AIWalker

Feb 12, 2025 · Artificial Intelligence

Goku: How HKU and ByteDance’s New Model Sets New Benchmarks in Commercial Image and Video Generation

The paper presents Goku, a rectified‑flow transformer that jointly generates high‑quality images and videos at commercial scale, detailing its novel architecture, massive high‑quality data pipeline, efficient large‑scale training tricks, and state‑of‑the‑art results on GenEval, DPG‑Bench, VBench and UCF‑101.