Fix use_flow_sigmas (#10242)

use_flow_sigmas copy

Author: hlky

src/diffusers/schedulers/scheduling_deis_multistep.py

@@ -287,7 +287,7 @@ def set_timesteps(self, num_inference_steps: int, device: Union[str, torch.devic
         elif self.config.use_flow_sigmas:
             alphas = np.linspace(1, 1 / self.config.num_train_timesteps, num_inference_steps + 1)
             sigmas = 1.0 - alphas
-            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1]
+            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1].copy()
             timesteps = (sigmas * self.config.num_train_timesteps).copy()
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)

src/diffusers/schedulers/scheduling_dpmsolver_multistep.py

@@ -412,7 +412,7 @@ def set_timesteps(
         elif self.config.use_flow_sigmas:
             alphas = np.linspace(1, 1 / self.config.num_train_timesteps, num_inference_steps + 1)
             sigmas = 1.0 - alphas
-            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1]
+            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1].copy()
             timesteps = (sigmas * self.config.num_train_timesteps).copy()
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)

src/diffusers/schedulers/scheduling_dpmsolver_multistep_inverse.py

@@ -297,7 +297,7 @@ def set_timesteps(self, num_inference_steps: int = None, device: Union[str, torc
         elif self.config.use_flow_sigmas:
             alphas = np.linspace(1, 1 / self.config.num_train_timesteps, num_inference_steps + 1)
             sigmas = 1.0 - alphas
-            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1]
+            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1].copy()
             timesteps = (sigmas * self.config.num_train_timesteps).copy()
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)

src/diffusers/schedulers/scheduling_dpmsolver_singlestep.py

@@ -361,7 +361,7 @@ def set_timesteps(
         elif self.config.use_flow_sigmas:
             alphas = np.linspace(1, 1 / self.config.num_train_timesteps, num_inference_steps + 1)
             sigmas = 1.0 - alphas
-            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1]
+            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1].copy()
             timesteps = (sigmas * self.config.num_train_timesteps).copy()
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)

src/diffusers/schedulers/scheduling_sasolver.py

@@ -316,7 +316,7 @@ def set_timesteps(self, num_inference_steps: int = None, device: Union[str, torc
         elif self.config.use_flow_sigmas:
             alphas = np.linspace(1, 1 / self.config.num_train_timesteps, num_inference_steps + 1)
             sigmas = 1.0 - alphas
-            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1]
+            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1].copy()
             timesteps = (sigmas * self.config.num_train_timesteps).copy()
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)

src/diffusers/schedulers/scheduling_unipc_multistep.py

@@ -379,7 +379,7 @@ def set_timesteps(self, num_inference_steps: int, device: Union[str, torch.devic
         elif self.config.use_flow_sigmas:
             alphas = np.linspace(1, 1 / self.config.num_train_timesteps, num_inference_steps + 1)
             sigmas = 1.0 - alphas
-            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1]
+            sigmas = np.flip(self.config.flow_shift * sigmas / (1 + (self.config.flow_shift - 1) * sigmas))[:-1].copy()
             timesteps = (sigmas * self.config.num_train_timesteps).copy()
         else:
             sigmas = np.interp(timesteps, np.arange(0, len(sigmas)), sigmas)