Wan-Streamer v0.2：更高解析度，相同延遲

摘要

我們推出 Wan-Streamer v0.2，這是在即時串流、端到端影音互動模型上的延遲保持升級版本。v0.2 保留了 v0.1 的建模架構，但將互動輸出串流從 192x336 提升至 640x368，同時在 25 FPS 下維持約 200 毫秒的模型端訊號到訊號延遲。更高解析度的串流支援場景根基的中景代理，其在即時對話期間的姿態、視線、手勢、附近物體及局部場景佈局均保持清晰可辨。為支援更大的視覺串流而不增加用戶感知延遲，v0.2 保留了思考器作為單 GPU 低延遲路徑，負責串流感知、建立生成快取的短語言/狀態 Transformer 前向傳播，以及最終解碼。執行器則成為多 GPU 的 Ulysses 風格上下文並行群組，負責處理昂貴的下一個單元潛變量生成。每個執行器 rank 將傳入的 K/V 寫入預先分片的本地快取。長的高解析度潛在影片序列跨 rank 分片進行去噪，並透過 Ulysses 通訊聚合，而更短的音訊潛在序列則無需序列分片即可生成。在此拆分中，思考器的語言/狀態計算僅以 K/V 條件形式傳遞至執行器，因此執行器群組內部無需溝通單獨的語言序列。這使額外硬體集中於視覺生成，同時保持緊湊的思考器-執行器邊界，使得包含 350 毫秒雙向網路預算後的總遠端互動延遲約為 550 毫秒。

English

We present Wan-Streamer v0.2, a latency-preserving upgrade of the native-streaming, end-to-end audio-visual interaction model. v0.2 keeps the v0.1 modeling formulation, but raises the interactive output stream from 192x336 to 640x368 while preserving approximately 200 ms model-side signal-to-signal latency at 25 FPS. The higher-resolution stream supports scene-grounded mid-shot agents whose posture, gaze, hands, nearby objects, and local scene layout remain legible during real-time conversation. To support the larger visual stream without adding user-visible delay, v0.2 keeps the thinker as a single-GPU low-latency path for streaming perception, the short language/state Transformer pass that builds the generation cache, and final decoding. The performer becomes a multi-GPU Ulysses-style context-parallel group for the expensive next-unit latent generation. Each performer rank writes incoming K/V into a pre-sharded local cache. The long high-resolution latent video sequence is split across ranks for denoising and gathered through Ulysses communication, while the much shorter audio latent sequence is generated without sequence sharding. In this split, the thinker's language/state computation reaches the performer only as K/V conditioning, so no separate language sequence has to be communicated inside the performer group. This concentrates additional hardware on visual generation while preserving the compact thinker-performer boundary, keeping total remote interaction latency at approximately 550 ms when a 350 ms bidirectional network budget is included.