✷ The archive · 11 labs · 3,433 dispatches
The labs
No more opening a dozen official blogs every morning. First-hand releases from OpenAI, Anthropic, DeepMind and the rest, each with its substance pulled out.
OpenAI Research: Transfer of Adversarial Robustness Between Perturbation Types
OpenAI researchers found that adversarial robustness does not consistently transfer between different perturbation types, suggesting that defenses must be evaluated against a diverse range of attack vectors to be truly effective.
OpenAI MuseNet
MuseNet is a deep neural network capable of generating 4-minute musical compositions across 10 instruments by learning patterns of harmony, rhythm, and style from hundreds of thousands of MIDI files.
OpenAI Sparse Transformer
OpenAI introduced the Sparse Transformer, a deep neural network that uses a reformulated attention mechanism to model sequences 30x longer than previous Transformers by reducing algorithmic complexity from O(N^2) to O(N sqrt(N)).
OpenAI Five defeats Dota 2 world champions
OpenAI Five became the first AI to defeat esports world champions in a live livestreamed match, winning two back-to-back games against the Dota 2 team OG.
OpenAI Five Finals Event Announcement
OpenAI announced a final live event for OpenAI Five on April 13, 2019, to demonstrate the competence, scalability, and human-AI interaction capabilities of its Dota 2 AI.
Implicit Generation and Generalization Methods for Energy-Based Models
OpenAI researchers developed stable and scalable training methods for energy-based models (EBMs), achieving high-quality image generation and superior out-of-distribution generalization compared to likelihood-based models.
OpenAI Scholars 2019 Program Participants
OpenAI introduced its 2019 cohort of Scholars, a multidisciplinary group of researchers focusing on reinforcement learning, natural language processing, and generative models.
OpenAI LP: Transition to Capped-Profit Structure
OpenAI announced the creation of OpenAI LP, a hybrid capped-profit company designed to raise billions in capital for compute and talent while ensuring that the primary mission of safe AGI benefits all of humanity.
OpenAI Activation Atlases
OpenAI and Google researchers introduced Activation Atlases, a visualization technique that maps interactions between neurons to reveal how neural networks represent concepts and identify decision-making weaknesses.
Neural MMO: A Massively Multiagent Game Environment
OpenAI has released Neural MMO, a persistent, large-scale environment for reinforcement learning agents that demonstrates how increased population size and species diversity drive exploration and niche formation.
OpenAI Spinning Up in Deep RL Workshop Review
OpenAI hosted a workshop based on its Spinning Up in Deep RL resource package to scale mentorship and skill development in reinforcement learning for a diverse group of participants.
AI Safety Needs Social Scientists
OpenAI argues that long-term AI safety and alignment require the integration of social science to account for human cognitive biases and the complexities of human values.
OpenAI GPT-2 Release and Implications
OpenAI introduced GPT-2, a 1.5 billion parameter unsupervised language model capable of zero-shot task performance and coherent text generation, while implementing a staged release strategy to mitigate misuse risks.
OpenAI Research: Computational Limitations in Robust Classification
OpenAI researchers have identified classification tasks where robust classifiers exist but are computationally impossible to learn, linking the hardness of robust learning to the existence of cryptographic primitives.
OpenAI Summer Fellows 2018 Final Projects
OpenAI's 2018 Summer Fellows completed research projects focusing on reinforcement learning generalization, distributed training patterns, and generative model expressivity.
How AI Training Scales: Predicting Parallelizability via Gradient Noise Scale
OpenAI discovered that the gradient noise scale, a statistical metric quantifying the signal-to-noise ratio of network gradients, can predict the maximum useful batch size for neural network training across diverse tasks.
Quantifying Generalization in Reinforcement Learning
OpenAI introduces CoinRun, a procedurally generated training environment designed to quantify and improve an agent's ability to transfer learned skills to novel situations in reinforcement learning.
OpenAI Spinning Up in Deep RL
OpenAI has released Spinning Up in Deep RL, an educational resource providing code, tutorials, and documentation to help practitioners master deep reinforcement learning.
OpenAI Learning Concepts with Energy Functions
OpenAI has developed a technique using energy functions to enable agents to learn and extract concepts that can be transferred across dissimilar environments without retraining.
OpenAI Plan Online Learn Offline (POLO) Framework
OpenAI introduces the Plan Online Learn Offline (POLO) framework, which combines local model-based control and global value function learning to enable efficient learning in complex simulated control tasks.
OpenAI Random Network Distillation (RND) for Reinforcement Learning
OpenAI introduces Random Network Distillation (RND), a curiosity-driven exploration method that for the first time exceeds average human performance on Montezuma's Revenge without using human demonstrations or emulator state access.
OpenAI Iterated Amplification for Complex Goal Learning
OpenAI proposes iterated amplification, an AI safety technique that enables the specification of complex, beyond-human-scale goals by decomposing tasks into simpler sub-tasks to generate training signals.
OpenAI Scholars 2019 Program Announcement
OpenAI has opened applications for its second cohort of OpenAI Scholars, providing stipends and mentorship to individuals from underrepresented groups to study deep learning and open-source a project.
OpenAI Fellows Winter 2019 and Interns Summer 2019 Programs
OpenAI announced the opening of applications for its Winter 2019 Fellows program and Summer 2019 Internships, designed to integrate researchers from diverse backgrounds and students into its AI research efforts.
FFJORD: Free-form continuous dynamics for scalable reversible generative models
OpenAI introduces FFJORD, a continuous-time invertible generative model that uses Hutchinson's trace estimator to enable unrestricted neural network architectures for scalable density estimation.
OpenAI Scholars 2018 Final Projects
OpenAI published a showcase of final projects from the 2018 Scholars program, featuring diverse applications of machine learning in music generation, intuitive physics, and natural language processing.
OpenAI Five: The International 2018 Results
OpenAI Five competed against world-class professional Dota 2 players at The International 2018, demonstrating high-level gameplay despite losing both matches under 'Real Dota' rules.
OpenAI Large-scale Study of Curiosity-Driven Learning
OpenAI researchers demonstrated that agents trained solely on intrinsic curiosity rewards—based on prediction error—can achieve high performance across 54 benchmark environments without any hand-designed extrinsic rewards.
OpenAI Five Benchmark Results
OpenAI Five won a best-of-three series against a team of 99.95th percentile Dota players, demonstrating advanced AI capabilities in handling complexity and uncertainty.
OpenAI Dactyl: Learning Dexterity through Domain Randomization
OpenAI developed Dactyl, a system that trains a human-like robot hand in simulation to manipulate physical objects with high dexterity, transferring those skills to the real world without fine-tuning.
OpenAI Variational Option Discovery Algorithms
OpenAI introduces Variational Autoencoding Learning of Options by Reinforcement (VALOR) and a curriculum learning approach to stabilize the discovery of diverse behavioral modes in reinforcement learning agents.
OpenAI Scholars 2018: Meet our Scholars
OpenAI introduced its 2018 Scholars cohort, featuring a diverse group of researchers focusing on areas such as reinforcement learning, language modeling, and the intersection of AI and the arts.
OpenAI Five Benchmark
OpenAI announced a benchmark match for OpenAI Five, featuring an expanded hero pool and updated game mechanics to test the AI against high-percentile human players.
Glow: Better Reversible Generative Models
OpenAI introduces Glow, a flow-based generative model using invertible 1x1 convolutions to generate high-resolution images with exact latent-variable inference and efficient sampling.
Learning Montezuma’s Revenge from a Single Demonstration
OpenAI trained an RL agent to achieve a record high score of 74,500 in Montezuma’s Revenge by using a single human demonstration to create a reverse curriculum of starting states.
OpenAI Five
OpenAI Five, a team of five neural networks trained via self-play on Dota 2, began defeating amateur human teams and showed that scaled-up reinforcement learning can handle long-horizon, partially observed tasks without fundamental algorithmic advances.
OpenAI Retro Contest Results
OpenAI concluded its first Retro Contest, where the top-performing AI agents for Sonic the Hedgehog were developed by tuning or extending existing algorithms like PPO and Rainbow DQN.
OpenAI Learning Policy Representations in Multiagent Systems
OpenAI introduces a general learning framework that treats agent modeling as a representation learning problem to understand agent behavior in multiagent systems using minimal interaction data.
OpenAI Improving Language Understanding with Unsupervised Learning
OpenAI demonstrated that a Transformer-based model pre-trained via unsupervised language modeling and then fine-tuned on small supervised datasets can achieve state-of-the-art results across diverse language tasks, including commonsense reasoning.
GamePad: A learning environment for theorem proving
OpenAI introduced GamePad, a system designed to apply machine learning methods to theorem proving within the Coq proof assistant to automate tactic prediction and position evaluation.
OpenAI Fellows Program Fall 2018
OpenAI launched the Fellows program in Fall 2018 to provide a pathway for individuals without formal AI backgrounds to transition into artificial intelligence research.
OpenAI Gym Retro Release
OpenAI has released the full version of Gym Retro, a reinforcement learning platform that expands the available game library to over 1,000 titles across multiple classic consoles to enable research into agent generalization.
OpenAI AI and Compute Analysis
OpenAI analysis reveals that compute used for the largest AI training runs has grown exponentially since 2012, doubling every 3.4 months and far outpacing Moore's Law.
AI Safety via Debate
OpenAI proposes a safety technique that trains AI agents to debate topics to help humans supervise systems that perform tasks beyond human cognitive capabilities.
OpenAI Evolved Policy Gradients (EPG)
OpenAI introduces Evolved Policy Gradients (EPG), a metalearning approach that evolves loss functions to help RL agents generalize to new tasks within a task family.
OpenAI Gotta Learn Fast Benchmark
OpenAI has introduced a new reinforcement learning benchmark based on the Sonic the Hedgehog video game franchise to measure transfer learning and few-shot learning performance.
OpenAI Retro Contest and Gym Retro Release
OpenAI launched the Retro Contest to measure reinforcement learning generalization across unseen video game levels and released Gym Retro, a platform integrating 30 SEGA Genesis and 62 Atari 2600 games into Gym.
OpenAI Variance Reduction for Policy Gradient with Action-Dependent Factorized Baselines
OpenAI researchers developed a bias-free action-dependent baseline for policy gradient methods to reduce gradient estimate variance, enabling faster learning in high-dimensional action spaces.
OpenAI OT-GAN: Improving GANs using optimal transport
OpenAI introduces Optimal Transport GAN (OT-GAN), a variant of generative adversarial nets that uses a new mini-batch energy distance metric to improve training stability and image generation quality.
OpenAI Hackathon Report March 2018
OpenAI hosted its first hackathon on March 3, 2018, bringing together 100 AI community members to develop projects spanning healthcare, safety, and reinforcement learning.