Show HN: A 20-line kernel that fixes float32 variance under XLA

Category: ai-ml

Tags: online-learning, streaming-ml, jax, drift-detection, robot-control, continual-learning

Score: 6.3/10 (Innovation: 6, Technical: 7, Documentation: 7, Utility: 5)

Dense-Armor is a Python library for online learning and streaming statistics, offering a broad set of primitives including drift detection, anomaly scoring, safe robot command enforcement, and continual learning. It targets systems where data arrives one sample at a time, with JAX as the backend and bridges to scikit-learn, PyTorch, and TensorFlow.

Target audience: ML engineers and robotics developers working with streaming data and online learning

Repository: https://github.com/tatopenn-cell/Dense-Armor · Python · AGPL-3.0 · 9 stars

View on Hacker News