ComfyUI Node
SDXL Advanced Block Merge (Tensor Prism)
A ComfyUI node in Tensor_Prism/Merge with 41 inputs and 1 output.
SDXL Advanced Block Merge (Tensor Prism)
- model_A
- model_B
- model_C
- MODEL
◄merge_method▾►
◄default_unet_ratio0.50►
◄memory_limit_gb8.0►
◄force_cpufalse►
◄batch_size50►
◄auto_memory_managementtrue►
◄ties_global_alpha_A0.50►
◄ties_global_alpha_B0.50►
◄rescale_output_magnitudesfalse►
◄iterations1►
◄a_delta_factor1.00►
◄b_delta_factor1.00►
◄precision_modeauto►
◄aggressive_cleanuptrue►
◄out_ratio0.50►
◄time_embed_ratio0.50►
◄label_emb_ratio0.50►
◄input_block_00_ratio0.50►
◄output_block_00_ratio0.50►
◄input_block_01_ratio0.50►
◄output_block_01_ratio0.50►
◄input_block_02_ratio0.50►
◄output_block_02_ratio0.50►
◄input_block_03_ratio0.50►
◄output_block_03_ratio0.50►
◄input_block_04_ratio0.50►
◄output_block_04_ratio0.50►
◄input_block_05_ratio0.50►
◄output_block_05_ratio0.50►
◄input_block_06_ratio0.50►
◄output_block_06_ratio0.50►
◄input_block_07_ratio0.50►
◄output_block_07_ratio0.50►
◄input_block_08_ratio0.50►
◄output_block_08_ratio0.50►
◄middle_block_00_ratio0.50►
◄middle_block_01_ratio0.50►
◄middle_block_02_ratio0.50►
CategoryTensor_Prism/Merge
Inputs (41)
| Name | Type | Default | Description |
|---|---|---|---|
| model_A | MODEL | — | |
| model_B | MODEL | — | |
| merge_method | COMBO | 3 options: Linear Interpolation, Add Difference, TIES-Merging (Simplified) | |
| default_unet_ratio | FLOAT | 0.500–1 | — |
| memory_limit_gb | FLOAT | 8.01–64 | — |
| force_cpu | BOOLEAN | false | — |
| batch_size | INT | 501–500 | — |
| auto_memory_management | BOOLEAN | true | — |
| model_Copt | MODEL | — | |
| ties_global_alpha_Aopt | FLOAT | 0.500–1 | — |
| ties_global_alpha_Bopt | FLOAT | 0.500–1 | — |
| rescale_output_magnitudesopt | BOOLEAN | false | — |
| iterationsopt | INT | 11–100 | — |
| a_delta_factoropt | FLOAT | 1.00-2–2 | — |
| b_delta_factoropt | FLOAT | 1.00-2–2 | — |
| precision_modeopt | COMBO | auto | 3 options: auto, fp16, fp32 |
| aggressive_cleanupopt | BOOLEAN | true | — |
| out_ratioopt | FLOAT | 0.500–1 | — |
| time_embed_ratioopt | FLOAT | 0.500–1 | — |
| label_emb_ratioopt | FLOAT | 0.500–1 | — |
| input_block_00_ratioopt | FLOAT | 0.500–1 | — |
| output_block_00_ratioopt | FLOAT | 0.500–1 | — |
| input_block_01_ratioopt | FLOAT | 0.500–1 | — |
| output_block_01_ratioopt | FLOAT | 0.500–1 | — |
| input_block_02_ratioopt | FLOAT | 0.500–1 | — |
| output_block_02_ratioopt | FLOAT | 0.500–1 | — |
| input_block_03_ratioopt | FLOAT | 0.500–1 | — |
| output_block_03_ratioopt | FLOAT | 0.500–1 | — |
| input_block_04_ratioopt | FLOAT | 0.500–1 | — |
| output_block_04_ratioopt | FLOAT | 0.500–1 | — |
| input_block_05_ratioopt | FLOAT | 0.500–1 | — |
| output_block_05_ratioopt | FLOAT | 0.500–1 | — |
| input_block_06_ratioopt | FLOAT | 0.500–1 | — |
| output_block_06_ratioopt | FLOAT | 0.500–1 | — |
| input_block_07_ratioopt | FLOAT | 0.500–1 | — |
| output_block_07_ratioopt | FLOAT | 0.500–1 | — |
| input_block_08_ratioopt | FLOAT | 0.500–1 | — |
| output_block_08_ratioopt | FLOAT | 0.500–1 | — |
| middle_block_00_ratioopt | FLOAT | 0.500–1 | — |
| middle_block_01_ratioopt | FLOAT | 0.500–1 | — |
| middle_block_02_ratioopt | FLOAT | 0.500–1 | — |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| MODEL | MODEL | — |