ComfyUI Node
LoRA Truncate Rank (CCN)
Fast rank reduction for SVD-extracted LoRAs by directly truncating weight matrices. Runs in seconds vs. minutes for full re-decomposition. NOT valid for trained LoRAs — their rank dimensions are not ordered by importance.
LoRA Truncate Rank (CCN)
- filepath
◄lora_name▾►
◄new_rank32►
◄filename_prefixloras/CCN_truncated_lora►
◄output_dtypematch_original►
◄verbosetrue►
CategoryCCN/lora
Inputs (5)
| Name | Type | Default | Description |
|---|---|---|---|
| lora_name | COMBO | Source LoRA file. Must be from SVD extraction, not training. | |
| new_rank | INT | 321–4096 | Target rank. Must be less than the source LoRA's rank. Layers already at or below this rank are left unchanged. |
| filename_prefix | STRING | loras/CCN_truncated_lora | Output path and filename prefix (relative to output directory). |
| output_dtype | COMBO | match_original | Data type for saved tensors. fp8_e4m3 halves size vs fp16 with minimal quality loss. |
| verbose | BOOLEAN | true | Print per-layer rank changes to console. |
Outputs (1)
| Name | Type | Description |
|---|---|---|
| filepath | STRING | — |