Add an experimental LoraSave node to extract model loras.

The model_diff input should be connected to the output of a
ModelMergeSubtract node.

comfy_extras/nodes_lora_extract.py

@@ -0,0 +1,91 @@
+import torch
+import comfy.model_management
+import comfy.utils
+import folder_paths
+import os
+import logging
+
+CLAMP_QUANTILE = 0.99
+
+def extract_lora(diff, rank):
+    conv2d = (len(diff.shape) == 4)
+    kernel_size = None if not conv2d else diff.size()[2:4]
+    conv2d_3x3 = conv2d and kernel_size != (1, 1)
+    out_dim, in_dim = diff.size()[0:2]
+    rank = min(rank, in_dim, out_dim)
+
+    if conv2d:
+        if conv2d_3x3:
+            diff = diff.flatten(start_dim=1)
+        else:
+            diff = diff.squeeze()
+
+
+    U, S, Vh = torch.linalg.svd(diff.float())
+    U = U[:, :rank]
+    S = S[:rank]
+    U = U @ torch.diag(S)
+    Vh = Vh[:rank, :]
+
+    dist = torch.cat([U.flatten(), Vh.flatten()])
+    hi_val = torch.quantile(dist, CLAMP_QUANTILE)
+    low_val = -hi_val
+
+    U = U.clamp(low_val, hi_val)
+    Vh = Vh.clamp(low_val, hi_val)
+    if conv2d:
+        U = U.reshape(out_dim, rank, 1, 1)
+        Vh = Vh.reshape(rank, in_dim, kernel_size[0], kernel_size[1])
+    return (U, Vh)
+
+class LoraSave:
+    def __init__(self):
+        self.output_dir = folder_paths.get_output_directory()
+
+    @classmethod
+    def INPUT_TYPES(s):
+        return {"required": {"filename_prefix": ("STRING", {"default": "loras/ComfyUI_extracted_lora"}),
+                              "rank": ("INT", {"default": 8, "min": 1, "max": 1024, "step": 1}),
+                            },
+                "optional": {"model_diff": ("MODEL",),},
+    }
+    RETURN_TYPES = ()
+    FUNCTION = "save"
+    OUTPUT_NODE = True
+
+    CATEGORY = "_for_testing"
+
+    def save(self, filename_prefix, rank, model_diff=None):
+        if model_diff is None:
+            return {}
+
+        full_output_folder, filename, counter, subfolder, filename_prefix = folder_paths.get_save_image_path(filename_prefix, self.output_dir)
+
+        output_sd = {}
+        prefix_key = "diffusion_model."
+        stored = set()
+
+        comfy.model_management.load_models_gpu([model_diff], force_patch_weights=True)
+        sd = model_diff.model_state_dict(filter_prefix=prefix_key)
+
+        for k in sd:
+            if k.endswith(".weight"):
+                weight_diff = sd[k]
+                if weight_diff.ndim < 2:
+                    continue
+                try:
+                    out = extract_lora(weight_diff, rank)
+                    output_sd["{}.lora_up.weight".format(k[:-7])] = out[0].contiguous().half().cpu()
+                    output_sd["{}.lora_down.weight".format(k[:-7])] = out[1].contiguous().half().cpu()
+                except:
+                    logging.warning("Could not generate lora weights for key {}, is the weight difference a zero?".format(k))
+
+        output_checkpoint = f"{filename}_{counter:05}_.safetensors"
+        output_checkpoint = os.path.join(full_output_folder, output_checkpoint)
+
+        comfy.utils.save_torch_file(output_sd, output_checkpoint, metadata=None)
+        return {}
+
+NODE_CLASS_MAPPINGS = {
+    "LoraSave": LoraSave
+}

main.py

@@ -247,6 +247,7 @@ if __name__ == "__main__":
     folder_paths.add_model_folder_path("clip", os.path.join(folder_paths.get_output_directory(), "clip"))
     folder_paths.add_model_folder_path("vae", os.path.join(folder_paths.get_output_directory(), "vae"))
     folder_paths.add_model_folder_path("diffusion_models", os.path.join(folder_paths.get_output_directory(), "diffusion_models"))
+    folder_paths.add_model_folder_path("loras", os.path.join(folder_paths.get_output_directory(), "loras"))
 
     if args.input_directory:
         input_dir = os.path.abspath(args.input_directory)

nodes.py

@@ -2101,6 +2101,7 @@ def init_builtin_extra_nodes():
         "nodes_controlnet.py",
         "nodes_hunyuan.py",
         "nodes_flux.py",
+        "nodes_lora_extract.py",
     ]
 
     import_failed = []