Merge branch 'main' into ruff_ci

Author: cyyever

src/transformers/generation/flax_utils.py

@@ -221,7 +221,7 @@ def _expand_to_num_beams(tensor, num_beams):
     def _adapt_logits_for_beam_search(self, logits):
         """
         This function can be overwritten in the specific modeling_flax_<model-name>.py classes to allow for custom beam
-        search behavior. Note that the only model that overwrites this method is [`~transformes.FlaxMarianMTModel`].
+        search behavior. Note that the only model that overwrites this method is [`~transformers.FlaxMarianMTModel`].
         """
         return logits
 

src/transformers/kernels/yoso/fast_lsh_cumulation_cuda.cu

@@ -779,12 +779,12 @@ __global__ void lsh_weighted_cumulation_ver4_step2_cuda_kernel(
 
     __syncthreads();
 
-    int num_distint_query = query_counter[0];
+    int num_distinct_query = query_counter[0];
 
-    if (num_distint_query > 0) {
-      for (int idx_base = 0; idx_base < num_distint_query; idx_base = idx_base + num_warps) {
+    if (num_distinct_query > 0) {
+      for (int idx_base = 0; idx_base < num_distinct_query; idx_base = idx_base + num_warps) {
         int idx = idx_base + warp_idx;
-        if (idx < num_distint_query) {
+        if (idx < num_distinct_query) {
           int query_idx = inserted_query[idx];
           int batch_idx__query_idx = batch_idx * num_query + query_idx;
 
@@ -813,7 +813,7 @@ __global__ void lsh_weighted_cumulation_ver4_step2_cuda_kernel(
       }
     } else {
 
-      // all computation is completed if num_distint_query == 0
+      // all computation is completed if num_distinct_query == 0
       break;
 
     }

src/transformers/models/clap/modeling_clap.py

@@ -1717,7 +1717,7 @@ def forward(
         >>> model = ClapModel.from_pretrained("laion/clap-htsat-unfused")
         >>> processor = AutoProcessor.from_pretrained("laion/clap-htsat-unfused")
 
-        >>> input_text = ["Sound of a dog", "Sound of vaccum cleaner"]
+        >>> input_text = ["Sound of a dog", "Sound of vacuum cleaner"]
 
         >>> inputs = processor(text=input_text, audios=audio_sample, return_tensors="pt", padding=True)
 

src/transformers/models/clipseg/modeling_clipseg.py

@@ -625,7 +625,7 @@ def forward(
                 input_ids.to(dtype=torch.int, device=last_hidden_state.device).argmax(dim=-1),
             ]
         else:
-            # The config gets updated `eos_token_id` from PR #24773 (so the use of exta new tokens is possible)
+            # The config gets updated `eos_token_id` from PR #24773 (so the use of extra new tokens is possible)
             pooled_output = last_hidden_state[
                 torch.arange(last_hidden_state.shape[0], device=last_hidden_state.device),
                 # We need to get the first position of `eos_token_id` value (`pad_token_ids` might equal to `eos_token_id`)

src/transformers/models/deepseek_vl/image_processing_deepseek_vl.py

@@ -355,7 +355,7 @@ def pad_to_square(
             background_color (`int` or `tuple[int, int, int]`, *optional*, defaults to 0):
                 The color to use for the padding. Can be an integer for single channel or a
                 tuple of integers representing for multi-channel images. If passed as integer
-                in mutli-channel mode, it will default to `0` in subsequent channels.
+                in multi-channel mode, it will default to `0` in subsequent channels.
             data_format (`str` or `ChannelDimension`, *optional*):
                 The channel dimension format for the output image. Can be one of:
                     - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.

src/transformers/models/deepseek_vl/image_processing_deepseek_vl_fast.py

@@ -113,7 +113,7 @@ def pad_to_square(
             background_color (`int` or `tuple[int, int, int]`, *optional*, defaults to 0):
                 The color to use for the padding. Can be an integer for single channel or a
                 tuple of integers representing for multi-channel images. If passed as integer
-                in mutli-channel mode, it will default to `0` in subsequent channels.
+                in multi-channel mode, it will default to `0` in subsequent channels.
 
         Returns:
             `torch.Tensor`: The padded images.

src/transformers/models/deepseek_vl_hybrid/image_processing_deepseek_vl_hybrid.py

@@ -428,7 +428,7 @@ def pad_to_square(
             background_color (`int` or `tuple[int, int, int]`, *optional*, defaults to 0):
                 The color to use for the padding. Can be an integer for single channel or a
                 tuple of integers representing for multi-channel images. If passed as integer
-                in mutli-channel mode, it will default to `0` in subsequent channels.
+                in multi-channel mode, it will default to `0` in subsequent channels.
             data_format (`str` or `ChannelDimension`, *optional*):
                 The channel dimension format for the output image. Can be one of:
                     - `"channels_first"` or `ChannelDimension.FIRST`: image in (num_channels, height, width) format.

src/transformers/models/deepseek_vl_hybrid/image_processing_deepseek_vl_hybrid_fast.py

@@ -147,7 +147,7 @@ def pad_to_square(
             background_color (`int` or `tuple[int, int, int]`, *optional*, defaults to 0):
                 The color to use for the padding. Can be an integer for single channel or a
                 tuple of integers representing for multi-channel images. If passed as integer
-                in mutli-channel mode, it will default to `0` in subsequent channels.
+                in multi-channel mode, it will default to `0` in subsequent channels.
 
         Returns:
             `torch.Tensor`: The padded images.

src/transformers/models/deprecated/gptsan_japanese/modeling_gptsan_japanese.py

@@ -75,9 +75,9 @@ def load_balancing_loss_func(router_probs: torch.Tensor, expert_indices: torch.T
 
     Args:
         router_probs (`torch.Tensor`):
-            Probability assigned to each expert per token. Shape: [batch_size, seqeunce_length, num_experts].
+            Probability assigned to each expert per token. Shape: [batch_size, sequence_length, num_experts].
         expert_indices (`torch.Tensor`):
-            Indices tensor of shape [batch_size, seqeunce_length] identifying the selected expert for a given token.
+            Indices tensor of shape [batch_size, sequence_length] identifying the selected expert for a given token.
 
     Returns:
         The auxiliary loss.

src/transformers/models/dpt/image_processing_dpt_fast.py

@@ -62,7 +62,7 @@
 class DPTFastImageProcessorKwargs(DefaultFastImageProcessorKwargs):
     """
     ensure_multiple_of (`int`, *optional*, defaults to 1):
-        If `do_resize` is `True`, the image is resized to a size that is a multiple of this value. Can be overidden
+        If `do_resize` is `True`, the image is resized to a size that is a multiple of this value. Can be overridden
         by `ensure_multiple_of` in `preprocess`.
     do_pad (`bool`, *optional*, defaults to `False`):
         Whether to apply center padding. This was introduced in the DINOv2 paper, which uses the model in
@@ -72,7 +72,7 @@ class DPTFastImageProcessorKwargs(DefaultFastImageProcessorKwargs):
         DINOv2 paper, which uses the model in combination with DPT.
     keep_aspect_ratio (`bool`, *optional*, defaults to `False`):
         If `True`, the image is resized to the largest possible size such that the aspect ratio is preserved. Can
-        be overidden by `keep_aspect_ratio` in `preprocess`.
+        be overridden by `keep_aspect_ratio` in `preprocess`.
     do_reduce_labels (`bool`, *optional*, defaults to `self.do_reduce_labels`):
         Whether or not to reduce all label values of segmentation maps by 1. Usually used for datasets where 0
         is used for background, and background itself is not included in all classes of a dataset (e.g.