Precise Editing and Flexible Referencing for Interactable Worlds
EditWorld adds streaming edits and image references to video world models, leading WBench-Editing.
EditWorld is a video world model that moves beyond navigation to precise modification by streaming editing instructions and reference images during autoregressive generation. It uses Gated Causal Attention for time-varying conditions and a Sparse Context mechanism that keeps historical context bounded for long-horizon inference. Training combines autoregressive and bidirectional objectives with annealed self-resampling, plus a dedicated synthesis and annotation pipeline. On the new WBench-Editing benchmark it reports an overall score of 73.8 and an editing score of 80.0, ahead of prior methods on editing metrics.
- Streams editing instructions and reference images during autoregressive generation.
- Gated Causal Attention and Sparse Context bound long-horizon context.
- Scores 73.8 overall and 80.0 on editing in WBench-Editing.
- Uses joint autoregressive and bidirectional training with annealed self-resampling.
Full article156 words · extracted from huggingface.co · click to collapse
We present EditWorld, a video world model for precise editing and flexible referencing in interactable worlds. Existing video world models primarily focus on navigation, letting users explore generated worlds but offering limited control over how existing world content is modified. EditWorld extends world modeling from exploration to precise modification by streaming editing instructions and reference images during autoregressive generation. To support these capabilities, EditWorld introduces Gated Causal Attention for temporally varying editing conditions and reference images, together with a Sparse Context mechanism that maintains a bounded historical context for long-horizon inference. We further adopt joint autoregressive and bidirectional training with annealed self-resampling, and construct a dedicated data synthesis and annotation pipeline that provides supervision for world editing. We also present WBench-Editing to systematically evaluate streaming world editing capabilities. EditWorld achieves the best overall performance on WBench-Editing with an overall score of 73.8 and an editing score of 80.0, substantially outperforming existing methods on editing-related metrics. https://github.com/leoisufa/EditWorld
Text extracted automatically; images, tables and formatting may be missing. Original: https://huggingface.co/papers/2609.34470