Skip to main content
This documentation was AI-generated. If you find any errors or have suggestions for improvement, please feel free to contribute! Edit on GitHub
The ModelMergeCosmosPredict2_14B node allows you to merge two AI models by adjusting the influence of different model components. It provides fine-grained control over how much each part of the second model contributes to the final merged model, using blending weights for specific model layers and components.

Inputs

ParameterData TypeRequiredRangeDescription
model1MODELYes-The base model to merge with
model2MODELYes-The secondary model to merge into the base model
pos_embedder.FLOATYes0.0 - 1.0Position embedder blending weight (default: 1.0)
x_embedder.FLOATYes0.0 - 1.0Input embedder blending weight (default: 1.0)
t_embedder.FLOATYes0.0 - 1.0Time embedder blending weight (default: 1.0)
t_embedding_norm.FLOATYes0.0 - 1.0Time embedding normalization blending weight (default: 1.0)
blocks.0.FLOATYes0.0 - 1.0Block 0 blending weight (default: 1.0)
blocks.1.FLOATYes0.0 - 1.0Block 1 blending weight (default: 1.0)
blocks.2.FLOATYes0.0 - 1.0Block 2 blending weight (default: 1.0)
blocks.3.FLOATYes0.0 - 1.0Block 3 blending weight (default: 1.0)
blocks.4.FLOATYes0.0 - 1.0Block 4 blending weight (default: 1.0)
blocks.5.FLOATYes0.0 - 1.0Block 5 blending weight (default: 1.0)
blocks.6.FLOATYes0.0 - 1.0Block 6 blending weight (default: 1.0)
blocks.7.FLOATYes0.0 - 1.0Block 7 blending weight (default: 1.0)
blocks.8.FLOATYes0.0 - 1.0Block 8 blending weight (default: 1.0)
blocks.9.FLOATYes0.0 - 1.0Block 9 blending weight (default: 1.0)
blocks.10.FLOATYes0.0 - 1.0Block 10 blending weight (default: 1.0)
blocks.11.FLOATYes0.0 - 1.0Block 11 blending weight (default: 1.0)
blocks.12.FLOATYes0.0 - 1.0Block 12 blending weight (default: 1.0)
blocks.13.FLOATYes0.0 - 1.0Block 13 blending weight (default: 1.0)
blocks.14.FLOATYes0.0 - 1.0Block 14 blending weight (default: 1.0)
blocks.15.FLOATYes0.0 - 1.0Block 15 blending weight (default: 1.0)
blocks.16.FLOATYes0.0 - 1.0Block 16 blending weight (default: 1.0)
blocks.17.FLOATYes0.0 - 1.0Block 17 blending weight (default: 1.0)
blocks.18.FLOATYes0.0 - 1.0Block 18 blending weight (default: 1.0)
blocks.19.FLOATYes0.0 - 1.0Block 19 blending weight (default: 1.0)
blocks.20.FLOATYes0.0 - 1.0Block 20 blending weight (default: 1.0)
blocks.21.FLOATYes0.0 - 1.0Block 21 blending weight (default: 1.0)
blocks.22.FLOATYes0.0 - 1.0Block 22 blending weight (default: 1.0)
blocks.23.FLOATYes0.0 - 1.0Block 23 blending weight (default: 1.0)
blocks.24.FLOATYes0.0 - 1.0Block 24 blending weight (default: 1.0)
blocks.25.FLOATYes0.0 - 1.0Block 25 blending weight (default: 1.0)
blocks.26.FLOATYes0.0 - 1.0Block 26 blending weight (default: 1.0)
blocks.27.FLOATYes0.0 - 1.0Block 27 blending weight (default: 1.0)
blocks.28.FLOATYes0.0 - 1.0Block 28 blending weight (default: 1.0)
blocks.29.FLOATYes0.0 - 1.0Block 29 blending weight (default: 1.0)
blocks.30.FLOATYes0.0 - 1.0Block 30 blending weight (default: 1.0)
blocks.31.FLOATYes0.0 - 1.0Block 31 blending weight (default: 1.0)
blocks.32.FLOATYes0.0 - 1.0Block 32 blending weight (default: 1.0)
blocks.33.FLOATYes0.0 - 1.0Block 33 blending weight (default: 1.0)
blocks.34.FLOATYes0.0 - 1.0Block 34 blending weight (default: 1.0)
blocks.35.FLOATYes0.0 - 1.0Block 35 blending weight (default: 1.0)
final_layer.FLOATYes0.0 - 1.0Final layer blending weight (default: 1.0)
Note: All blending weight parameters accept values between 0.0 and 1.0, where 0.0 means no contribution from model2 and 1.0 means full contribution from model2 for that specific component.

Outputs

Output NameData TypeDescription
modelMODELThe merged model combining features from both input models