Skip to content
Accelerating Gemma 4: faster inference with multi-token prediction drafters — txtfeed | txtfeed