Context compression finally works in production: new research cuts LLM input 16x without the accuracy hit

LCLMs compress LLM context before decode — 8.8x faster at 16x compression, beating every KV cache method tested. Open-sourced by NYU and Columbia.
HotHardware

NVIDIA Unlocks Higher Quality Textures With Neural Compression For 4X VRAM Savings

If you read this site, chances are pretty good that you've probably played around with a neural image upscaler at some point. These programs, like the popular "waifu2x", use a pre-trained neural ...
EDN

Data compression tutorial: Part 1

Images transmitted over the world wide web are an excellent example of why data compression is important. Suppose we need to download a digitized color photograph over a computer's 33.6 kbps modem. If ...
TechSpot

AMD's neural texture compression could reduce storage and VRAM requirements

Something to look forward to: High-resolution textures are a primary factor behind the growing install sizes and VRAM usage in modern blockbuster games. Nvidia proposed a neural-network-based method ...