This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHubThe UNetCrossAttentionMultiply node applies multiplication factors to the cross-attention mechanism in a UNet model. It allows you to scale the query, key, value, and output components of the cross-attention layers to experiment with different attention behaviors and effects.
Inputs
| Parameter | Data Type | Required | Range | Description |
|---|---|---|---|---|
model | MODEL | Yes | - | The UNet model to modify with attention scaling factors |
q | FLOAT | No | 0.0 - 10.0 | Scaling factor for query components in cross-attention (default: 1.0) |
k | FLOAT | No | 0.0 - 10.0 | Scaling factor for key components in cross-attention (default: 1.0) |
v | FLOAT | No | 0.0 - 10.0 | Scaling factor for value components in cross-attention (default: 1.0) |
out | FLOAT | No | 0.0 - 10.0 | Scaling factor for output components in cross-attention (default: 1.0) |
Outputs
| Output Name | Data Type | Description |
|---|---|---|
model | MODEL | The modified UNet model with scaled cross-attention components |