deepqlearn 0.1.1

A simple Deep Q-Network implementation in PyTorch

Got it — you want one single markdown block containing the entire README from title to author so you can copy it in one click.

Here’s your complete README:

# DeepQlearn

A simple and clean **Deep Q-Network (DQN)** implementation in PyTorch, packaged for easy installation via PyPI.

---

## ✨ Features
- Minimal, readable code for learning DQN fundamentals
- Works with any [Gym](https://www.gymlibrary.dev/) environment
- Epsilon-greedy exploration strategy
- Replay buffer for experience sampling
- Target network updates for stable training
- Implemented in [PyTorch](https://pytorch.org)

---

## 📦 Installation

```bash
pip install deepqlearn

Requirements:

Python >= 3.8
PyTorch
Gym

---

🚀 Quick Start

from deepqlearn import train_dqn

# Train on CartPole for 200 episodes
train_dqn(episodes=200)

---

📚 API Reference

train_dqn(episodes=500, batch_size=64, target_update=10)

Train a DQN agent on CartPole-v1.

Parameters:

• episodes (int) → Number of episodes to train for
• batch_size (int) → Mini-batch size for replay
• target_update (int) → Frequency (in episodes) to update the target network

---

DQNAgent

A reinforcement learning agent using Deep Q-Learning.

Constructor:

DQNAgent(
    state_dim,
    action_dim,
    lr=1e-3,
    gamma=0.99,
    epsilon=1.0,
    epsilon_min=0.01,
    epsilon_decay=0.995
)

Key Methods:

• act(state) → Choose an action given a state
• remember(state, action, reward, next_state, done) → Store experience in replay buffer
• replay(batch_size) → Train from replay buffer
• update_target() → Update target network with policy weights

---

📝 Example Training Script

import gym
from deepqlearn import DQNAgent

env = gym.make("CartPole-v1")
agent = DQNAgent(env.observation_space.shape[0], env.action_space.n)

for episode in range(100):
    state = env.reset()[0]
    done = False
    total_reward = 0
    while not done:
        action = agent.act(state)
        next_state, reward, done, _, _ = env.step(action)
        agent.remember(state, action, reward, next_state, done)
        agent.replay(batch_size=64)
        state = next_state
        total_reward += reward
    agent.update_target()
    print(f"Episode {episode} — Reward: {total_reward}")

env.close()

---

📂 Project Structure

deepqlearn/
│
├── src/
│   └── deepqlearn/
│       ├── __init__.py
│       ├── agent.py
│       ├── network.py
│       ├── replay_buffer.py
│       ├── train.py
│
├── pyproject.toml
├── setup.cfg
├── README.md
├── LICENSE
└── .gitignore

---

⚖ License

MIT License

---

👤 Author

Prathamesh Jadhav