NNT Define Multihead Attention is a ComfyUI node in the NNT Neural Network Toolkit/Transformers category from ComfyUI Neural Network Toolkit NNT . Inputs…

Neural Network Toolkit (NNT) for ComfyUI is an extensive set of custom ComfyUI nodes for designing, training, and fine-tuning neural networks. This toolkit allows defining models, layers, training workflows, transformers, and tensor operations in a visual manner using nodes.

ComfyUI Neural Network Toolkit NNT 

embed_dim

num_heads

dropout

bias

add_bias_kv

add_zero_attn

batch_first

kdim

vdim

LAYER_STACK

NntDefineMultiheadAttention

LIST

NNT Define Multihead Attention

Name	Type	Default	Description
embed_dim	INT	51264–2048	—
num_heads	INT	81–32	—
dropout	FLOAT	0.10–0.9	—
bias	COMBO	True	2 options: True, False
add_bias_kv	COMBO	False	2 options: True, False
add_zero_attn	COMBO	False	2 options: True, False
batch_first	COMBO	True	2 options: True, False
kdim	INT	00–2048	—
vdim	INT	00–2048	—
LAYER_STACKopt	LIST		—

NNT Define Multihead Attention

Inputs (10)

Outputs (1)