Merge branch 'main' into requirements-custom-blocks

Author: sayakpaul

src/diffusers/models/autoencoders/vae.py

@@ -286,19 +286,16 @@ def forward(
 
         sample = self.conv_in(sample)
 
-        upscale_dtype = next(iter(self.up_blocks.parameters())).dtype
         if torch.is_grad_enabled() and self.gradient_checkpointing:
             # middle
             sample = self._gradient_checkpointing_func(self.mid_block, sample, latent_embeds)
-            sample = sample.to(upscale_dtype)
 
             # up
             for up_block in self.up_blocks:
                 sample = self._gradient_checkpointing_func(up_block, sample, latent_embeds)
         else:
             # middle
             sample = self.mid_block(sample, latent_embeds)
-            sample = sample.to(upscale_dtype)
 
             # up
             for up_block in self.up_blocks:

src/diffusers/modular_pipelines/modular_pipeline.py

@@ -320,6 +320,7 @@ def from_pretrained(
             "cache_dir",
             "force_download",
             "local_files_only",
+            "local_dir",
             "proxies",
             "resume_download",
             "revision",
@@ -336,11 +337,10 @@ def from_pretrained(
             module_file=module_file,
             class_name=class_name,
             **hub_kwargs,
-            **kwargs,
         )
         expected_kwargs, optional_kwargs = block_cls._get_signature_keys(block_cls)
         block_kwargs = {
-            name: kwargs.pop(name) for name in kwargs if name in expected_kwargs or name in optional_kwargs
+            name: kwargs.get(name) for name in kwargs if name in expected_kwargs or name in optional_kwargs
         }
 
         return block_cls(**block_kwargs)

src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet.py

@@ -355,7 +355,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/controlnet_sd3/pipeline_stable_diffusion_3_controlnet_inpainting.py

@@ -373,7 +373,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/pag/pipeline_pag_sd_3.py

@@ -326,7 +326,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/pag/pipeline_pag_sd_3_img2img.py

@@ -342,7 +342,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3.py

@@ -336,7 +336,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_img2img.py

@@ -361,7 +361,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/pipelines/stable_diffusion_3/pipeline_stable_diffusion_3_inpaint.py

@@ -367,7 +367,7 @@ def _get_clip_prompt_embeds(
         prompt_embeds = prompt_embeds.repeat(1, num_images_per_prompt, 1)
         prompt_embeds = prompt_embeds.view(batch_size * num_images_per_prompt, seq_len, -1)
 
-        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt, 1)
+        pooled_prompt_embeds = pooled_prompt_embeds.repeat(1, num_images_per_prompt)
         pooled_prompt_embeds = pooled_prompt_embeds.view(batch_size * num_images_per_prompt, -1)
 
         return prompt_embeds, pooled_prompt_embeds

src/diffusers/utils/dynamic_modules_utils.py

@@ -254,6 +254,7 @@ def get_cached_module_file(
     token: Optional[Union[bool, str]] = None,
     revision: Optional[str] = None,
     local_files_only: bool = False,
+    local_dir: Optional[str] = None,
 ):
     """
     Prepares Downloads a module from a local folder or a distant repo and returns its path inside the cached
@@ -332,6 +333,7 @@ def get_cached_module_file(
                 force_download=force_download,
                 proxies=proxies,
                 local_files_only=local_files_only,
+                local_dir=local_dir,
             )
             submodule = "git"
             module_file = pretrained_model_name_or_path + ".py"
@@ -355,6 +357,7 @@ def get_cached_module_file(
                 force_download=force_download,
                 proxies=proxies,
                 local_files_only=local_files_only,
+                local_dir=local_dir,
                 token=token,
             )
             submodule = os.path.join("local", "--".join(pretrained_model_name_or_path.split("/")))
@@ -415,6 +418,7 @@ def get_cached_module_file(
                     token=token,
                     revision=revision,
                     local_files_only=local_files_only,
+                    local_dir=local_dir,
                 )
     return os.path.join(full_submodule, module_file)
 
@@ -431,7 +435,7 @@ def get_class_from_dynamic_module(
     token: Optional[Union[bool, str]] = None,
     revision: Optional[str] = None,
     local_files_only: bool = False,
-    **kwargs,
+    local_dir: Optional[str] = None,
 ):
     """
     Extracts a class from a module file, present in the local folder or repository of a model.
@@ -496,5 +500,6 @@ def get_class_from_dynamic_module(
         token=token,
         revision=revision,
         local_files_only=local_files_only,
+        local_dir=local_dir,
     )
     return get_class_in_module(class_name, final_module)