How AI
Training Scales
How AI Training Scales
We’ve discovered that the gradient noise scale, a simple statistical metric, predicts the parallelizability of neural network training on a wide range of tasks.
How AI Training Scales
Research
OpenAI conducts fundamental, long-term research toward the creation of safe AGI.
How AI
Learning
Dexterity
Learning Dexterity
We’ve trained a human-like robot hand to manipulate physical objects with unprecedented dexterity.
Learning Dexterity
Improving Language Understanding
with Unsupervised Learning
Improving Language Understanding with Unsupervised Learning
We’ve obtained state-of-the-art results on a suite of diverse language tasks with a scalable, task-agnostic system, which we’re also releasing.
Improving Language Understanding
with Unsupervised Learning
with Unsupervised Learning
OpenAI Charter
OpenAI Charter
We’re releasing a charter that describes the principles we use to execute on OpenAI’s mission.
OpenAI Charter
Retro Contest
Retro Contest
We’re launching a transfer learning contest that measures a reinforcement learning algorithm’s ability to generalize from previous experience.
Retro Contest
Ingredients for
Robotics Research
Ingredients for Robotics Research
We’re releasing eight simulated robotics environments and a Baselines implementation of Hindsight Experience Replay, all developed for our research over the past year.
Ingredients for Robotics Research
Block-Sparse GPU Kernels
Block-Sparse GPU Kernels
We’re releasing highly-optimized GPU kernels for an underexplored class of neural network architectures: networks with block-sparse weights.
Block-Sparse GPU Kernels
Competitive Self-Play
Competitive Self-Play
We’ve found that self-play allows simulated AIs to discover physical skills like tackling, ducking, faking, kicking, catching, and diving for the ball, without explicitly designing an environment with these skills in mind.
Competitive Self-Play
OpenAI Baselines:
ACKTR & A2C
OpenAI Baselines: ACKTR & A2C
We’re releasing two new OpenAI Baselines implementations: ACKTR and A2C. A2C is a synchronous, deterministic variant of Asynchronous Advantage Actor Critic (A3C) which we’ve found gives equal performance.
OpenAI Baselines: ACKTR & A2C
Better Exploration with Parameter Noise
Better Exploration with Parameter Noise
We’ve found that adding adaptive noise to the parameters of reinforcement learning algorithms frequently boosts performance.
Better Exploration with Parameter Noise
Proximal Policy Optimization
Proximal Policy Optimization
We’re releasing a new class of reinforcement learning algorithms, Proximal Policy Optimization (PPO), which perform comparably or better than state-of-the-art approaches while being much simpler to implement and tune.
Proximal Policy Optimization
OpenAI Baselines: DQN
OpenAI Baselines
We’re open-sourcing OpenAI Baselines, our internal effort to reproduce reinforcement learning algorithms with performance on par with published results.
OpenAI Baselines
Robots that Learn
Robots that Learn
We’ve created a robotics system, trained entirely in simulation and deployed on a physical robot, which can learn a new task after seeing it done once.
Robots that Learn
Unsupervised Sentiment Neuron
Unsupervised
Sentiment Neuron
We’ve developed an unsupervised system which learns an excellent representation of sentiment, despite being trained only to predict the next character in the text of Amazon reviews.
Unsupervised
Sentiment Neuron
Sentiment Neuron
Evolution
Strategies
Evolution Strategies
We've discovered that evolution strategies (ES), an optimization technique that's been known for decades, rivals the performance of standard reinforcement learning (RL) techniques on modern RL benchmarks (e.g. Atari/MuJoCo), while overcoming many of RL's inconveniences.
Evolution Strategies
Generative Models
Generative Models
Four projects that share a common theme of enhancing or using generative models.
Generative Models
Research by OpenAI
We regularly publish our research, which focuses on reinforcement learning, robotics, and generative models.