Skip to main content
The UNetSelfAttentionMultiply node applies multiplication factors to the query, key, value, and output components of the self-attention mechanism in a UNet model. It allows you to scale different parts of the attention computation to experiment with how attention weights affect the model’s behavior.

Inputs

ParameterDescriptionData TypeRequiredRange
modelThe UNet model to modify with attention scaling factorsMODELYes-
qMultiplication factor for query component (default: 1.0)FLOATNo0.0 - 10.0
kMultiplication factor for key component (default: 1.0)FLOATNo0.0 - 10.0
vMultiplication factor for value component (default: 1.0)FLOATNo0.0 - 10.0
outMultiplication factor for output component (default: 1.0)FLOATNo0.0 - 10.0

Outputs

Output NameDescriptionData Type
MODELThe modified UNet model with scaled attention componentsMODEL
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub

Source fingerprint (SHA-256): 7a6039eb2faae56437a5eb6fe01be6d38e53c0632175a3405a1e24a476d4da82