|
|
|
|
|
|
by ShamelessC
932 days ago
|
|
|
The only difference is the label, really. The underlying transformer architecture and the approach of using a codebook is identical to a large language model. The same approach was also used originally for image generation in DALL-E 1. |
|
|