Nodes/ComfyUI-Tensor-Prism-Node-Pack/Analyze Model Weights (Tensor Prism)
ComfyUI Node

Analyze Model Weights (Tensor Prism)

A ComfyUI node in Tensor_Prism/Analysis with 13 inputs and 1 output.

By Arctenox·Created 11 months ago·Updated 6 months ago· 2
Analyze Model Weights (Tensor Prism)
  • model_a
  • model_b
  • merge_recipe
optimization_method
global_alpha0.50
optimization_strength0.80
target_mean_weight0.50
target_std_dev-1.00
smooth_weightsfalse
smoothing_strength0.30
use_model_a_as_basetrue
fp16_modetrue
verbosetrue
auto_detect_optimalfalse
CategoryTensor_Prism/Analysis

Inputs (13)

NameTypeDefaultDescription
model_aMODEL
model_bMODEL
optimization_methodCOMBO5 options: combined, similarity, variance, gradient_magnitude, entropy
global_alphaFLOAT0.500–1Starting point for optimization
optimization_strengthFLOAT0.800–1How much to trust automated optimization
target_mean_weightFLOAT0.50-1–1Target mean weight across all blocks (-1 = auto-detect)
target_std_devFLOAT-1.00-1–0.5Target standard deviation (-1 = auto, 0 = uniform, higher = more variation)
smooth_weightsBOOLEANfalseApply smoothing to reduce weight variance
smoothing_strengthFLOAT0.300–1How much to smooth (0=none, 1=completely flatten)
use_model_a_as_baseoptBOOLEANtrue
fp16_modeoptBOOLEANtrue
verboseoptBOOLEANtrue
auto_detect_optimaloptBOOLEANfalseAuto-detect optimal mean weight based on model similarity

Outputs (1)

NameTypeDescription
merge_recipeMERGE_RECIPE