Dual-Flow Transformers: Decoupling the Primary Prefill Path from Additional Decode Computation

Abstract

arXiv:2608.12385v1 Announce Type: new Abstract: As large language models serve more requests, cumulative inference cost is becoming increasingly important relative to one-time training cost. The two inference phases stress hardware differently: prompt prefill is parallel and typically compute-bound, whereas autoregressive decode is sequential and often memory-bandwidth-bound. Conventional width or depth scaling increases both costs together because every added layer is evaluated in both phases.