engblogs

Explore the latest in open-source GPU programming for neural networks with Triton.

Generally capable agents are trained to play a variety of games and tasks without human interaction data, marking an important step toward more adaptive AI agents.

Exploring the development of AI agents with general and adaptive behavior, capable of learning and reacting to new conditions and playing a variety of games and tasks.

Exploring the advancements in machine learning for high-accuracy protein structure prediction on a large scale

Sharing high-quality protein structure predictions for all human proteins and 20 other organisms using AlphaFold 2 system.

Sharing knowledge among web engineers at SoundCloud and being part of the 'Web Collective'

Introducing Melting Pot: a scalable evaluation suite for multi-agent reinforcement learning, designed to test generalisation to novel social situations and assess a broad range of social interactions with 21 MARL substrates and over 85 unique test scenarios.

Exploring the efficacy of large language models trained on code

Reflecting on the inaugural year of EleutherAI.

Upcoming API security updates requiring action from developers.

This role has been filled

Enhancing language model performance through training on a carefully curated dataset