Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data
arxiv.org
September 17, 2026
2 min read
🔥🔥🔥🔥🔥
50/100
Summary
Computer Science > Artificial Intelligence [Submitted on 16 Sep 2026] Title:Infinite-Parameter LLMs: Generating and Adapting Weights from Live Data View PDF HTML (experimental)Abstract:The scaling laws hold that a language model grows more capable with more parameters and more training data, and Mixture-of-Experts (MoE) architectures have ridden these laws to remarkable results, activating only a ...