Fix typos in strings and comments

Signed-off-by: co63oc <[email protected]>

Author: co63oc

examples/cogvideo/train_cogvideox_image_to_video_lora.py

@@ -555,7 +555,7 @@ def _load_dataset_from_local_path(self):
 
         if any(not path.is_file() for path in instance_videos):
             raise ValueError(
-                "Expected '--video_column' to be a path to a file in `--instance_data_root` containing line-separated paths to video data but found atleast one path that is not a valid file."
+                "Expected '--video_column' to be a path to a file in `--instance_data_root` containing line-separated paths to video data but found at least one path that is not a valid file."
             )
 
         return instance_prompts, instance_videos

examples/cogvideo/train_cogvideox_lora.py

@@ -539,7 +539,7 @@ def _load_dataset_from_local_path(self):
 
         if any(not path.is_file() for path in instance_videos):
             raise ValueError(
-                "Expected '--video_column' to be a path to a file in `--instance_data_root` containing line-separated paths to video data but found atleast one path that is not a valid file."
+                "Expected '--video_column' to be a path to a file in `--instance_data_root` containing line-separated paths to video data but found at least one path that is not a valid file."
             )
 
         return instance_prompts, instance_videos

examples/research_projects/multi_subject_dreambooth_inpainting/README.md

@@ -2,7 +2,7 @@
 
 Please note that this project is not actively maintained. However, you can open an issue and tag @gzguevara.
 
-[DreamBooth](https://arxiv.org/abs/2208.12242) is a method to personalize text2image models like stable diffusion given just a few(3~5) images of a subject. This project consists of **two parts**. Training Stable Diffusion for inpainting requieres prompt-image-mask pairs. The Unet of inpainiting models have 5 additional input channels (4 for the encoded masked-image and 1 for the mask itself).
+[DreamBooth](https://arxiv.org/abs/2208.12242) is a method to personalize text2image models like stable diffusion given just a few(3~5) images of a subject. This project consists of **two parts**. Training Stable Diffusion for inpainting requires prompt-image-mask pairs. The Unet of inpainting models have 5 additional input channels (4 for the encoded masked-image and 1 for the mask itself).
 
 **The first part**, the `multi_inpaint_dataset.ipynb` notebook, demonstrates how make a 🤗 dataset of prompt-image-mask pairs. You can, however, skip the first part and move straight to the second part with the example datasets in this project. ([cat toy dataset masked](https://huggingface.co/datasets/gzguevara/cat_toy_masked), [mr. potato head dataset masked](https://huggingface.co/datasets/gzguevara/mr_potato_head_masked))
 
@@ -73,7 +73,7 @@ accelerate launch train_multi_subject_dreambooth_inpaint.py \
 
 ## 3. Results
 
-A [![Weights & Biases](https://img.shields.io/badge/Weights%20&%20Biases-Report-blue)](https://wandb.ai/gzguevara/uncategorized/reports/Multi-Subject-Dreambooth-for-Inpainting--Vmlldzo2MzY5NDQ4?accessToken=y0nya2d7baguhbryxaikbfr1203amvn1jsmyl07vk122mrs7tnph037u1nqgse8t) is provided showing the training progress by every 50 steps. Note, the reported  weights & baises run was performed on a A100 GPU with the following stetting:
+A [![Weights & Biases](https://img.shields.io/badge/Weights%20&%20Biases-Report-blue)](https://wandb.ai/gzguevara/uncategorized/reports/Multi-Subject-Dreambooth-for-Inpainting--Vmlldzo2MzY5NDQ4?accessToken=y0nya2d7baguhbryxaikbfr1203amvn1jsmyl07vk122mrs7tnph037u1nqgse8t) is provided showing the training progress by every 50 steps. Note, the reported weights & biases run was performed on a A100 GPU with the following stetting:
 
 ```bash
 accelerate launch train_multi_subject_dreambooth_inpaint.py \

src/diffusers/hooks/faster_cache.py

@@ -146,7 +146,7 @@ class FasterCacheConfig:
     alpha_low_frequency: float = 1.1
     alpha_high_frequency: float = 1.1
 
-    # n as described in CFG-Cache explanation in the paper - dependant on the model
+    # n as described in CFG-Cache explanation in the paper - dependent on the model
     unconditional_batch_skip_range: int = 5
     unconditional_batch_timestep_skip_range: Tuple[int, int] = (-1, 641)
 

src/diffusers/hooks/hooks.py

@@ -43,9 +43,9 @@ def initialize_hook(self, module: torch.nn.Module) -> torch.nn.Module:
         """
         return module
 
-    def deinitalize_hook(self, module: torch.nn.Module) -> torch.nn.Module:
+    def deinitialize_hook(self, module: torch.nn.Module) -> torch.nn.Module:
         r"""
-        Hook that is executed when a model is deinitalized.
+        Hook that is executed when a model is deinitialized.
 
         Args:
             module (`torch.nn.Module`):
@@ -192,7 +192,7 @@ def remove_hook(self, name: str, recurse: bool = True) -> None:
             else:
                 self._fn_refs[index + 1].forward = old_forward
 
-            self._module_ref = hook.deinitalize_hook(self._module_ref)
+            self._module_ref = hook.deinitialize_hook(self._module_ref)
             del self.hooks[name]
             self._hook_order.pop(index)
             self._fn_refs.pop(index)

src/diffusers/hooks/layerwise_casting.py

@@ -60,9 +60,9 @@ def initialize_hook(self, module: torch.nn.Module):
         module.to(dtype=self.storage_dtype, non_blocking=self.non_blocking)
         return module
 
-    def deinitalize_hook(self, module: torch.nn.Module):
+    def deinitialize_hook(self, module: torch.nn.Module):
         raise NotImplementedError(
-            "LayerwiseCastingHook does not support deinitalization. A model once enabled with layerwise casting will "
+            "LayerwiseCastingHook does not support deinitialization. A model once enabled with layerwise casting will "
             "have casted its weights to a lower precision dtype for storage. Casting this back to the original dtype "
             "will lead to precision loss, which might have an impact on the model's generation quality. The model should "
             "be re-initialized and loaded in the original dtype."

src/diffusers/loaders/peft.py

@@ -250,7 +250,7 @@ def load_lora_adapter(
 
             rank = {}
             for key, val in state_dict.items():
-                # Cannot figure out rank from lora layers that don't have atleast 2 dimensions.
+                # Cannot figure out rank from lora layers that don't have at least 2 dimensions.
                 # Bias layers in LoRA only have a single dimension
                 if "lora_B" in key and val.ndim > 1:
                     # Check out https://github.com/huggingface/peft/pull/2419 for the `^` symbol.

src/diffusers/models/autoencoders/autoencoder_kl.py

@@ -63,7 +63,7 @@ class AutoencoderKL(ModelMixin, ConfigMixin, FromOriginalModelMixin, PeftAdapter
             Synthesis with Latent Diffusion Models](https://arxiv.org/abs/2112.10752) paper.
         force_upcast (`bool`, *optional*, default to `True`):
             If enabled it will force the VAE to run in float32 for high image resolution pipelines, such as SD-XL. VAE
-            can be fine-tuned / trained to a lower range without loosing too much precision in which case
+            can be fine-tuned / trained to a lower range without losing too much precision in which case
             `force_upcast` can be set to `False` - see: https://huggingface.co/madebyollin/sdxl-vae-fp16-fix
         mid_block_add_attention (`bool`, *optional*, default to `True`):
             If enabled, the mid_block of the Encoder and Decoder will have attention blocks. If set to false, the

src/diffusers/models/autoencoders/autoencoder_kl_allegro.py

@@ -715,7 +715,7 @@ class AutoencoderKLAllegro(ModelMixin, ConfigMixin):
             Synthesis with Latent Diffusion Models](https://arxiv.org/abs/2112.10752) paper.
         force_upcast (`bool`, default to `True`):
             If enabled it will force the VAE to run in float32 for high image resolution pipelines, such as SD-XL. VAE
-            can be fine-tuned / trained to a lower range without loosing too much precision in which case
+            can be fine-tuned / trained to a lower range without losing too much precision in which case
             `force_upcast` can be set to `False` - see: https://huggingface.co/madebyollin/sdxl-vae-fp16-fix
     """
 

src/diffusers/models/autoencoders/autoencoder_kl_cogvideox.py

@@ -983,7 +983,7 @@ class AutoencoderKLCogVideoX(ModelMixin, ConfigMixin, FromOriginalModelMixin):
             Synthesis with Latent Diffusion Models](https://arxiv.org/abs/2112.10752) paper.
         force_upcast (`bool`, *optional*, default to `True`):
             If enabled it will force the VAE to run in float32 for high image resolution pipelines, such as SD-XL. VAE
-            can be fine-tuned / trained to a lower range without loosing too much precision in which case
+            can be fine-tuned / trained to a lower range without losing too much precision in which case
             `force_upcast` can be set to `False` - see: https://huggingface.co/madebyollin/sdxl-vae-fp16-fix
     """