Latent Factorization

1-Bit AI Models Actually Work — On-Device AI's Cost Inflection Point May Be Near

LittleBit compresses AI to 1-bit via 'latent factorization' with minimal loss. If viable, costs drop — phones, factories, even servers could run large