New Release
Scaling Kubernetes to 2,500 nodes
Kubernetes has achieved a significant milestone, now effectively managing 2,500 nodes to enhance cloud infrastructure capabilities.
OpenAI NewsOriginal →
2,472 stories
Kubernetes has achieved a significant milestone, now effectively managing 2,500 nodes to enhance cloud infrastructure capabilities.
OpenAI's new block-sparse GPU kernels promise to revolutionize performance for neural networks in AI applications.
OpenAI introduces L₀ regularization to enhance sparse neural networks, boosting efficiency and reducing overfitting.
OpenAI introduces new examples to enhance the interpretability of AI models for educators and learners alike.
OpenAI unveils a new hierarchical reinforcement learning algorithm that speeds up navigation task solving.
New robotics techniques enable simulated controllers to adapt to real-world changes, enhancing reliability in dynamic environments.
OpenAI's new asymmetric actor-critic model boosts image-based robot learning efficiency by 30%.
New research enhances robotic control through sim-to-real transfer with dynamics randomization.
OpenAI unveils innovative techniques for improving robotic grasping through domain randomization and generative models.
Meta-learning agents showcase superior adaptability in simulated robot wrestling, defeating stronger opponents with rapid learning capabilities.
OpenAI's new self-play method transforms AI training by allowing autonomous skill discovery without predefined environments.
New research uncovers unexpected nonlinear capabilities in deep linear networks, challenging traditional views on model limitations.
OpenAI's new algorithm, LOLA, enhances AI's ability to model and collaborate with other agents in strategic settings.
OpenAI's new model enhances AI learning by integrating opponent strategies for superior decision-making.
OpenAI launches new reinforcement learning baselines, ACKTR and A2C, enhancing efficiency for developers and researchers.
OpenAI's latest breakthrough in Dota 2 showcases AI's rapid advancement from novice to pro through self-play techniques.
An AI bot has triumphed over elite Dota 2 players in 1v1 matches, showcasing its self-learning capabilities.
OpenAI introduces RL-Teacher, an open-source tool for enhancing AI training through human feedback.
OpenAI introduces adaptive noise to improve exploration in reinforcement learning, enhancing performance across various applications.
OpenAI introduces Proximal Policy Optimization, a powerful yet simpler reinforcement learning algorithm for developers and researchers.
New adversarial inputs expose vulnerabilities in neural networks, raising concerns for AI security in critical applications like self-driving cars.
OpenAI's Hindsight Experience Replay promises to revolutionize AI training efficiency by enhancing reinforcement learning.
OpenAI introduces teacher-student curriculum learning models to enhance personalized education and engagement.