Skip to main content

> ML_ALGORITHM // MASKED-GENERATIVE-IMAGE-TRANSFORMER-MASKGIT_v1.0

MaskGIT (Masked Generative Image Transformer)

Non-autoregressive generative vision Transformer that produces high-fidelity images in just 8-16 parallel iterative decoding steps.

Discrete Token Generative Transformersdeep-generativemoderate-posthoclarge (>100k)
Back to All Algorithms
Computational Complexity
Training Complexity:O(epochs * batch_size * transformer_pass)
Inference Complexity:O(8 to 16 * transformer_pass)
Hardware Profile
CPU Friendly:No
Requires GPU:Yes
Memory Footprint:high
Interpretability & Data
Interpretability Tier:moderate-posthoc
Training Data Needs:large (>100k)

Interpretability Assessment

Non-autoregressive decoding progressively unmasks highest-confidence image tokens in constant iterations.

Suitable Tasks & Supported Modalities

Suitable Tasks:
image generationimage editingimage in painting
Supported Modalities:
image

Implementing Libraries

PyTorchLinux Foundation / PyTorch Foundation · v2.4.1
View Spec
kerashub

Foundational Literature

Common Pitfalls & Warnings
  • Linear or suboptimal unmasking schedules leave uncoordinated visual artifacts across distant unmasked patches