engblogs

This blog post discusses how multiple teams at Pinterest collaborated to build a large-scale, highly-flexible, and cost-efficient user signal platform service that indexes user events in near real-time, constructs user sequences, and is easy to use for online service requests and ML training & inferences.

Speeding up development of large AI models by utilizing a modular approach

Exploring the brand story and history of Modular

Exploring the modular approach to work and its impact on productivity and collaboration.

Rebuilding AI in response to the changing economy with a modular approach.

A discussion on the benefits and advantages of using a next-generation AI developer platform with modular capabilities.

Increasing development velocity of giant AI models through modularity.

Debugging data to improve ML applications by identifying and fixing data quality issues for better accuracy and fairness.

A deep-learning system that uses exterior data to explore the internal structures, voids, and cracks of materials.

Stability AI introduces DeepFloyd IF, an advanced text-to-image model for research labs to explore and experiment with text-to-image generation techniques.

Stability AI introduces StableVicuna, an open source RLHF LLM chatbot trained via reinforced learning from human feedback.

Machine learning on Kubernetes: wisdom learned at Snorkel AI