reward-scope-ai/reward-scope
Real-time reward debugging and hacking detection for reinforcement learning
When training reinforcement learning agents, it often looks like the agent is improving because its reward score is rising, but its actual behavior is broken. This tool helps you catch these 'reward hacking' issues by monitoring your agent's training in real time. It takes your ongoing training data as input and provides a live dashboard and alerts, showing you exactly how your agent is learning and flagging problematic exploitation patterns. This is for machine learning researchers, engineers, and practitioners working with reinforcement learning models who need to ensure their agents learn desirable behaviors.
Use this if you are training reinforcement learning agents and need to detect when they are exploiting the reward function in unintended ways, rather than genuinely learning the desired task.
Not ideal if you are working with supervised or unsupervised learning models, as its features are specifically designed for reinforcement learning training analysis.
Stars
18
Forks
2
Language
Python
License
MIT
Category
Last pushed
Dec 29, 2025
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/reward-scope-ai/reward-scope"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
DLR-RM/stable-baselines3
PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.
google-deepmind/dm_control
Google DeepMind's software stack for physics-based simulation and Reinforcement Learning...
Denys88/rl_games
RL implementations
pytorch/rl
A modular, primitive-first, python-first PyTorch library for Reinforcement Learning.
yandexdataschool/Practical_RL
A course in reinforcement learning in the wild