This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHubThe UNetTemporalAttentionMultiply node applies multiplication factors to different types of attention mechanisms in a temporal UNet model. It modifies the model by adjusting the weights of self-attention and cross-attention layers, distinguishing between structural and temporal components. This allows fine-tuning of how much influence each attention type has on the model’s output.
Inputs
| Parameter | Data Type | Required | Range | Description |
|---|---|---|---|---|
model | MODEL | Yes | - | The input model to modify with attention multipliers |
self_structural | FLOAT | No | 0.0 - 10.0 | Multiplier for self-attention structural components (default: 1.0) |
self_temporal | FLOAT | No | 0.0 - 10.0 | Multiplier for self-attention temporal components (default: 1.0) |
cross_structural | FLOAT | No | 0.0 - 10.0 | Multiplier for cross-attention structural components (default: 1.0) |
cross_temporal | FLOAT | No | 0.0 - 10.0 | Multiplier for cross-attention temporal components (default: 1.0) |
Outputs
| Output Name | Data Type | Description |
|---|---|---|
model | MODEL | The modified model with adjusted attention weights |