Merge branch 'sd3' into sd3_5_support

README.md

@@ -953,6 +953,13 @@ https://github.com/kohya-ss/sd-scripts/pull/1290) frodo821 氏に感謝します
 - `gen_imgs.py` のプロンプトオプションに、保存時のファイル名を指定する `--f` オプションを追加しました。また同スクリプトで Diffusers ベースのキーを持つ LoRA の重みに対応しました。
 
 
+### Oct 27, 2024 / 2024-10-27:
+
+- `svd_merge_lora.py` VRAM usage has been reduced. However, main memory usage will increase (32GB is sufficient).
+- This will be included in the next release.
+- `svd_merge_lora.py` のVRAM使用量を削減しました。ただし、メインメモリの使用量は増加します（32GBあれば十分です）。
+- これは次回リリースに含まれます。
+
 ### Oct 26, 2024 / 2024-10-26: 
 
 - Fixed a bug in `svd_merge_lora.py`, `sdxl_merge_lora.py`, and `resize_lora.py` where the hash value of LoRA metadata was not correctly calculated when the `save_precision` was different from the  `precision` used in the calculation. See issue [#1722](https://github.com/kohya-ss/sd-scripts/pull/1722) for details. Thanks to JujoHotaru for raising the issue.

networks/svd_merge_lora.py

@@ -301,10 +301,10 @@ def merge_lora_models(models, ratios, lbws, new_rank, new_conv_rank, device, mer
             # make original weight if not exist
             if lora_module_name not in merged_sd:
                 weight = torch.zeros((out_dim, in_dim, *kernel_size) if conv2d else (out_dim, in_dim), dtype=merge_dtype)
-                if device:
-                    weight = weight.to(device)
             else:
                 weight = merged_sd[lora_module_name]
+            if device:
+                weight = weight.to(device)
 
             # merge to weight
             if device:
@@ -336,13 +336,16 @@ def merge_lora_models(models, ratios, lbws, new_rank, new_conv_rank, device, mer
                 conved = torch.nn.functional.conv2d(down_weight.permute(1, 0, 2, 3), up_weight).permute(1, 0, 2, 3)
                 weight = weight + ratio * conved * scale
 
-            merged_sd[lora_module_name] = weight
+            merged_sd[lora_module_name] = weight.to("cpu")
 
     # extract from merged weights
     logger.info("extract new lora...")
     merged_lora_sd = {}
     with torch.no_grad():
         for lora_module_name, mat in tqdm(list(merged_sd.items())):
+            if device:
+                mat = mat.to(device)
+
             conv2d = len(mat.size()) == 4
             kernel_size = None if not conv2d else mat.size()[2:4]
             conv2d_3x3 = conv2d and kernel_size != (1, 1)
@@ -381,7 +384,7 @@ def merge_lora_models(models, ratios, lbws, new_rank, new_conv_rank, device, mer
 
             merged_lora_sd[lora_module_name + ".lora_up.weight"] = up_weight.to("cpu").contiguous()
             merged_lora_sd[lora_module_name + ".lora_down.weight"] = down_weight.to("cpu").contiguous()
-            merged_lora_sd[lora_module_name + ".alpha"] = torch.tensor(module_new_rank)
+            merged_lora_sd[lora_module_name + ".alpha"] = torch.tensor(module_new_rank, device="cpu")
 
     # build minimum metadata
     dims = f"{new_rank}"