up

Author: sayakpaul

src/diffusers/hooks/context_parallel.py

@@ -283,8 +283,8 @@ def _find_submodule_by_name(model: torch.nn.Module, name: str) -> torch.nn.Modul
         return model
     first_atom, remaining_name = name.split(".", 1) if "." in name else (name, "")
     if first_atom == "*":
-        if not isinstance(model, torch.nn.ModuleList):
-            raise ValueError("Wildcard '*' can only be used with ModuleList")
+        if not isinstance(model, torch.nn.Modulelist):
+            raise ValueError("Wildcard '*' can only be used with Modulelist")
         submodules = []
         for submodule in model:
             subsubmodules = _find_submodule_by_name(submodule, remaining_name)

src/diffusers/hooks/first_block_cache.py

@@ -227,7 +227,7 @@ def apply_first_block_cache(module: torch.nn.Module, config: FirstBlockCacheConf
     remaining_blocks = []
 
     for name, submodule in module.named_children():
-        if name not in _ALL_TRANSFORMER_BLOCK_IDENTIFIERS or not isinstance(submodule, torch.nn.ModuleList):
+        if name not in _ALL_TRANSFORMER_BLOCK_IDENTIFIERS or not isinstance(submodule, torch.nn.Modulelist):
             continue
         for index, block in enumerate(submodule):
             remaining_blocks.append((f"{name}.{index}", block))

src/diffusers/hooks/group_offloading.py

@@ -471,7 +471,7 @@ def apply_group_offloading(
       memory, but can be slower due to the excessive number of device synchronizations.
 
     Group offloading is a middle ground between the two methods. It works by offloading groups of internal layers,
-    (either `torch.nn.ModuleList` or `torch.nn.Sequential`). This method uses lower memory than module-level
+    (either `torch.nn.Modulelist` or `torch.nn.Sequential`). This method uses lower memory than module-level
     offloading. It is also faster than leaf-level/sequential offloading, as the number of device synchronizations is
     reduced.
 
@@ -576,7 +576,7 @@ def _apply_group_offloading(module: torch.nn.Module, config: GroupOffloadingConf
 
 def _apply_group_offloading_block_level(module: torch.nn.Module, config: GroupOffloadingConfig) -> None:
     r"""
-    This function applies offloading to groups of torch.nn.ModuleList or torch.nn.Sequential blocks. In comparison to
+    This function applies offloading to groups of torch.nn.Modulelist or torch.nn.Sequential blocks. In comparison to
     the "leaf_level" offloading, which is more fine-grained, this offloading is done at the top-level blocks.
     """
 
@@ -586,12 +586,12 @@ def _apply_group_offloading_block_level(module: torch.nn.Module, config: GroupOf
         )
         config.num_blocks_per_group = 1
 
-    # Create module groups for ModuleList and Sequential blocks
+    # Create module groups for Modulelist and Sequential blocks
     modules_with_group_offloading = set()
     unmatched_modules = []
     matched_module_groups = []
     for name, submodule in module.named_children():
-        if not isinstance(submodule, (torch.nn.ModuleList, torch.nn.Sequential)):
+        if not isinstance(submodule, (torch.nn.Modulelist, torch.nn.Sequential)):
             unmatched_modules.append((name, submodule))
             modules_with_group_offloading.add(name)
             continue

src/diffusers/hooks/layer_skip.py

@@ -218,10 +218,10 @@ def _apply_layer_skip_hook(module: torch.nn.Module, config: LayerSkipConfig, nam
             )
 
     transformer_blocks = _get_submodule_from_fqn(module, config.fqn)
-    if transformer_blocks is None or not isinstance(transformer_blocks, torch.nn.ModuleList):
+    if transformer_blocks is None or not isinstance(transformer_blocks, torch.nn.Modulelist):
         raise ValueError(
             f"Could not find {config.fqn} in the provided module, or configured `fqn` (fully qualified name) does not identify "
-            f"a `torch.nn.ModuleList`. Please provide a valid `fqn` that identifies a stack of transformer blocks."
+            f"a `torch.nn.Modulelist`. Please provide a valid `fqn` that identifies a stack of transformer blocks."
         )
     if len(config.indices) == 0:
         raise ValueError("Layer index list is empty. Please provide a non-empty list of layer indices to skip.")

src/diffusers/hooks/utils.py

@@ -21,7 +21,7 @@ def _get_identifiable_transformer_blocks_in_module(module: torch.nn.Module):
     module_list_with_transformer_blocks = []
     for name, submodule in module.named_modules():
         name_endswith_identifier = any(name.endswith(identifier) for identifier in _ALL_TRANSFORMER_BLOCK_IDENTIFIERS)
-        is_modulelist = isinstance(submodule, torch.nn.ModuleList)
+        is_modulelist = isinstance(submodule, torch.nn.Modulelist)
         if name_endswith_identifier and is_modulelist:
             module_list_with_transformer_blocks.append((name, submodule))
     return module_list_with_transformer_blocks

src/diffusers/image_processor.py

@@ -63,7 +63,7 @@ def is_valid_image_imagelist(images):
     - A list of valid images.
 
     Args:
-        images (`Union[np.ndarray, torch.Tensor, PIL.Image.Image, List]`):
+        images (`Union[np.ndarray, torch.Tensor, PIL.Image.Image, list]`):
             The image(s) to check. Can be a batch of images (4D tensor/array), a single image, or a list of valid
             images.
 

src/diffusers/loaders/ip_adapter.py

@@ -830,7 +830,7 @@ def LinearStrengthModel(start, finish, size):
         # Single value for all layers of all IP-Adapters
         if isinstance(scale, scale_type):
             scale = [scale for _ in range(num_ip_adapters)]
-        # List of per-layer scales for a single IP-Adapter
+        # list of per-layer scales for a single IP-Adapter
         elif _is_valid_type(scale, list[scale_type]) and num_ip_adapters == 1:
             scale = [scale]
         # Invalid scale type

src/diffusers/loaders/lora_base.py

@@ -547,7 +547,7 @@ def fuse_lora(
         > [!WARNING] > This is an experimental API.
 
         Args:
-            components: (`list[str]`): List of LoRA-injectable components to fuse the LoRAs into.
+            components: (`list[str]`): list of LoRA-injectable components to fuse the LoRAs into.
             lora_scale (`float`, defaults to 1.0):
                 Controls how much to influence the outputs with the LoRA parameters.
             safe_fusing (`bool`, defaults to `False`):
@@ -627,7 +627,7 @@ def unfuse_lora(self, components: list[str] = [], **kwargs):
         > [!WARNING] > This is an experimental API.
 
         Args:
-            components (`list[str]`): List of LoRA-injectable components to unfuse LoRA from.
+            components (`list[str]`): list of LoRA-injectable components to unfuse LoRA from.
             unfuse_unet (`bool`, defaults to `True`): Whether to unfuse the UNet LoRA parameters.
             unfuse_text_encoder (`bool`, defaults to `True`):
                 Whether to unfuse the text encoder LoRA parameters. If the text encoder wasn't monkey-patched with the
@@ -956,7 +956,7 @@ def set_lora_device(self, adapter_names: list[str], device: torch.device | str |
 
         Args:
             adapter_names (`list[str]`):
-                List of adapters to send device to.
+                list of adapters to send device to.
             device (`Union[torch.device, str, int]`):
                 Device to send the adapters to. Can be either a torch device, a str or an integer.
         """

src/diffusers/loaders/lora_pipeline.py

@@ -543,7 +543,7 @@ def fuse_lora(
         > [!WARNING] > This is an experimental API.
 
         Args:
-            components: (`list[str]`): List of LoRA-injectable components to fuse the LoRAs into.
+            components: (`list[str]`): list of LoRA-injectable components to fuse the LoRAs into.
             lora_scale (`float`, defaults to 1.0):
                 Controls how much to influence the outputs with the LoRA parameters.
             safe_fusing (`bool`, defaults to `False`):
@@ -580,7 +580,7 @@ def unfuse_lora(self, components: list[str] = ["unet", "text_encoder"], **kwargs
         > [!WARNING] > This is an experimental API.
 
         Args:
-            components (`list[str]`): List of LoRA-injectable components to unfuse LoRA from.
+            components (`list[str]`): list of LoRA-injectable components to unfuse LoRA from.
             unfuse_unet (`bool`, defaults to `True`): Whether to unfuse the UNet LoRA parameters.
             unfuse_text_encoder (`bool`, defaults to `True`):
                 Whether to unfuse the text encoder LoRA parameters. If the text encoder wasn't monkey-patched with the
@@ -1992,7 +1992,7 @@ def unfuse_lora(self, components: list[str] = ["transformer", "text_encoder"], *
         > [!WARNING] > This is an experimental API.
 
         Args:
-            components (`list[str]`): List of LoRA-injectable components to unfuse LoRA from.
+            components (`list[str]`): list of LoRA-injectable components to unfuse LoRA from.
         """
         transformer = getattr(self, self.transformer_name) if not hasattr(self, "transformer") else self.transformer
         if hasattr(transformer, "_transformer_norm_layers") and transformer._transformer_norm_layers:
@@ -3889,7 +3889,7 @@ def fuse_lora(
         Fuses the LoRA parameters into the original parameters of the corresponding blocks.
 
         Args:
-            components: (`list[str]`): List of LoRA-injectable components to fuse the LoRAs into.
+            components: (`list[str]`): list of LoRA-injectable components to fuse the LoRAs into.
             lora_scale (`float`, defaults to 1.0):
                 Controls how much to influence the outputs with the LoRA parameters.
             safe_fusing (`bool`, defaults to `False`):
@@ -3919,7 +3919,7 @@ def unfuse_lora(self, components: list[str] = ["transformer"], **kwargs):
         Reverses the effect of [`pipe.fuse_lora()`].
 
         Args:
-            components (`list[str]`): List of LoRA-injectable components to unfuse LoRA from.
+            components (`list[str]`): list of LoRA-injectable components to unfuse LoRA from.
         """
         super().unfuse_lora(components=components, **kwargs)
 

src/diffusers/loaders/textual_inversion.py

@@ -11,7 +11,7 @@
 # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 # See the License for the specific language governing permissions and
 # limitations under the License.
-from typing import List, Optional
+from typing import Optional, list
 
 import safetensors
 import torch
@@ -127,14 +127,14 @@ def maybe_convert_prompt(self, prompt: str | list[str], tokenizer: "PreTrainedTo
         Returns:
             `str` or list of `str`: The converted prompt
         """
-        if not isinstance(prompt, List):
+        if not isinstance(prompt, list):
             prompts = [prompt]
         else:
             prompts = prompt
 
         prompts = [self._maybe_convert_prompt(p, tokenizer) for p in prompts]
 
-        if not isinstance(prompt, List):
+        if not isinstance(prompt, list):
             return prompts[0]
 
         return prompts