Metadata-Version: 2.4
Name: aditya-epsgreedy-20251112
Version: 0.0.1
Summary: A tiny, well-tested epsilon-greedy action selector for Reinforcement Learning with action masking. Created by Aditya Singh for B.Tech AI (RL Practicum, NMIMS MPSTME).
Project-URL: Homepage, https://pypi.org/project/aditya-epsgreedy-20251112/
Project-URL: Repository, https://github.com/salmoneatenbybear/aditya-epsgreedy-20251112
Author-email: Aditya Singh <salmoneatenbybear@gmail.com>
License: MIT
License-File: LICENSE
Keywords: RL,action-selection,aditya-singh,ai,bandit,epsilon-greedy,reinforcement-learning
Classifier: Development Status :: 4 - Beta
Classifier: Intended Audience :: Education
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering :: Artificial Intelligence
Requires-Python: >=3.8
Description-Content-Type: text/markdown

# Epsilon-Greedy

A minimal, tested epsilon-greedy action selector for Reinforcement Learning.
Includes action masking and tie-aware probability allocation.

## Usage

```python
from epsgreedy import epsilon_greedy

action_values = [1.2, 0.5, 1.2, -0.3]   # Q(s, ·)
# Mask out invalid actions (e.g., action 1 is invalid here)
invalid = [False, True, False, False]

a, p = epsilon_greedy(action_values, epsilon=0.1, invalid=invalid)
print(a, p)  # chosen action index, probabilities over actions
```

## Install
```bash
pip install aditya-epsgreedy-20251112
```

## Testing
```bash
pip install pytest
pytest -q
```

## License
MIT