NVlabs / ASPIRE

Framework where coding agents discover and refine robot sensorimotor skills as reusable code in place of gradient-based training.

AI4Finance-Foundation / FinRL

Open-source Python framework for training and backtesting financial trading agents with deep reinforcement learning.

UFund-Me / Qbot

Open-source AI-oriented quantitative investment platform covering data, strategy research, backtesting, simulated and live trading.

pytorch / pytorch

Python library providing GPU-accelerated tensor computation like NumPy, plus a tape-based autograd system for building deep neural networks.