diffsynth 2.1.3__tar.gz → 2.1.5__tar.gz

{diffsynth-2.1.3 → diffsynth-2.1.5}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: diffsynth
-Version: 2.1.3
+Version: 2.1.5
 Summary: Enjoy the magic of Diffusion models!
 Author: ModelScope Team
 License-Expression: Apache-2.0
@@ -107,6 +107,8 @@ See also:
 
 > Currently, the development personnel of this project are limited, with most of the work handled by [Artiprocher](https://github.com/Artiprocher) and [mi804](https://github.com/mi804). Therefore, the progress of new feature development will be relatively slow, and the speed of responding to and resolving issues is limited. We apologize for this and ask developers to understand.
 
+- **August 25, 2026** We have open-sourced [DiffSynth-WebUI](https://github.com/modelscope/DiffSynth-WebUI), enabling one-click private deployment of LoRA training services. Combined with the model quantization feature, you can train large models even with consumer-grade GPUs.
+
 - **August 19, 2026** We have released the model quantization feature. It provides a unified `QuantizeConfig` entry point supporting multiple quantization backends including bitsandbytes, torchao, and comfy-kitchen, with capabilities such as online quantization, loading pre-quantized weights, mixed quantization, saving quantized models, and quantization + LoRA training. For details, please refer to the [documentation](/docs/en/Pipeline_Usage/Quantization.md).
 
 - **August 17, 2026** MiniMax-Music3 open-sourced, welcome a new member to the audio model family! Support includes text-to-music generation and low VRAM inference. For details, please refer to the [documentation](/docs/en/Model_Details/MiniMax-Music3.md) and [example code](/examples/minimax_music3/).
@@ -802,6 +804,9 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
 | FLUX.1 | [black-forest-labs/FLUX.1-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-dev) | [code](/examples/flux/model_inference/FLUX.1-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev.py) |
 | FLUX.1 | [black-forest-labs/FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) | [code](/examples/flux/model_inference/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Krea-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Krea-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Krea-dev.py) |
 | FLUX.1 | [black-forest-labs/FLUX.1-Kontext-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Kontext-dev) | [code](/examples/flux/model_inference/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Kontext-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Kontext-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Kontext-dev.py) |
+| FLUX.1 | [black-forest-labs/FLUX.1-Fill-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Fill-dev) | [code](/examples/flux/model_inference/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Fill-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Fill-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Fill-dev.py) |
+| FLUX.1 | [black-forest-labs/FLUX.1-Redux-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Redux-dev) | [code](/examples/flux/model_inference/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Redux-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Redux-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Redux-dev.py) |
+| FLUX.1 | [HuanJue/Insert-Anything](https://www.modelscope.cn/models/HuanJue/Insert-Anything) | [code](/examples/flux/model_inference/Insert-Anything.py) | [code](/examples/flux/model_inference_low_vram/Insert-Anything.py) | - | - | [code](/examples/flux/model_training/lora/Insert-Anything.sh) | [code](/examples/flux/model_training/validate_lora/Insert-Anything.py) |
 | FLUX.1 | [alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta](https://www.modelscope.cn/models/alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Inpainting-Beta.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Inpainting-Beta.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Inpainting-Beta.py) |
 | FLUX.1 | [InstantX/FLUX.1-dev-Controlnet-Union-alpha](https://www.modelscope.cn/models/InstantX/FLUX.1-dev-Controlnet-Union-alpha) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Union-alpha.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Union-alpha.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Union-alpha.py) |
 | FLUX.1 | [jasperai/Flux.1-dev-Controlnet-Upscaler](https://www.modelscope.cn/models/jasperai/Flux.1-dev-Controlnet-Upscaler) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Upscaler.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Upscaler.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Upscaler.py) |

{diffsynth-2.1.3 → diffsynth-2.1.5}/README.md

@@ -40,6 +40,8 @@ See also:
 
 > Currently, the development personnel of this project are limited, with most of the work handled by [Artiprocher](https://github.com/Artiprocher) and [mi804](https://github.com/mi804). Therefore, the progress of new feature development will be relatively slow, and the speed of responding to and resolving issues is limited. We apologize for this and ask developers to understand.
 
+- **August 25, 2026** We have open-sourced [DiffSynth-WebUI](https://github.com/modelscope/DiffSynth-WebUI), enabling one-click private deployment of LoRA training services. Combined with the model quantization feature, you can train large models even with consumer-grade GPUs.
+
 - **August 19, 2026** We have released the model quantization feature. It provides a unified `QuantizeConfig` entry point supporting multiple quantization backends including bitsandbytes, torchao, and comfy-kitchen, with capabilities such as online quantization, loading pre-quantized weights, mixed quantization, saving quantized models, and quantization + LoRA training. For details, please refer to the [documentation](/docs/en/Pipeline_Usage/Quantization.md).
 
 - **August 17, 2026** MiniMax-Music3 open-sourced, welcome a new member to the audio model family! Support includes text-to-music generation and low VRAM inference. For details, please refer to the [documentation](/docs/en/Model_Details/MiniMax-Music3.md) and [example code](/examples/minimax_music3/).
@@ -735,6 +737,9 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
 | FLUX.1 | [black-forest-labs/FLUX.1-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-dev) | [code](/examples/flux/model_inference/FLUX.1-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev.py) |
 | FLUX.1 | [black-forest-labs/FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) | [code](/examples/flux/model_inference/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Krea-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Krea-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Krea-dev.py) |
 | FLUX.1 | [black-forest-labs/FLUX.1-Kontext-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Kontext-dev) | [code](/examples/flux/model_inference/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Kontext-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Kontext-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Kontext-dev.py) |
+| FLUX.1 | [black-forest-labs/FLUX.1-Fill-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Fill-dev) | [code](/examples/flux/model_inference/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Fill-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Fill-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Fill-dev.py) |
+| FLUX.1 | [black-forest-labs/FLUX.1-Redux-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Redux-dev) | [code](/examples/flux/model_inference/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Redux-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Redux-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Redux-dev.py) |
+| FLUX.1 | [HuanJue/Insert-Anything](https://www.modelscope.cn/models/HuanJue/Insert-Anything) | [code](/examples/flux/model_inference/Insert-Anything.py) | [code](/examples/flux/model_inference_low_vram/Insert-Anything.py) | - | - | [code](/examples/flux/model_training/lora/Insert-Anything.sh) | [code](/examples/flux/model_training/validate_lora/Insert-Anything.py) |
 | FLUX.1 | [alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta](https://www.modelscope.cn/models/alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Inpainting-Beta.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Inpainting-Beta.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Inpainting-Beta.py) |
 | FLUX.1 | [InstantX/FLUX.1-dev-Controlnet-Union-alpha](https://www.modelscope.cn/models/InstantX/FLUX.1-dev-Controlnet-Union-alpha) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Union-alpha.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Union-alpha.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Union-alpha.py) |
 | FLUX.1 | [jasperai/Flux.1-dev-Controlnet-Upscaler](https://www.modelscope.cn/models/jasperai/Flux.1-dev-Controlnet-Upscaler) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Upscaler.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Upscaler.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Upscaler.py) |

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/core/data/operators.py

@@ -307,8 +307,8 @@ class LoadAudioWithTorchaudio(DataProcessingOperator, FrameSamplerByRateMixin):
                 padding = target_samples - current_samples
                 waveform = torch.nn.functional.pad(waveform, (0, padding))
             return waveform, sample_rate
-        except:
-            warnings.warn(f"Cannot load audio in {data}. The audio will be `None`.")
+        except Exception as e:
+            warnings.warn(f"Cannot load audio in {data} due to {e}. The audio will be `None`.")
             return None
 
 

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/core/loader/config.py

@@ -75,11 +75,11 @@ class ModelConfig:
     def parse_download_source(self):
         if self.download_source is None:
             if os.environ.get('DIFFSYNTH_DOWNLOAD_SOURCE') is not None:
-                return os.environ.get('DIFFSYNTH_DOWNLOAD_SOURCE')
+                return os.environ.get('DIFFSYNTH_DOWNLOAD_SOURCE').lower()
             else:
                 return "modelscope"
         else:
-            return self.download_source
+            return self.download_source.lower()
         
     def parse_skip_download(self):
         if self.skip_download is None:

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/core/loader/file.py

@@ -1,5 +1,5 @@
 from safetensors import safe_open
-import torch, hashlib
+import torch, hashlib, json
 
 
 def load_state_dict(file_path, torch_dtype=None, device="cpu", pin_memory=False, verbose=0):
@@ -34,10 +34,26 @@ def load_state_dict_from_safetensors(file_path, torch_dtype=None, device="cpu"):
 
 
 def load_metadata_from_safetensors(file_path):
+    if isinstance(file_path, list):
+        metadata = {}
+        for file_path_ in file_path:
+            for key, value in load_metadata_from_safetensors(file_path_).items():
+                metadata[key] = merge_metadata_value(metadata[key], value) if key in metadata else value
+        return metadata
     with safe_open(file_path, framework="pt", device="cpu") as f:
         return f.metadata() or {}
 
 
+def merge_metadata_value(old_value, new_value):
+    try:
+        old_items, new_items = json.loads(old_value), json.loads(new_value)
+    except (json.JSONDecodeError, TypeError):
+        return new_value
+    if isinstance(old_items, list) and isinstance(new_items, list):
+        return json.dumps(old_items + [i for i in new_items if i not in old_items])
+    return new_value
+
+
 def load_state_dict_from_bin(file_path, torch_dtype=None, device="cpu"):
     state_dict = torch.load(file_path, map_location=device, weights_only=True)
     if len(state_dict) == 1:

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/core/loader/model.py

@@ -37,7 +37,7 @@ def load_model(model_class, path, config=None, torch_dtype=torch.bfloat16, devic
             devices = [vram_config[k] for k in ("offload_device", "onload_device", "preparing_device", "computation_device")]
             load_device = [d for d in devices if d != "disk"][0]
             disk_map = DiskMap(path, load_device, torch_dtype=None, state_dict_converter=state_dict_converter)
-            metadata = load_metadata_from_safetensors(path[0] if isinstance(path, list) else path)
+            metadata = load_metadata_from_safetensors(path)
             model = quantize.prepare_for_prequantized_load(model, compute_dtype=vram_config["computation_dtype"])
             model = enable_vram_management(model, module_map, vram_config=vram_config, disk_map=disk_map, vram_limit=vram_limit, quantize=quantize, metadata=metadata)
         else:
@@ -119,6 +119,9 @@ def load_model(model_class, path, config=None, torch_dtype=torch.bfloat16, devic
         # Because some models override the behavior of `to()`,
         # especially those from libraries like Transformers.
         model = model.to(dtype=torch_dtype, device=device)
+    if quantize is not None:
+        # Downstream steps (e.g. LoRA hot-loading) need the config to handle the quantized layers.
+        model.quantize_config = quantize
     if hasattr(model, "eval"):
         model = model.eval()
     return model

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/core/quant/backends/bitsandbytes.py

@@ -11,26 +11,27 @@ except ImportError:
     BITSANDBYTES_AVAILABLE = False
 
 
-def _assign_params_from_state_dict(self, state_dict, prefix, local_metadata, strict, missing_keys, unexpected_keys, error_msgs):
-    """`_load_from_state_dict` replacement for shells: assigns tensors as-is, since packed weights do not pass the standard shape check."""
-    local_names = set()
-    for name, current in list(self._parameters.items()) + list(self._buffers.items()):
-        if current is None:
-            continue
-        local_names.add(name)
-        key = prefix + name
-        if key in state_dict:
-            value = state_dict[key]
-            if name in self._parameters:
-                self._parameters[name] = value if isinstance(value, torch.nn.Parameter) else torch.nn.Parameter(value, requires_grad=False)
-            else:
-                self._buffers[name] = value
-        elif strict:
-            missing_keys.append(key)
-    if strict:
-        for key in state_dict:
-            if key.startswith(prefix) and key[len(prefix):].split(".", 1)[0] not in local_names:
-                unexpected_keys.append(key)
+if BITSANDBYTES_AVAILABLE:
+    class BitsAndBytesLinear4bit(bnb.nn.Linear4bit):
+        def _load_from_state_dict(self, state_dict, prefix, local_metadata, strict, missing_keys, unexpected_keys, error_msgs):
+            local_names = set()
+            for name, current in list(self._parameters.items()) + list(self._buffers.items()):
+                if current is None:
+                    continue
+                local_names.add(name)
+                key = prefix + name
+                if key in state_dict:
+                    value = state_dict[key]
+                    if name in self._parameters:
+                        self._parameters[name] = value if isinstance(value, torch.nn.Parameter) else torch.nn.Parameter(value, requires_grad=False)
+                    else:
+                        self._buffers[name] = value
+                elif strict:
+                    missing_keys.append(key)
+            if strict:
+                for key in state_dict:
+                    if key.startswith(prefix) and key[len(prefix):].split(".", 1)[0] not in local_names:
+                        unexpected_keys.append(key)
 
 
 @register_quant_backend("bitsandbytes")
@@ -55,7 +56,7 @@ class BitsAndBytesQuantBackend(QuantBackend):
         }
 
     def quantized_linear_classes(self):
-        return (bnb.nn.Linear4bit,)
+        return (BitsAndBytesLinear4bit,)
 
     def create_quantized_linear(self, linear, compute_device=None, model_device=None):
         """The `meta` shell avoids allocating an fp weight; bnb quantizes while `Params4bit` moves to `compute_device`."""
@@ -65,7 +66,7 @@ class BitsAndBytesQuantBackend(QuantBackend):
             compute_device = linear.weight.device
 
         with torch.device("meta"):
-            quant_linear = bnb.nn.Linear4bit(
+            quant_linear = BitsAndBytesLinear4bit(
                 linear.in_features,
                 linear.out_features,
                 bias=linear.bias is not None,
@@ -89,7 +90,7 @@ class BitsAndBytesQuantBackend(QuantBackend):
 
     def create_quantized_linear_shell(self, linear, compute_dtype):
         with torch.device("meta"):
-            shell = bnb.nn.Linear4bit(
+            shell = BitsAndBytesLinear4bit(
                 linear.in_features,
                 linear.out_features,
                 bias=linear.bias is not None,
@@ -98,7 +99,6 @@ class BitsAndBytesQuantBackend(QuantBackend):
                 quant_type=self.config.quant_type,
                 quant_storage=self.config.quant_storage,
             )
-        shell._load_from_state_dict = _assign_params_from_state_dict.__get__(shell)
         return shell
 
     def unflatten_state_dict(self, state_dict, metadata):

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/core/vram/layers.py

@@ -442,7 +442,7 @@ class AutoWrappedLinear(torch.nn.Linear, AutoTorchModule, LoRAHotLoadMixin):
         return out
 
 
-class AutoWrappedQuantizedModule(AutoTorchModule, LoRAHotLoadMixin):
+class AutoWrappedQuantizedModule(torch.nn.Linear, AutoTorchModule, LoRAHotLoadMixin):
     def __init__(
         self,
         module: torch.nn.Module,
@@ -467,7 +467,8 @@ class AutoWrappedQuantizedModule(AutoTorchModule, LoRAHotLoadMixin):
                 "Disk offload for quantized layers requires both `disk_map` and `quantize`, "
                 "so each layer can rebuild its packed weight and quant state lazily."
             )
-        super().__init__(
+        AutoTorchModule.__init__(
+            self,
             offload_dtype,
             offload_device,
             onload_dtype,
@@ -478,6 +479,8 @@ class AutoWrappedQuantizedModule(AutoTorchModule, LoRAHotLoadMixin):
             computation_device,
             vram_limit,
         )
+        self.in_features = module.in_features
+        self.out_features = module.out_features
         self.module = module
         self.name = name
         self.disk_offload = disk_offload

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/diffusion/base_pipeline.py

@@ -240,6 +240,13 @@ class BasePipeline(torch.nn.Module):
                 module.computation_device = device
                 
     
+    def check_quant_hot_load(self, module: torch.nn.Module):
+        # A quantized weight cannot absorb a fused LoRA, so hot-loading is the only option for it.
+        if getattr(module, "quantize_config", None) is None:
+            return module
+        return self.enable_lora_hot_loading(module)
+
+
     def load_lora(
         self,
         module: torch.nn.Module,
@@ -249,6 +256,7 @@ class BasePipeline(torch.nn.Module):
         state_dict=None,
         verbose=1,
     ):
+        module = self.check_quant_hot_load(module)
         if state_dict is None:
             if isinstance(lora_config, str):
                 lora = load_state_dict(lora_config, torch_dtype=self.torch_dtype, device=self.device)
@@ -370,7 +378,7 @@ class BasePipeline(torch.nn.Module):
             "computation_dtype": self.torch_dtype,
             "computation_device": self.device,
         }
-        model = enable_vram_management(model, module_map, vram_config=vram_config)
+        model = enable_vram_management(model, module_map, vram_config=vram_config, quantize=getattr(model, "quantize_config", None))
         return model
 
     def compile_pipeline(self, mode: str = "default", dynamic: bool = True, fullgraph: bool = False, compile_models: list = None, **kwargs):

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/diffusion/flow_match.py

@@ -302,7 +302,7 @@ class FlowMatchScheduler():
         num_train_timesteps = 1000
         if special_case == "stage2":
             sigmas = torch.Tensor([0.909375, 0.725, 0.421875])
-        elif special_case == "ditilled_stage1":
+        elif special_case == "distilled_stage1":
             sigmas = torch.Tensor([1.0, 0.99375, 0.9875, 0.98125, 0.975, 0.909375, 0.725, 0.421875])
         else:
             dynamic_shift_len = dynamic_shift_len or 4096

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/diffusion/loss.py

@@ -63,7 +63,7 @@ def FlowMatchSFTAudioVideoLoss(pipe: BasePipeline, **inputs):
     return loss
 
 
-def FlowMatchSFTMiniMaxH3AudioVideoLoss(pipe: BasePipeline, **inputs):
+def FlowMatchSFTMiniMaxH3AudioVideoLoss(pipe: BasePipeline, training_cfg_scale: float = 1.0, inputs_nega: dict | None = None, **inputs):
     max_timestep_boundary = int(inputs.get("max_timestep_boundary", 1) * len(pipe.scheduler.timesteps))
     min_timestep_boundary = int(inputs.get("min_timestep_boundary", 0) * len(pipe.scheduler.timesteps))
 
@@ -81,11 +81,34 @@ def FlowMatchSFTMiniMaxH3AudioVideoLoss(pipe: BasePipeline, **inputs):
         training_target_audio = pipe.scheduler_audio.training_target(inputs["audio_input_latents"], audio_noise, timestep_audio)
 
     models = {name: getattr(pipe, name) for name in pipe.in_iteration_models}
+    if training_cfg_scale > 1.0:
+        if not inputs_nega:
+            raise ValueError(
+                "MiniMax-H3 CFG-aware training requires unconditional inputs. "
+                "When using split training, rebuild the data cache with the same "
+                "--training_cfg_scale value."
+            )
+        inputs_uncond = {**inputs, **inputs_nega}
+        inputs_uncond["use_gradient_checkpointing"] = False
+        inputs_uncond["use_gradient_checkpointing_offload"] = False
+        with torch.no_grad():
+            noise_pred_uncond, noise_pred_audio_uncond = pipe.model_fn(
+                **models, **inputs_uncond,
+                timestep_video=timestep_video, timestep_audio=timestep_audio,
+            )
+
     noise_pred, noise_pred_audio = pipe.model_fn(
         **models, **inputs,
         timestep_video=timestep_video, timestep_audio=timestep_audio,
     )
 
+    if training_cfg_scale > 1.0:
+        # The checkpoint's conditional prediction has CFG distilled into it.
+        # Rearrange the CFG equation to recover the raw velocity fitted to the
+        # standard flow-matching target, using the current model as the teacher.
+        noise_pred = (noise_pred + (training_cfg_scale - 1.0) * noise_pred_uncond) / training_cfg_scale
+        noise_pred_audio = (noise_pred_audio + (training_cfg_scale - 1.0) * noise_pred_audio_uncond) / training_cfg_scale
+
     loss = torch.nn.functional.mse_loss(noise_pred.float(), training_target.float())
     loss = loss * pipe.scheduler.training_weight(timestep_video)
     if "audio_input_latents" in inputs:

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/diffusion/training_module.py

@@ -193,9 +193,25 @@ class DiffusionTrainingModule(torch.nn.Module):
             quant_config = QuantizeConfig(method=method, exclude_modules=exclude_modules)
             if not quant_config.backend.capabilities().get("is_differentiable", True):
                 raise ValueError(f"Quantization method `{method}` is not differentiable, so it cannot be used for training (frozen quantized layers must pass gradients through to LoRA branches). Choose a method whose backend declares `is_differentiable=True`.")
-            quant_map[model_string] = quant_config
+            quant_map[self.normalize_quant_key(model_string)] = quant_config
         return quant_map
 
+    def normalize_quant_key(self, model_string):
+        if isinstance(model_string, str) and model_string.startswith("[") and model_string.endswith("]"):
+            try:
+                model_string = json.loads(model_string)
+            except json.JSONDecodeError:
+                raise ValueError(f"Failed to parse quant option model string: `{model_string}`. A multi-file model must be written as a JSON list of files, matching its `--model_paths` entry.")
+        if isinstance(model_string, list):
+            return tuple(model_string)
+        return model_string
+
+    def get_quant_config(self, quant_map, model_string):
+        quant_config = quant_map.get(self.normalize_quant_key(model_string))
+        if quant_config is None and len(quant_map) > 0:
+            print(f"No quant option matches `{model_string}`. This model is loaded without quantization. Parsed quant options: {({key: config.method for key, config in quant_map.items()})}.")
+        return quant_config
+
     def parse_model_configs(self, model_paths, model_id_with_origin_paths, fp8_models=None, offload_models=None, quant_options=None, device="cpu"):
         fp8_models = [] if fp8_models is None else fp8_models.split(",")
         offload_models = [] if offload_models is None else offload_models.split(",")
@@ -209,7 +225,7 @@ class DiffusionTrainingModule(torch.nn.Module):
                     offload=path in offload_models,
                     device=device
                 )
-                model_configs.append(ModelConfig(path=path, quantize=quant_map.get(path), **vram_config))
+                model_configs.append(ModelConfig(path=path, quantize=self.get_quant_config(quant_map, path), **vram_config))
         if model_id_with_origin_paths is not None:
             model_id_with_origin_paths = model_id_with_origin_paths.split(",")
             for model_id_with_origin_path in model_id_with_origin_paths:
@@ -219,7 +235,7 @@ class DiffusionTrainingModule(torch.nn.Module):
                     device=device
                 )
                 config = self.parse_path_or_model_id(model_id_with_origin_path)
-                model_configs.append(ModelConfig(model_id=config.model_id, origin_file_pattern=config.origin_file_pattern, quantize=quant_map.get(model_id_with_origin_path), **vram_config))
+                model_configs.append(ModelConfig(model_id=config.model_id, origin_file_pattern=config.origin_file_pattern, quantize=self.get_quant_config(quant_map, model_id_with_origin_path), **vram_config))
         return model_configs
     
 
@@ -343,7 +359,7 @@ class DiffusionTrainingModule(torch.nn.Module):
         # TODO: set `remove_unnecessary_params` to `True` by default
         remove_unnecessary_params=False,
         # TODO: move `loss_required_params` to `loss.py`
-        loss_required_params=("input_latents", "max_timestep_boundary", "min_timestep_boundary", "first_frame_latents", "video_latents", "audio_input_latents", "num_inference_steps"),
+        loss_required_params=("input_latents", "max_timestep_boundary", "min_timestep_boundary", "first_frame_latents", "video_latents", "audio_input_latents", "num_inference_steps", "cfg_scale"),
         force_remove_params_shared=tuple(),
         force_remove_params_posi=tuple(),
         force_remove_params_nega=tuple(),

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/pipelines/ace_step.py

@@ -70,7 +70,8 @@ class AceStepPipeline(BasePipeline):
         pipe.conditioner = model_pool.fetch_model("ace_step_conditioner")
         pipe.dit = model_pool.fetch_model("ace_step_dit")
         pipe.vae = model_pool.fetch_model("ace_step_vae")
-        pipe.vae.remove_weight_norm()
+        if pipe.vae is not None:
+            pipe.vae.remove_weight_norm()
         pipe.tokenizer_model = model_pool.fetch_model("ace_step_tokenizer")
 
         if text_tokenizer_config is not None:

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/pipelines/ltx2_audio_video.py

@@ -213,7 +213,7 @@ class LTX2AudioVideoPipeline(BasePipeline):
         progress_bar_cmd=tqdm,
     ):
         # Scheduler
-        self.scheduler.set_timesteps(num_inference_steps, denoising_strength=denoising_strength, special_case="ditilled_stage1" if use_distilled_pipeline else None)
+        self.scheduler.set_timesteps(num_inference_steps, denoising_strength=denoising_strength, special_case="distilled_stage1" if use_distilled_pipeline else None)
         # Inputs
         inputs_posi = {
             "prompt": prompt,

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/pipelines/minimax_h3_audio_video.py

@@ -58,7 +58,21 @@ class MiniMaxH3Pipeline(BasePipeline):
         model_configs: list[ModelConfig] = [],
         processor_config: ModelConfig = ModelConfig(model_id="MiniMax/MiniMax-H3", origin_file_pattern="FL2VA/processor/"),
         vram_limit: float = None,
+        redirect_common_files: bool = True,
     ):
+        if redirect_common_files:
+            redirect_dict = {
+                "MiniMax/MiniMax-H3": "MiniMaxAI/MiniMax-H3",
+            }
+            for model_config in model_configs:
+                if model_config.require_downloading() and model_config.parse_download_source() == "huggingface":
+                    if model_config.model_id is not None and model_config.model_id in redirect_dict:
+                        print(f"The model is detected to be downloading from HuggingFace. {model_config.model_id} is redirected to {redirect_dict[model_config.model_id]}. You can use `redirect_common_files=False` to disable file redirection.")
+                        model_config.model_id = redirect_dict[model_config.model_id]
+            if processor_config is not None and processor_config.require_downloading() and processor_config.parse_download_source() == "huggingface":
+                if processor_config.model_id is not None and processor_config.model_id in redirect_dict:
+                    print(f"The model is detected to be downloading from HuggingFace. {processor_config.model_id} is redirected to {redirect_dict[processor_config.model_id]}. You can use `redirect_common_files=False` to disable file redirection.")
+                    processor_config.model_id = redirect_dict[processor_config.model_id]
         pipe = MiniMaxH3Pipeline(device=device, torch_dtype=torch_dtype)
         model_pool = pipe.download_and_load_models(model_configs, vram_limit)
         pipe.text_encoder = model_pool.fetch_model("minimax_h3_text_encoder")

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/pipelines/minimax_music3.py

@@ -45,7 +45,21 @@ class MiniMaxMusic3Pipeline(BasePipeline):
         model_configs: list[ModelConfig] = [],
         tokenizer_config: ModelConfig = None,
         vram_limit: float = None,
+        redirect_common_files: bool = True,
     ):
+        if redirect_common_files:
+            redirect_dict = {
+                "MiniMax/MiniMax-Music3": "MiniMaxAI/MiniMax-Music3",
+            }
+            for model_config in model_configs:
+                if model_config.require_downloading() and model_config.parse_download_source() == "huggingface":
+                    if model_config.model_id is not None and model_config.model_id in redirect_dict:
+                        print(f"The model is detected to be downloading from HuggingFace. {model_config.model_id} is redirected to {redirect_dict[model_config.model_id]}. You can use `redirect_common_files=False` to disable file redirection.")
+                        model_config.model_id = redirect_dict[model_config.model_id]
+            if tokenizer_config is not None and tokenizer_config.require_downloading() and tokenizer_config.parse_download_source() == "huggingface":
+                if tokenizer_config.model_id is not None and tokenizer_config.model_id in redirect_dict:
+                    print(f"The model is detected to be downloading from HuggingFace. {tokenizer_config.model_id} is redirected to {redirect_dict[tokenizer_config.model_id]}. You can use `redirect_common_files=False` to disable file redirection.")
+                    tokenizer_config.model_id = redirect_dict[tokenizer_config.model_id]
         pipe = MiniMaxMusic3Pipeline(device=device, torch_dtype=torch_dtype)
         model_pool = pipe.download_and_load_models(model_configs, vram_limit)
         pipe.text_encoder = model_pool.fetch_model("minimax_music3_text_encoder")
@@ -348,4 +362,4 @@ class MiniMaxMusic3Unit_Vocoder(PipelineUnit):
             waveform_chunks.append(waveform[...,
                                             left:waveform.shape[-1] - right])
         song = torch.cat(waveform_chunks, dim=-1)[0]
-        return {"audio": song.float().clamp(-1.0, 1.0)}
+        return {"audio": song.float().clamp(-1.0, 1.0).cpu()}

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth/version.py

@@ -1,5 +1,5 @@
 # Make sure to modify __release_datetime__ to release time when making official release.
-__version__ = '2.1.3'
+__version__ = '2.1.5'
 # default release datetime for branches under active development is set
 # to be a time far-far-away-into-the-future
 __release_datetime__ = '2099-10-13 08:56:12'

{diffsynth-2.1.3 → diffsynth-2.1.5}/diffsynth.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: diffsynth
-Version: 2.1.3
+Version: 2.1.5
 Summary: Enjoy the magic of Diffusion models!
 Author: ModelScope Team
 License-Expression: Apache-2.0
@@ -107,6 +107,8 @@ See also:
 
 > Currently, the development personnel of this project are limited, with most of the work handled by [Artiprocher](https://github.com/Artiprocher) and [mi804](https://github.com/mi804). Therefore, the progress of new feature development will be relatively slow, and the speed of responding to and resolving issues is limited. We apologize for this and ask developers to understand.
 
+- **August 25, 2026** We have open-sourced [DiffSynth-WebUI](https://github.com/modelscope/DiffSynth-WebUI), enabling one-click private deployment of LoRA training services. Combined with the model quantization feature, you can train large models even with consumer-grade GPUs.
+
 - **August 19, 2026** We have released the model quantization feature. It provides a unified `QuantizeConfig` entry point supporting multiple quantization backends including bitsandbytes, torchao, and comfy-kitchen, with capabilities such as online quantization, loading pre-quantized weights, mixed quantization, saving quantized models, and quantization + LoRA training. For details, please refer to the [documentation](/docs/en/Pipeline_Usage/Quantization.md).
 
 - **August 17, 2026** MiniMax-Music3 open-sourced, welcome a new member to the audio model family! Support includes text-to-music generation and low VRAM inference. For details, please refer to the [documentation](/docs/en/Model_Details/MiniMax-Music3.md) and [example code](/examples/minimax_music3/).
@@ -802,6 +804,9 @@ https://github.com/Artiprocher/DiffSynth-Studio/assets/35051019/59fb2f7b-8de0-44
 | FLUX.1 | [black-forest-labs/FLUX.1-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-dev) | [code](/examples/flux/model_inference/FLUX.1-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev.py) |
 | FLUX.1 | [black-forest-labs/FLUX.1-Krea-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Krea-dev) | [code](/examples/flux/model_inference/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Krea-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Krea-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Krea-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Krea-dev.py) |
 | FLUX.1 | [black-forest-labs/FLUX.1-Kontext-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Kontext-dev) | [code](/examples/flux/model_inference/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Kontext-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Kontext-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Kontext-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Kontext-dev.py) |
+| FLUX.1 | [black-forest-labs/FLUX.1-Fill-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Fill-dev) | [code](/examples/flux/model_inference/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Fill-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Fill-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Fill-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Fill-dev.py) |
+| FLUX.1 | [black-forest-labs/FLUX.1-Redux-dev](https://www.modelscope.cn/models/black-forest-labs/FLUX.1-Redux-dev) | [code](/examples/flux/model_inference/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_training/full/FLUX.1-Redux-dev.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-Redux-dev.py) | [code](/examples/flux/model_training/lora/FLUX.1-Redux-dev.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-Redux-dev.py) |
+| FLUX.1 | [HuanJue/Insert-Anything](https://www.modelscope.cn/models/HuanJue/Insert-Anything) | [code](/examples/flux/model_inference/Insert-Anything.py) | [code](/examples/flux/model_inference_low_vram/Insert-Anything.py) | - | - | [code](/examples/flux/model_training/lora/Insert-Anything.sh) | [code](/examples/flux/model_training/validate_lora/Insert-Anything.py) |
 | FLUX.1 | [alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta](https://www.modelscope.cn/models/alimama-creative/FLUX.1-dev-Controlnet-Inpainting-Beta) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Inpainting-Beta.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Inpainting-Beta.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Inpainting-Beta.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Inpainting-Beta.py) |
 | FLUX.1 | [InstantX/FLUX.1-dev-Controlnet-Union-alpha](https://www.modelscope.cn/models/InstantX/FLUX.1-dev-Controlnet-Union-alpha) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Union-alpha.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Union-alpha.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Union-alpha.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Union-alpha.py) |
 | FLUX.1 | [jasperai/Flux.1-dev-Controlnet-Upscaler](https://www.modelscope.cn/models/jasperai/Flux.1-dev-Controlnet-Upscaler) | [code](/examples/flux/model_inference/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_inference_low_vram/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_training/full/FLUX.1-dev-Controlnet-Upscaler.sh) | [code](/examples/flux/model_training/validate_full/FLUX.1-dev-Controlnet-Upscaler.py) | [code](/examples/flux/model_training/lora/FLUX.1-dev-Controlnet-Upscaler.sh) | [code](/examples/flux/model_training/validate_lora/FLUX.1-dev-Controlnet-Upscaler.py) |