total_timesteps,
learning_rate=3e-4,
n_steps=2048,
batch_size=64,
)
# Train
logger.info(f"\n🎓 Training for {total_timesteps} timesteps...\n")
optimizer.train(total_timesteps=total_timesteps)
# Backtest
logger.info("\n📊 Running backtest...\n")
env = RWAYieldEnv(episode_length=720)
backtester = Backtester(optimizer, env)
results = backtester.run_backtest(n_episodes=10)
if results.get("rl_agent"):
import numpy as np
returns = [r["total_return"] for r in results["rl_agent"]]
sharpes = [r["sharpe"] for r in results["rl_agent"]]
drawdowns = [r["max_drawdown"] for r in results["rl_agent"]]
print("\n" + "=" * 60)
print("📊 BACKTEST RESULTS (10 episodes)")
print("=" * 60)
print(f" Avg Return: {np.mean(returns):+.2f}%")
print(f" Std Return: {np.std(returns):.2f}%")
print(f" Avg Sharpe: {np.mean(sharpes):.3f}")
print(f" Avg Max DD: {np.mean(drawdowns)*100:.2f}%")
print(f" Best Return: {max(returns):+.2f}%")
print(f" Worst Return: {min(returns):+.2f}%")
logger.info("\n✅ Training complete!")
if __name__ == "__main__":
main()
|