Compressing Streaming Neural Audio Encoders via Latent-Space Distillation

这条英文动态主要涉及模型能力与工程。原文要点：System-wide Dictation on Apple devices runs entirely on-device, and the speech it transcribes reaches the foundation model through a tokenizer: an encoder that maps short windows of waveform onto the representation the language model reads. Because that model is sparsely activated under Instruction-Following Pruning, only a small subset of its experts occupies DRAM at any time, so the always-on tokenizer competes for...