theluyuan/DiffSynth-Studio
1
0
Fork 0
mirror of https://github.com/modelscope/DiffSynth-Studio.git synced 2026-03-18 22:08:13 +00:00
Code Issues Packages Projects Releases Wiki Activity

Qwen image split training Bug Fix (#926)

Browse Source 
* bugfix
This commit is contained in:
Zhongjie Duan
2025-09-17 20:53:46 +08:00
committed by GitHub
parent e9f29bc402
commit b0abdaffb4
3 changed files with 10 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

5
diffsynth/pipelines/qwen_image.py
View File

@@ -140,8 +140,9 @@ class QwenImagePipeline(BasePipeline):
timestep_id = torch.randint(0, self.scheduler.num_train_timesteps, (1,)) timestep_id = torch.randint(0, self.scheduler.num_train_timesteps, (1,))
timestep = self.scheduler.timesteps[timestep_id].to(dtype=self.torch_dtype, device=self.device) timestep = self.scheduler.timesteps[timestep_id].to(dtype=self.torch_dtype, device=self.device)
inputs["latents"] = self.scheduler.add_noise(inputs["input_latents"], inputs["noise"], timestep) noise = torch.randn_like(inputs["input_latents"])
training_target = self.scheduler.training_target(inputs["input_latents"], inputs["noise"], timestep) inputs["latents"] = self.scheduler.add_noise(inputs["input_latents"], noise, timestep)
training_target = self.scheduler.training_target(inputs["input_latents"], noise, timestep)
noise_pred = self.model_fn(**inputs, timestep=timestep) noise_pred = self.model_fn(**inputs, timestep=timestep)

4
diffsynth/trainers/utils.py
View File

@@ -421,10 +421,12 @@ class DiffusionTrainingModule(torch.nn.Module):
return state_dict return state_dict
def transfer_data_to_device(self, data, device): def transfer_data_to_device(self, data, device, torch_float_dtype=None):
for key in data: for key in data:
if isinstance(data[key], torch.Tensor): if isinstance(data[key], torch.Tensor):
data[key] = data[key].to(device) data[key] = data[key].to(device)
if torch_float_dtype is not None and data[key].dtype in [torch.float, torch.float16, torch.bfloat16]:
data[key] = data[key].to(torch_float_dtype)
return data return data

6
examples/qwen_image/model_training/train.py
View File

@@ -85,8 +85,10 @@ class QwenImageTrainingModule(DiffusionTrainingModule):
def forward(self, data, inputs=None, return_inputs=False): def forward(self, data, inputs=None, return_inputs=False):
# Inputs # Inputs
if inputs is None: inputs = self.forward_preprocess(data) if inputs is None:
else: inputs = self.transfer_data_to_device(inputs, self.pipe.device) inputs = self.forward_preprocess(data)
else:
inputs = self.transfer_data_to_device(inputs, self.pipe.device, self.pipe.torch_dtype)
if return_inputs: return inputs if return_inputs: return inputs
# Loss # Loss