ByteDanceBAGEL
BAGEL 7B MoT
BAGEL unifies text and image understanding, image generation and image editing.
Model design
The model has seven billion active parameters and fourteen billion total parameters. Its mixture-of-transformer-experts design uses separate image encoders for semantic and pixel-level features. Training includes interleaved language, image, video and web data.BAGEL 7B MoT official model card
Practical use
Practical guidance. BAGEL is useful when one local model must both inspect and change an image. Test whether it preserves untouched content during an edit, not only whether the requested new object appears.
Specifications & access
| Detail | Value and conditions |
|---|---|
| Access | Downloadable |
| Model identifiers | BAGEL-7B-MoT |
| License | Apache 2.0BAGEL 7B MoT official model card |
| Access | Weights are published in the official ByteDance Seed repository.BAGEL 7B MoT official model card |
Related models
- Seedream 5.0 Pro
ByteDance ยท Seedream
Image generation and editing model for text-rich graphics, marked-up edits and layer separation.
Launch date not confirmedAPI