[Schedulers] Fix callback steps (#5261)

* fix all

* make fix copies

* make fix copies

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_inpaint_legacy.py

@@ -744,7 +744,8 @@ class StableDiffusionInpaintPipelineLegacy(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         # use original latents corresponding to unmasked portions of the image
         latents = (init_latents_orig * mask) + (latents * (1 - mask))

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_instruct_pix2pix.py

@@ -378,7 +378,8 @@ class StableDiffusionInstructPix2PixPipeline(DiffusionPipeline, TextualInversion
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_latent_upscale.py

@@ -472,7 +472,8 @@ class StableDiffusionLatentUpscalePipeline(DiffusionPipeline):
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_ldm3d.py

@@ -674,7 +674,8 @@ class StableDiffusionLDM3DPipeline(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_model_editing.py

@@ -802,7 +802,8 @@ class StableDiffusionModelEditingPipeline(DiffusionPipeline, TextualInversionLoa
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_panorama.py

@@ -770,7 +770,8 @@ class StableDiffusionPanoramaPipeline(DiffusionPipeline, TextualInversionLoaderM
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             if circular_padding:

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_pix2pix_zero.py

@@ -1006,7 +1006,8 @@ class StableDiffusionPix2PixZeroPipeline(DiffusionPipeline):
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         # 8. Compute the edit directions.
         edit_direction = self.construct_direction(source_embeds, target_embeds).to(prompt_embeds.device)
@@ -1283,7 +1284,8 @@ class StableDiffusionPix2PixZeroPipeline(DiffusionPipeline):
                 ):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         inverted_latents = latents.detach().clone()
 

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_sag.py

@@ -712,7 +712,8 @@ class StableDiffusionSAGPipeline(DiffusionPipeline, TextualInversionLoaderMixin)
                     if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                         progress_bar.update()
                         if callback is not None and i % callback_steps == 0:
-                            callback(i, t, latents)
+                            step_idx = i // getattr(self.scheduler, "order", 1)
+                            callback(step_idx, t, latents)
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_diffusion_upscale.py

@@ -756,7 +756,8 @@ class StableDiffusionUpscalePipeline(DiffusionPipeline, TextualInversionLoaderMi
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16

src/diffusers/pipelines/stable_diffusion/pipeline_stable_unclip.py

@@ -925,7 +925,8 @@ class StableUnCLIPPipeline(DiffusionPipeline, TextualInversionLoaderMixin, LoraL
             latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
 
             if callback is not None and i % callback_steps == 0:
-                callback(i, t, latents)
+                step_idx = i // getattr(self.scheduler, "order", 1)
+                callback(step_idx, t, latents)
 
         if not output_type == "latent":
             image = self.vae.decode(latents / self.vae.config.scaling_factor, return_dict=False)[0]

src/diffusers/pipelines/stable_diffusion/pipeline_stable_unclip_img2img.py

@@ -821,7 +821,8 @@ class StableUnCLIPImg2ImgPipeline(DiffusionPipeline, TextualInversionLoaderMixin
             latents = self.scheduler.step(noise_pred, t, latents, **extra_step_kwargs, return_dict=False)[0]
 
             if callback is not None and i % callback_steps == 0:
-                callback(i, t, latents)
+                step_idx = i // getattr(self.scheduler, "order", 1)
+                callback(step_idx, t, latents)
 
         # 9. Post-processing
         if not output_type == "latent":

src/diffusers/pipelines/stable_diffusion_safe/pipeline_stable_diffusion_safe.py

@@ -674,7 +674,8 @@ class StableDiffusionPipelineSafe(DiffusionPipeline):
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         # 8. Post-processing
         image = self.decode_latents(latents)

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl.py

@@ -877,7 +877,8 @@ class StableDiffusionXLPipeline(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_img2img.py

@@ -1028,7 +1028,8 @@ class StableDiffusionXLImg2ImgPipeline(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_inpaint.py

@@ -1352,7 +1352,8 @@ class StableDiffusionXLInpaintPipeline(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16

src/diffusers/pipelines/stable_diffusion_xl/pipeline_stable_diffusion_xl_instruct_pix2pix.py

@@ -923,7 +923,8 @@ class StableDiffusionXLInstructPix2PixPipeline(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16

src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_adapter.py

@@ -799,7 +799,8 @@ class StableDiffusionAdapterPipeline(DiffusionPipeline):
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if output_type == "latent":
             image = latents

src/diffusers/pipelines/t2i_adapter/pipeline_stable_diffusion_xl_adapter.py

@@ -970,7 +970,8 @@ class StableDiffusionXLAdapterPipeline(
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if not output_type == "latent":
             # make sure the VAE is in float32 mode, as it overflows in float16

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth.py

@@ -664,7 +664,8 @@ class TextToVideoSDPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lora
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if output_type == "latent":
             return TextToVideoSDPipelineOutput(frames=latents)

src/diffusers/pipelines/text_to_video_synthesis/pipeline_text_to_video_synth_img2img.py

@@ -736,7 +736,8 @@ class VideoToVideoSDPipeline(DiffusionPipeline, TextualInversionLoaderMixin, Lor
                 if i == len(timesteps) - 1 or ((i + 1) > num_warmup_steps and (i + 1) % self.scheduler.order == 0):
                     progress_bar.update()
                     if callback is not None and i % callback_steps == 0:
-                        callback(i, t, latents)
+                        step_idx = i // getattr(self.scheduler, "order", 1)
+                        callback(step_idx, t, latents)
 
         if output_type == "latent":
             return TextToVideoSDPipelineOutput(frames=latents)