arXiv cs.AI / cs.LG / cs.CL·7d agoWeight Pair Encoding: Inducing a Smaller Grammar in Neural Network Weights#compression#quantization#grammarAI research