techtimes

Cognition SWE-1.7: RL on Top of RL Yields Near-Frontier Code at Low Cost

Cognition launched SWE-1.7 on Wednesday, the most capable model it has trained, and with it made a specific claim about reinforcement learning that the broader AI research community will be evaluating ...
GitHub

Reinforcement Learning-Constrained Truncated Diffusion Modeling in End-to-End Autonomous Driving

Generative diffusion models for end-to-end autonomous driving often suffer from mode collapse, tending to generate conservative and homogeneous behaviors. While DiffusionDrive employs predefined ...
GitHub

ReinFlow: Fine-tuning Flow Matching Policy with Online Reinforcement Learning

To fully reproduce our experiments, please refer to ReproduceExps.md. To download our training data and reproduce the plots in the paper, please refer to ...
Interesting Engineering

Watch: Korean humanoid robot performs viral K-POP dance, learns by watching videos

While China dominates humanoid robotics headlines, a Korean firm showcased a humanoid learning complex motions through an open-source AI framework. In a recent demonstration, ROBOTIS’ AI Sapiens ...