Update cldm/cldm.py

cldm/cldm.py

@@ -320,14 +320,14 @@ class ControlLDM(LatentDiffusion):
     @torch.no_grad()
     def get_input(self, batch, k, bs=None, *args, **kwargs):
         # x, c = super().get_input(batch, self.first_stage_key, *args, **kwargs)
-        x, c, t = super().get_input(batch, self.first_stage_key, *args, **kwargs) # 接收t
+        x, c, t = super().get_input(batch, self.first_stage_key, *args, **kwargs) 
         control = batch[self.control_key]
         if bs is not None:
             control = control[:bs]
         control = control.to(self.device)
         control = einops.rearrange(control, 'b h w c -> b c h w')
         control = control.to(memory_format=torch.contiguous_format).float()
-        # 设置条件；根据返回的t；设置输入的条件图；高层的条件图为全黑的
+
         # control_processed=[]
         # for i in range(len(t)):
         #   control = batch[self.control_key][i].unsqueeze(0)
@@ -336,7 +336,7 @@ class ControlLDM(LatentDiffusion):
         #     black_image = torch.zeros(control_shape, dtype=torch.float32, device=self.device)
         #     control = black_image
         #     if bs is not None:
-        #         control = control[:bs] # 如果定义了bs，则截取前bs个
+        #         control = control[:bs] 
         #     control = control.to(self.device)
         #     control = einops.rearrange(control, 'b h w c -> b c h w')
         #     control = control.to(memory_format=torch.contiguous_format).float()
@@ -349,10 +349,8 @@ class ControlLDM(LatentDiffusion):
         #     control = control.to(memory_format=torch.contiguous_format).float()
         #     control_processed.append(control)
 
-        # # 将列表中的张量拼接在一起，形成一个新的张量，沿着新的第一个维度
         # control = torch.cat(control_processed, dim=0)
-        # 重写的get_input也需要返回t
-        # 进入ddpm中的shared_setp()
+
         return x, dict(c_crossattn=[c], c_concat=[control]), t
 
     def apply_model(self, x_noisy, t, cond, *args, **kwargs):