AWhat changed
Trained bidirectional image-token prediction on random masks and generated images through parallel prediction and iterative refinement rather than fixed raster-order decoding.
Scientific scope
Trained bidirectional image-token prediction on random masks and generated images through parallel prediction and iterative refinement rather than fixed raster-order decoding. Discrete image tokenization and masked modeling have predecessors. Parallel refinement changes the generation procedure and is not exact sampling of the raster-autoregressive model. Reported speedup depends on token count, schedule and hardware.
Evidence
The linked original sources describe the stated contribution and its evidence limits. WRS has not independently rerun the experiments or certified a proof. Discrete image tokenization and masked modeling have predecessors. Parallel refinement changes the generation procedure and is not exact sampling of the raster-autoregressive model. Reported speedup depends on token count, schedule and hardware.
Contributors
- Source-listed authors and technical contributors · SCIENCE: Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
- MaskGIT: Masked Generative Image Transformer
- Chang et al.: MaskGIT bidirectional masked image refinement — author/institutional project
Provisional WRS assessment, edition 2026-10-16.1. Independent expert calibration is pending. This grade does not certify correctness or assign ownership.
Full scientific record · Browse contribution grades · Request a correction