Alaya-EVOKE：從線性擴展監督到無盡世界

摘要

互動式世界模型必須支援持久記憶、即時互動與長時域生成，然而這些需求對模型施加了相互衝突的要求。在去噪器上下文或鍵值快取中維護歷史會產生不斷增長的成本，迫使模型在會話長度與記憶保留之間進行取捨；而低延遲互動依賴於少步生成，其能力受到教師模型的限制。Evoke 透過外部化持久世界狀態並重新設計教師模型以實現長時域互動生成，解決了這兩項限制。場景幾何被維護在一個外部、以相機為索引的世界狀態庫中，僅從中檢索與視圖相關的資訊，從而使去噪器上下文在會話增長時保持有界。我們並未將教師模型視為固定的生成器，而是為其設計了長時域監督能力：其稀疏注意力結合了分塊分組、對所選遠距幀的檢索，以及線性注意力全局狀態，在記憶體與計算上實現線性增長，同時能夠在長時域上進行監督。這種監督暴露了在短時間窗口內仍保持局部合理性的內容漂移，而每塊條件化則允許在整個序列中進行提示變更與事件控制。在自強制展開下應用的 30 秒分佈匹配目標，將這兩種能力轉移至一個不使用無分類器引導的三步學生模型，在保留即時條件化的同時，增強了對長期漂移的抵抗能力。憑藉有界上下文與循環外部記憶，Evoke 支援開放式、持續演化的生成；在單張 H200 上、解析度為 384×640 時，每個 1.5 秒的區塊僅需 2.11 秒即可生成。作為一個三步世界模型，Evoke 在 WBench 上取得了最先進的效能，同時在 VBench-Long 與 VBench-2.0 上保持競爭力。

English

Interactive world models must support persistent memory, responsive interaction, and long-horizon generation, yet these requirements place conflicting demands on the model. Maintaining history in the denoiser context or key-value cache incurs growing cost, forcing a trade-off between session length and retained memory, while low-latency interaction relies on few-step generation whose capabilities are bounded by its teacher. Evoke addresses both limitations by externalizing persistent world state and redesigning the teacher for long-horizon interactive generation. Scene geometry is maintained in an external, camera-indexed world state bank, from which only view-relevant information is retrieved, keeping the denoiser context bounded as the session grows. Rather than treating the teacher as a fixed generator, we design it for long-horizon supervision: its sparse attention combines chunk-wise grouping, retrieval of selected distant frames, and a linear-attention global state, yielding linear growth in memory and compute while enabling supervision over long horizons. Such supervision exposes content drift that stays locally plausible within short windows, while per-chunk conditioning enables prompt changes and event control throughout the sequence. A 30-second distribution-matching objective, applied under self-forced rollouts, transfers both capabilities to a three-step student that uses no classifier-free guidance, improving resistance to long-term drift while preserving responsive conditioning. With bounded context and recurrent external memory, Evoke supports open-ended, continuously evolving generation; on a single H200 at 384times 640, each 1.5,s chunk is generated in 2.11,s. As a three-step world model, Evoke achieves state-of-the-art performance on WBench while remaining competitive on VBench-Long and VBench-2.0.