FlashHead: Efficient Drop-In Replacement for the Classification Head in Language Model Inference

ArXi:2603.14591v1 Announce Type: cross Language models are increasingly adopting smaller architectures optimized for consumer devices. In this setting, inference efficiency is the primary constraint. Meanwhile, vocabulary sizes continue to grow rapidly, making the classification head a critical bottleneck that accounts for up to 60\% of model parameters, and 50\% of inference compute. We