mirror of
https://github.com/hanzoui/workflow-templates-broken-zap-2026-06-08.git
synced 2026-10-30 20:44:35 +00:00
1.4 KiB
1.4 KiB
UNetTemporalAttentionMultiply
Category: _for_testing/attention_experiments
Description
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The 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 | Type | Default | Description |
|---|---|---|---|
| model | Model | — | |
| self_structural | Float | 1.0 | |
| self_temporal | Float | 1.0 | |
| cross_structural | Float | 1.0 | |
| cross_temporal | Float | 1.0 |
Outputs
| Output | Type |
|---|---|
| )], | |
| is_experimental=True, | |
| ) |
@classmethod
def execute(cls, model, self_structural, self_temporal, cross_structural, cross_temporal) -> io.NodeOutput:
m = model.clone()
sd = model.model_state_dict( | Model |
Source: comfy_extras/nodes_attention_multiply.py
Used in 2 template(s)