Engineering Insights

AI-written critiques help humans notice flaws

We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more often when shown our model’s critiques. Larger models are better at self-critiquing, with scale improving critique-writing more than summary-writing. This shows promise for using AI systems to assist human supervision of AI systems on difficult tasks.

Sponsored Post: G-Core Labs, Close, Wynter, Pinecone, Kinsta, Bridgecrew, IP2Location, StackHawk, InterviewCamp.io, Educative, Stream, Fauna, Triplebyte

Who's Hiring?  Close is building the sales communication platform of the future and we're looking for a Site Reliability Engineer to help us accomplish that goal. If you have a passion for databases (both SQL and NoSQL), significant experience building, managing, and monitoring infrastructure, databases, and backend services at scale and want to work with a rag-tag team of hardworking, but humble humans, then come check us out! Wynter is looking for system administrators, engineers, and develop

Techniques for training large neural networks

Large neural networks are at the core of many recent advances in AI, but training them is a difficult engineering and research challenge which requires orchestrating a cluster of GPUs to perform a single synchronized calculation.

Bringing forward the End-of-Life Date for Node.js 16

The Annotated Diffusion Model

Deep Q-Learning with Space Invaders

Build a REST API with NestJS, Prisma, PostgreSQL and Swagger

Learn how to build a backend REST API with NestJS, Prisma, PostgreSQL and Swagger. In this article, you will learn how to set up the project, build the API and document it with Swagger.

Best practices for deploying language models

Cohere, OpenAI, and AI21 Labs have developed a preliminary set of best practices applicable to any organization developing or deploying large language models.