Tag
37 articles
This article explains how automated AI systems can self-improve their alignment without sacrificing general capabilities, a breakthrough in AI safety research.
Learn to accelerate transformer training using NVIDIA's Transformer Engine with fused kernels, FP8, and BF16 optimizations in PyTorch.
LinkedIn has decided against expanding its data centers in the next year, instead focusing on maximizing existing GPU resources. The move reflects a broader industry trend toward sustainable AI growth and efficient resource management.
Learn how to check and optimize your phone's RAM usage with simple tools and methods that anyone can follow.
This explainer examines how China's Moonshot AI has developed highly efficient AI models like Kimi K3 that outperform traditional systems at a fraction of the cost, exploring the advanced techniques behind model optimization and their implications for the global AI landscape.
Learn how to build and optimize AI inference pipelines using TensorFlow, mimicking the specialized approach of companies like Etched in the chip industry.
Configure a quiet, efficient Linux desktop environment similar to the System76 Thelio Mira with optimized performance and productivity tools.
Google is reportedly developing a new AI chip designed to make its Gemini models run much more efficiently. The move aims to enhance performance while reducing operational costs and environmental impact.
This article explains how AI-driven design optimization enables Tesla to create innovative vehicle configurations like the six-seat Model Y Long Wheelbase through machine learning algorithms and optimization techniques.
Learn how to build and optimize AI inference pipelines using TensorFlow, similar to what companies like Etched are developing for specialized AI chips.
This article explains how AI-driven price optimization works in e-commerce, using Amazon Prime Day SSD deals as an example to illustrate advanced machine learning techniques that dynamically adjust product prices in real-time.
This explainer explores NVIDIA's cuTile, a tile-based GPU programming interface that simplifies high-performance kernel development for compute-intensive tasks like matrix operations, while maintaining performance close to hand-optimized CUDA code.
