InternVL offers open-source multimodal large language models combining vision transformers and LLMs, featuring CascadeRL training and competitive benchmarks like GPT-4o.
GMR is a Python library that retargets human motion from multiple formats onto 17+ humanoid robots in real time on CPU, tuned for RL tracking policies and whole-body teleoperation.
Graph-R1 trains large language models with reinforcement learning to reason over knowledge graphs, cycling through think-query-retrieve-rethink steps for complex knowledge tasks.
GS-Playground combines 3D Gaussian Splatting rendering with a velocity-impulse physics engine to enable large-scale visual reinforcement learning at up to 10^4 FPS. Preview release with core simulation API and demos.
ML-From-Scratch offers bare-bones Python implementations of key machine learning algorithms using only NumPy, focusing on transparency over efficiency. Explore how it demystifies ML fundamentals.
UI-Voyager is a 4B parameter AI agent achieving 81% success on AndroidWorld by self-evolving with SSIM-based trajectory correction, no human labels needed.
motion_tracking_controller is a C++ ROS 2 package deploying RL-trained motion tracking policies on legged robots with ONNX inference and embedded robot control metadata.
FinRL-Trading offers a modular Python framework for quantitative trading focused on a weight-centric architecture unifying backtesting and live execution, with classical and DRL portfolio methods.
Alibaba’s VRAG models reasoning as a dynamic DAG with multimodal memory and RL-based fine-grained credit assignment, supporting text, image, and video retrieval in a unified framework.
TradeMaster offers a full pipeline for RL-based quantitative trading with 13+ algorithms and a rigorous 6-axis, 17-measure evaluation framework across multiple asset classes and trading tasks.
FinRL provides an open-source three-layer architecture for financial reinforcement learning with 5 DRL agents and 14+ data sources. Great for learning DRL in finance.
ToddlerBot offers a full Python stack for training, classifying, and deploying multi-skill humanoid locomotion policies using stereo depth data and reinforcement learning.
