Introducing STATE-Bench: A benchmark for AI agent memory
STATE-Bench is an open-source, memory-agnostic benchmark freely available to agent developers, researchers, and platform teams.
Topic: AI + Machine Learning
Conductor: Deterministic orchestration for multi-agent AI workflows
Conductor is an open-source CLI (MIT license, Microsoft org) that takes a different approach: you define your multi-agent workflows in YAML, and the routing between agents is deterministic. Jinja2 templates and expression evaluation handle conditions and branching. The orchestration layer consumes zero tokens. The structure is fixed at definition time—and that's the point.
PostgreSQL enters its AI era: Run PostgreSQL like a pro in the era of AI and rapid growth
Postgres can handle intelligent workloads and scale without sacrificing security, performance, reliability, or operational ease.
Microsoft’s open source journey: From 20,000 lines of Linux code to AI at global scale
From Linux kernel code to AI at scale, discover Microsoft’s open source evolution and impact.
Optimizing memory usage in large language models fine-tuning with KAITO: Best practices from Phi-3
The Cloud Native team at Azure is working to make AI on Kubernetes more cost-effective and approachable for a broader range of users.
Build a Hyperlight C guest to securely execute JavaScript
This article will show you how to create a “guest” application that uses the Hyperlight library.
What’s new with Microsoft in open-source and Kubernetes at KubeCon North America 2024
At Microsoft, we are committed to innovation in the cloud-native ecosystem through contributions and leadership from engineers across Azure.
On-Device Training: Training a model in browser
Continuing the ONNXRuntime On-Device Training blog series, we are introducing ONNX Runtime Training for Web.
Exploring DragGAN implementation using ONNX Runtime
Get a technical overview of the Microsoft implementation of the DragGAN2 algorithm using ONNX Runtime.
Recommenders joins LF AI & Data as new Sandbox project
LF AI & Data Foundation announced Recommenders as its latest Sandbox project.
Accelerating over 130,000 Hugging Face models with ONNX Runtime
ONNX models can be accelerated with ONNX Runtime, which works cross-platform and provides coverage for many cloud and language models.
Connect fluid dynamics, machine learning, and virtual reality with ONNX Runtime
Using ONNX Runtime to unlock the promise of developments in science for solving real world problems.
