LITM

A developer released TIS (Token Importance Scoring), a learned method for compressing the KV cache in large language models, achieving 100% accuracy on synthetic retrieval at 50% cache budget. The app…

Researchers at Hugging Face debate whether large language models can lose conceptual continuity while remaining coherent, proposing falsification-first controls to distinguish genuine architectural ga…