[Docs] Fix spelling and grammar mistakes (#28825)

* Fix typos and grammar mistakes in docs and examples

* Fix typos in docstrings and comments

* Fix spelling of `tokenizer` in model tests

* Remove erroneous spaces in decorators

* Remove extra spaces in Markdown link texts

CONTRIBUTING.md

@@ -295,7 +295,7 @@ repository such as [`hf-internal-testing`](https://huggingface.co/hf-internal-te
 to host these files and reference them by URL. We recommend placing documentation
 related images in the following repository:
 [huggingface/documentation-images](https://huggingface.co/datasets/huggingface/documentation-images).
-You can open a PR on this dataset repostitory and ask a Hugging Face member to merge it.
+You can open a PR on this dataset repository and ask a Hugging Face member to merge it.
 
 For more information about the checks run on a pull request, take a look at our [Checks on a Pull Request](https://huggingface.co/docs/transformers/pr_checks) guide.
 

docs/source/de/add_new_model.md

@@ -531,7 +531,7 @@ aber alle anderen sollten eine Initialisierung wie oben verwenden. Dies ist wie
 ```py
 def _init_weights(self, module):
     """Initialize the weights"""
-    if isinstnace(module, Wav2Vec2ForPreTraining):
+    if isinstance(module, Wav2Vec2ForPreTraining):
         module.project_hid.reset_parameters()
         module.project_q.reset_parameters()
         module.project_hid._is_hf_initialized = True

docs/source/de/testing.md

@@ -955,7 +955,7 @@ Einige Dekoratoren wie `@parameterized` schreiben Testnamen um, daher müssen `@
 `@require_*` müssen als letztes aufgeführt werden, damit sie korrekt funktionieren. Hier ist ein Beispiel für die korrekte Verwendung:
 
 ```python no-style
-@parameteriz ed.expand(...)
+@parameterized.expand(...)
 @slow
 def test_integration_foo():
 ```

docs/source/en/add_new_model.md

@@ -531,7 +531,7 @@ but all the other ones should use an initialization as above. This is coded like
 ```py
 def _init_weights(self, module):
     """Initialize the weights"""
-    if isinstnace(module, Wav2Vec2ForPreTraining):
+    if isinstance(module, Wav2Vec2ForPreTraining):
         module.project_hid.reset_parameters()
         module.project_q.reset_parameters()
         module.project_hid._is_hf_initialized = True

docs/source/en/community.md

@@ -10,14 +10,14 @@ This page regroups resources around 🤗 Transformers developed by the community
 
 | Resource     |      Description      |      Author      |
 |:----------|:-------------|------:|
-| [Hugging Face Transformers Glossary Flashcards](https://www.darigovresearch.com/huggingface-transformers-glossary-flashcards) | A set of flashcards based on the [Transformers Docs Glossary](glossary) that has been put into a form which can be easily learned/revised using [Anki ](https://apps.ankiweb.net/) an open source, cross platform app specifically designed for long term knowledge retention. See this [Introductory video on how to use the flashcards](https://www.youtube.com/watch?v=Dji_h7PILrw). | [Darigov Research](https://www.darigovresearch.com/) |
+| [Hugging Face Transformers Glossary Flashcards](https://www.darigovresearch.com/huggingface-transformers-glossary-flashcards) | A set of flashcards based on the [Transformers Docs Glossary](glossary) that has been put into a form which can be easily learned/revised using [Anki](https://apps.ankiweb.net/) an open source, cross platform app specifically designed for long term knowledge retention. See this [Introductory video on how to use the flashcards](https://www.youtube.com/watch?v=Dji_h7PILrw). | [Darigov Research](https://www.darigovresearch.com/) |
 
 ## Community notebooks:
 
 | Notebook     |      Description      |      Author      |      |
 |:----------|:-------------|:-------------|------:|
 | [Fine-tune a pre-trained Transformer to generate lyrics](https://github.com/AlekseyKorshuk/huggingartists) | How to generate lyrics in the style of your favorite artist by fine-tuning a GPT-2 model |  [Aleksey Korshuk](https://github.com/AlekseyKorshuk) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb) |
-| [Train T5 in Tensorflow 2 ](https://github.com/snapthat/TF-T5-text-to-text) | How to train T5 for any task using Tensorflow 2. This notebook demonstrates a Question & Answer task implemented in Tensorflow 2 using SQUAD | [Muhammad Harris](https://github.com/HarrisDePerceptron) |[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/snapthat/TF-T5-text-to-text/blob/master/snapthatT5/notebooks/TF-T5-Datasets%20Training.ipynb) |
+| [Train T5 in Tensorflow 2](https://github.com/snapthat/TF-T5-text-to-text) | How to train T5 for any task using Tensorflow 2. This notebook demonstrates a Question & Answer task implemented in Tensorflow 2 using SQUAD | [Muhammad Harris](https://github.com/HarrisDePerceptron) |[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/snapthat/TF-T5-text-to-text/blob/master/snapthatT5/notebooks/TF-T5-Datasets%20Training.ipynb) |
 | [Train T5 on TPU](https://github.com/patil-suraj/exploring-T5/blob/master/T5_on_TPU.ipynb)  | How to train T5 on SQUAD with Transformers and Nlp | [Suraj Patil](https://github.com/patil-suraj) |[![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/patil-suraj/exploring-T5/blob/master/T5_on_TPU.ipynb#scrollTo=QLGiFCDqvuil) |
 | [Fine-tune T5 for Classification and Multiple Choice](https://github.com/patil-suraj/exploring-T5/blob/master/t5_fine_tuning.ipynb)  | How to fine-tune T5 for classification and multiple choice tasks using a text-to-text format with PyTorch Lightning |  [Suraj Patil](https://github.com/patil-suraj) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/patil-suraj/exploring-T5/blob/master/t5_fine_tuning.ipynb) |
 | [Fine-tune DialoGPT on New Datasets and Languages](https://github.com/ncoop57/i-am-a-nerd/blob/master/_notebooks/2020-05-12-chatbot-part-1.ipynb)  | How to fine-tune the DialoGPT model on a new dataset for open-dialog conversational chatbots |  [Nathan Cooper](https://github.com/ncoop57) | [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ncoop57/i-am-a-nerd/blob/master/_notebooks/2020-05-12-chatbot-part-1.ipynb) |

docs/source/en/debugging.md

@@ -136,7 +136,7 @@ This means your GPU architecture is `8.6`.
 
 If you get `8, 6`, then you can set `TORCH_CUDA_ARCH_LIST="8.6"`. For multiple GPUs with different architectures, list them like `TORCH_CUDA_ARCH_LIST="6.1;8.6"`.
 
-It is also possible to not specifiy `TORCH_CUDA_ARCH_LIST` and the build program automatically queries the GPU architecture of the build. However, it may or may not match the actual GPU on the target machine which is why it is better to explicitly specfify the correct architecture.
+It is also possible to not specify `TORCH_CUDA_ARCH_LIST` and the build program automatically queries the GPU architecture of the build. However, it may or may not match the actual GPU on the target machine which is why it is better to explicitly specify the correct architecture.
 
 For training on multiple machines with the same setup, you'll need to make a binary wheel:
 

docs/source/en/deepspeed.md

@@ -394,7 +394,7 @@ Activation and gradient checkpointing trades speed for more GPU memory which all
 
 ### Optimizer and scheduler
 
-DeepSpeed and Transformers optimizer and scheduler can be mixed and matched as long as you don't enable `offload_optimizer`. When `offload_optimizer` is enabled, you could use a non-DeepSpeede optimizer (except for LAMB) as long as it has both a CPU and GPU implementation.
+DeepSpeed and Transformers optimizer and scheduler can be mixed and matched as long as you don't enable `offload_optimizer`. When `offload_optimizer` is enabled, you could use a non-DeepSpeed optimizer (except for LAMB) as long as it has both a CPU and GPU implementation.
 
 <Tip warning={true}>
 
@@ -626,7 +626,7 @@ deepspeed --num_gpus=1 examples/pytorch/translation/run_translation.py \
 DeepSpeed is still useful with just 1 GPU because you can:
 
 1. Offload some computations and memory to the CPU to make more GPU resources available to your model to use a larger batch size or fit a very large model that normally won't fit.
-2. Minimze memory fragmentation with it's smart GPU memory management system which also allows you to fit bigger models and data batches.
+2. Minimize memory fragmentation with it's smart GPU memory management system which also allows you to fit bigger models and data batches.
 
 <Tip>
 
@@ -851,7 +851,7 @@ checkpoint_dir = get_last_checkpoint(trainer.args.output_dir)
 fp32_model = load_state_dict_from_zero_checkpoint(trainer.model, checkpoint_dir)
 ```
 
-If you've enabled the `--load_best_model_at_end` parameter to track the best checkpoint in [`TraininArguments`], you can finish training first and save the final model explicitly. Then you can reload it as shown below:
+If you've enabled the `--load_best_model_at_end` parameter to track the best checkpoint in [`TrainingArguments`], you can finish training first and save the final model explicitly. Then you can reload it as shown below:
 
 ```py
 from deepspeed.utils.zero_to_fp32 import load_state_dict_from_zero_checkpoint
@@ -907,7 +907,7 @@ python zero_to_fp32.py . pytorch_model.bin
 
 <Tip>
 
-Run `python zero_to_fp32.py -h` for more usage details. The script requirees 2x the general RAM of the final fp32 weights.
+Run `python zero_to_fp32.py -h` for more usage details. The script requires 2x the general RAM of the final fp32 weights.
 
 </Tip>
 
@@ -1056,7 +1056,7 @@ train_batch_size = 1 * world_size
 # - if using `offload_param` you can manually finetune stage3_param_persistence_threshold to control
 # - which params should remain on gpus - the larger the value the smaller the offload size
 #
-# For indepth info on Deepspeed config see
+# For in-depth info on Deepspeed config see
 # https://huggingface.co/docs/transformers/main/main_classes/deepspeed
 
 # keeping the same format as json for consistency, except it uses lower case for true/false
@@ -1137,7 +1137,7 @@ This is a very basic example and you'll want to adapt it to your use case.
 
 ### Generate
 
-Using multiple GPUs with ZeRO-3 for generation requires sychronizing the GPUs by setting `synced_gpus=True` in the [`~GenerationMixin.generate`] method. Otherwise, if one GPU is finished generating before another one, the whole system hangs because the remaining GPUs haven't received the weight shard from the GPU that finished first.
+Using multiple GPUs with ZeRO-3 for generation requires synchronizing the GPUs by setting `synced_gpus=True` in the [`~GenerationMixin.generate`] method. Otherwise, if one GPU is finished generating before another one, the whole system hangs because the remaining GPUs haven't received the weight shard from the GPU that finished first.
 
 For Transformers>=4.28, if `synced_gpus` is automatically set to `True` if multiple GPUs are detected during generation.
 
@@ -1167,7 +1167,7 @@ The following sections provide a guide for resolving two of the most common issu
 
 ### DeepSpeed process killed at startup
 
-When the DeepSpeeed process is killed during launch without a traceback, that usually means the program tried to allocate more CPU memory than your system has or your process tried to allocate more CPU memory than allowed leading the OS kernel to terminate the process. In this case, check whether your configuration file has either `offload_optimizer`, `offload_param` or both configured to offload to the CPU. 
+When the DeepSpeed process is killed during launch without a traceback, that usually means the program tried to allocate more CPU memory than your system has or your process tried to allocate more CPU memory than allowed leading the OS kernel to terminate the process. In this case, check whether your configuration file has either `offload_optimizer`, `offload_param` or both configured to offload to the CPU. 
 
 If you have NVMe and ZeRO-3 setup, experiment with offloading to the NVMe ([estimate](https://deepspeed.readthedocs.io/en/latest/memory.html) the memory requirements for your model).
 

docs/source/en/model_doc/informer.md

@@ -18,7 +18,7 @@ rendered properly in your Markdown viewer.
 
 ## Overview
 
-The Informer model was proposed in [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting ](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
+The Informer model was proposed in [Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting](https://arxiv.org/abs/2012.07436) by Haoyi Zhou, Shanghang Zhang, Jieqi Peng, Shuai Zhang, Jianxin Li, Hui Xiong, and Wancai Zhang.
 
 This method introduces a Probabilistic Attention mechanism to select the "active" queries rather than the "lazy" queries and provides a sparse Transformer thus mitigating the quadratic compute and memory requirements of vanilla attention.
 

docs/source/en/model_doc/jukebox.md

@@ -27,7 +27,7 @@ The abstract from the paper is the following:
 *We introduce Jukebox, a model that generates music with singing in the raw audio domain. We tackle the long context of raw audio using a multiscale VQ-VAE to compress it to discrete codes, and modeling those using autoregressive Transformers. We show that the combined model at scale can generate high-fidelity and diverse songs with coherence up to multiple minutes. We can condition on artist and genre to steer the musical and vocal style, and on unaligned lyrics to make the singing more controllable. We are releasing thousands of non cherry-picked samples, along with model weights and code.*
 
 As shown on the following figure, Jukebox is made of 3 `priors` which are decoder only models. They follow the architecture described in [Generating Long Sequences with Sparse Transformers](https://arxiv.org/abs/1904.10509), modified to support longer context length.
-First, a autoencoder is used to encode the text lyrics. Next, the first (also called `top_prior`) prior attends to the last hidden states extracted from the lyrics encoder. The priors are linked to the previous priors respectively via an `AudioConditionner` module. The`AudioConditioner` upsamples the outputs of the previous prior to raw tokens at a certain audio frame per second resolution. 
+First, a autoencoder is used to encode the text lyrics. Next, the first (also called `top_prior`) prior attends to the last hidden states extracted from the lyrics encoder. The priors are linked to the previous priors respectively via an `AudioConditioner` module. The`AudioConditioner` upsamples the outputs of the previous prior to raw tokens at a certain audio frame per second resolution. 
 The metadata such as *artist, genre and timing* are passed to each prior, in the form of a start token and positional embedding for the timing data.  The hidden states are mapped to the closest codebook vector from the VQVAE in order to convert them to raw audio.
 
 ![JukeboxModel](https://gist.githubusercontent.com/ArthurZucker/92c1acaae62ebf1b6a951710bdd8b6af/raw/c9c517bf4eff61393f6c7dec9366ef02bdd059a3/jukebox.svg)
@@ -37,7 +37,7 @@ The original code can be found [here](https://github.com/openai/jukebox).
 
 ## Usage tips
 
-- This model only supports inference. This is for a few reasons, mostly because it requires a crazy amount of memory to train. Feel free to open a PR and add what's missing to have a full integration with the hugging face traineer!
+- This model only supports inference. This is for a few reasons, mostly because it requires a crazy amount of memory to train. Feel free to open a PR and add what's missing to have a full integration with the hugging face trainer!
 - This model is very slow, and takes 8h to generate a minute long audio using the 5b top prior on a V100 GPU. In order automaticallay handle the device on which the model should execute, use `accelerate`.
 - Contrary to the paper, the order of the priors goes from `0` to `1` as it felt more intuitive : we sample starting from `0`.
 - Primed sampling (conditioning the sampling on raw audio) requires more memory than ancestral sampling and should be used with `fp16` set to `True`.

docs/source/en/model_doc/markuplm.md

@@ -27,7 +27,7 @@ The model can be used for tasks like question answering on web pages or informat
 state-of-the-art results on 2 important benchmarks:
 - [WebSRC](https://x-lance.github.io/WebSRC/), a dataset for Web-Based Structural Reading Comprehension (a bit like SQuAD but for web pages)
 - [SWDE](https://www.researchgate.net/publication/221299838_From_one_tree_to_a_forest_a_unified_solution_for_structured_web_data_extraction), a dataset
-for information extraction from web pages (basically named-entity recogntion on web pages)
+for information extraction from web pages (basically named-entity recognition on web pages)
 
 The abstract from the paper is the following:
 

docs/source/en/model_doc/maskformer.md

@@ -39,7 +39,7 @@ This model was contributed by [francesco](https://huggingface.co/francesco). The
 
 ## Usage tips
 
--  MaskFormer's Transformer decoder is identical to the decoder of [DETR](detr). During training, the authors of DETR did find it helpful to use auxiliary losses in the decoder, especially to help the model output the correct number of objects of each class. If you set the parameter `use_auxilary_loss` of [`MaskFormerConfig`] to `True`, then prediction feedforward neural networks and Hungarian losses are added after each decoder layer (with the FFNs sharing parameters).
+-  MaskFormer's Transformer decoder is identical to the decoder of [DETR](detr). During training, the authors of DETR did find it helpful to use auxiliary losses in the decoder, especially to help the model output the correct number of objects of each class. If you set the parameter `use_auxiliary_loss` of [`MaskFormerConfig`] to `True`, then prediction feedforward neural networks and Hungarian losses are added after each decoder layer (with the FFNs sharing parameters).
 - If you want to train the model in a distributed environment across multiple nodes, then one should update the
   `get_num_masks` function inside in the `MaskFormerLoss` class of `modeling_maskformer.py`. When training on multiple nodes, this should be
   set to the average number of target masks across all nodes, as can be seen in the original implementation [here](https://github.com/facebookresearch/MaskFormer/blob/da3e60d85fdeedcb31476b5edd7d328826ce56cc/mask_former/modeling/criterion.py#L169).

docs/source/en/model_doc/mixtral.md

@@ -40,7 +40,7 @@ The original code can be found [here](https://github.com/mistralai/mistral-src).
 
 Mixtral-45B is a decoder-based LM with the following architectural choices:
 
-* Mixtral is a Mixture of Expert (MOE) model with 8 experts per MLP, with a total of 45B paramateres but the compute required is the same as a 14B model. This is because even though each experts have to be loaded in RAM (70B like ram requirement) each token from the hidden states are dipatched twice (top 2 routing) and thus the compute (the operation required at each foward computation) is just 2 X sequence_length. 
+* Mixtral is a Mixture of Expert (MOE) model with 8 experts per MLP, with a total of 45B paramateres but the compute required is the same as a 14B model. This is because even though each experts have to be loaded in RAM (70B like ram requirement) each token from the hidden states are dispatched twice (top 2 routing) and thus the compute (the operation required at each forward computation) is just 2 X sequence_length. 
 
 The following implementation details are shared with Mistral AI's first model [mistral](mistral):
 * Sliding Window Attention - Trained with 8k context length and fixed cache size, with a theoretical attention span of 128K tokens

docs/source/en/model_doc/mms.md

@@ -283,7 +283,7 @@ waveform = outputs.waveform[0]
 
 **Tips:**
 
-* The MMS-TTS checkpoints are trained on lower-cased, un-punctuated text. By default, the `VitsTokenizer` *normalizes* the inputs by removing any casing and punctuation, to avoid passing out-of-vocabulary characters to the model. Hence, the model is agnostic to casing and punctuation, so these should be avoided in the text prompt. You can disable normalisation by setting `noramlize=False` in the call to the tokenizer, but this will lead to un-expected behaviour and is discouraged.
+* The MMS-TTS checkpoints are trained on lower-cased, un-punctuated text. By default, the `VitsTokenizer` *normalizes* the inputs by removing any casing and punctuation, to avoid passing out-of-vocabulary characters to the model. Hence, the model is agnostic to casing and punctuation, so these should be avoided in the text prompt. You can disable normalisation by setting `normalize=False` in the call to the tokenizer, but this will lead to un-expected behaviour and is discouraged.
 * The speaking rate can be varied by setting the attribute `model.speaking_rate` to a chosen value. Likewise, the randomness of the noise is controlled by `model.noise_scale`:
 
 ```python

docs/source/en/model_doc/reformer.md

@@ -54,7 +54,7 @@ found [here](https://github.com/google/trax/tree/master/trax/models/reformer).
 
 Axial Positional Encodings were first implemented in Google's [trax library](https://github.com/google/trax/blob/4d99ad4965bab1deba227539758d59f0df0fef48/trax/layers/research/position_encodings.py#L29)
 and developed by the authors of this model's paper. In models that are treating very long input sequences, the
-conventional position id encodings store an embedings vector of size \\(d\\) being the `config.hidden_size` for
+conventional position id encodings store an embeddings vector of size \\(d\\) being the `config.hidden_size` for
 every position \\(i, \ldots, n_s\\), with \\(n_s\\) being `config.max_embedding_size`. This means that having
 a sequence length of \\(n_s = 2^{19} \approx 0.5M\\) and a `config.hidden_size` of \\(d = 2^{10} \approx 1000\\)
 would result in a position encoding matrix:

docs/source/en/model_doc/rwkv.md

@@ -89,7 +89,7 @@ In a traditional auto-regressive Transformer, attention is written as
 
 $$O = \hbox{softmax}(QK^{T} / \sqrt{d}) V$$
 
-with \\(Q\\), \\(K\\) and \\(V\\) are matrices of shape `seq_len x hidden_size` named query, key and value (they are actually bigger matrices with a batch dimension and an attention head dimension but we're only interested in the last two, which is where the matrix product is taken, so for the sake of simplicity we only consider those two). The product \\(QK^{T}\\) then has shape `seq_len x seq_len` and we can take the maxtrix product with \\(V\\) to get the output \\(O\\) of the same shape as the others.  
+with \\(Q\\), \\(K\\) and \\(V\\) are matrices of shape `seq_len x hidden_size` named query, key and value (they are actually bigger matrices with a batch dimension and an attention head dimension but we're only interested in the last two, which is where the matrix product is taken, so for the sake of simplicity we only consider those two). The product \\(QK^{T}\\) then has shape `seq_len x seq_len` and we can take the matrix product with \\(V\\) to get the output \\(O\\) of the same shape as the others.  
 
 Replacing the softmax by its value gives:
 
@@ -109,7 +109,7 @@ with \\(u\\) and \\(w\\) learnable parameters called in the code `time_first` an
 
 $$N_{i} = e^{u + K_{i}} V_{i} + \hat{N}_{i} \hbox{  where  } \hat{N}_{i} = e^{K_{i-1}} V_{i-1} + e^{w + K_{i-2}} V_{i-2} \cdots + e^{(i-2)w + K_{1}} V_{1}$$
 
-so \\(\hat{N}_{i}\\) (called `numerator_state` in the code) satistfies
+so \\(\hat{N}_{i}\\) (called `numerator_state` in the code) satisfies
 
 $$\hat{N}_{0} = 0 \hbox{  and  } \hat{N}_{j+1} = e^{K_{j}} V_{j} + e^{w} \hat{N}_{j}$$
 
@@ -117,7 +117,7 @@ and
 
 $$D_{i} = e^{u + K_{i}} + \hat{D}_{i} \hbox{  where  } \hat{D}_{i} = e^{K_{i-1}} + e^{w + K_{i-2}} \cdots + e^{(i-2)w + K_{1}}$$
 
-so \\(\hat{D}_{i}\\) (called `denominator_state` in the code) satistfies
+so \\(\hat{D}_{i}\\) (called `denominator_state` in the code) satisfies
 
 $$\hat{D}_{0} = 0 \hbox{  and  } \hat{D}_{j+1} = e^{K_{j}} + e^{w} \hat{D}_{j}$$
 

docs/source/en/model_doc/umt5.md

@@ -47,7 +47,7 @@ found [here](https://github.com/google-research/t5x).
 
 - UMT5 was only pre-trained on [mC4](https://huggingface.co/datasets/mc4) excluding any supervised training.
 Therefore, this model has to be fine-tuned before it is usable on a downstream task, unlike the original T5 model.
-- Since umT5 was pre-trained in an unsupervise manner, there's no real advantage to using a task prefix during single-task
+- Since umT5 was pre-trained in an unsupervised manner, there's no real advantage to using a task prefix during single-task
 fine-tuning. If you are doing multi-task fine-tuning, you should use a prefix.
 
 ## Differences with mT5?

docs/source/en/model_doc/unispeech-sat.md

@@ -31,7 +31,7 @@ this paper, we aim to improve the existing SSL framework for speaker representat
 introduced for enhancing the unsupervised speaker information extraction. First, we apply the multi-task learning to
 the current SSL framework, where we integrate the utterance-wise contrastive loss with the SSL objective function.
 Second, for better speaker discrimination, we propose an utterance mixing strategy for data augmentation, where
-additional overlapped utterances are created unsupervisely and incorporate during training. We integrate the proposed
+additional overlapped utterances are created unsupervisedly and incorporate during training. We integrate the proposed
 methods into the HuBERT framework. Experiment results on SUPERB benchmark show that the proposed system achieves
 state-of-the-art performance in universal representation learning, especially for speaker identification oriented
 tasks. An ablation study is performed verifying the efficacy of each proposed method. Finally, we scale up training

docs/source/en/model_doc/van.md

@@ -39,7 +39,7 @@ Tips:
 
 - VAN does not have an embedding layer, thus the `hidden_states` will have a length equal to the number of stages.
 
-The figure below illustrates the architecture of a Visual Aattention Layer. Taken from the [original paper](https://arxiv.org/abs/2202.09741).
+The figure below illustrates the architecture of a Visual Attention Layer. Taken from the [original paper](https://arxiv.org/abs/2202.09741).
 
 <img width="600" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/van_architecture.png"/>
 

docs/source/en/model_doc/wav2vec2.md

@@ -60,7 +60,7 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 
 🚀 Deploy
 
-- A blog post on how to deploy Wav2Vec2 for [Automatic Speech Recogntion with Hugging Face's Transformers & Amazon SageMaker](https://www.philschmid.de/automatic-speech-recognition-sagemaker).
+- A blog post on how to deploy Wav2Vec2 for [Automatic Speech Recognition with Hugging Face's Transformers & Amazon SageMaker](https://www.philschmid.de/automatic-speech-recognition-sagemaker).
 
 ## Wav2Vec2Config
 

docs/source/en/model_doc/wavlm.md

@@ -31,7 +31,7 @@ challenging. In this paper, we propose a new pre-trained model, WavLM, to solve
 WavLM is built based on the HuBERT framework, with an emphasis on both spoken content modeling and speaker identity
 preservation. We first equip the Transformer structure with gated relative position bias to improve its capability on
 recognition tasks. For better speaker discrimination, we propose an utterance mixing training strategy, where
-additional overlapped utterances are created unsupervisely and incorporated during model training. Lastly, we scale up
+additional overlapped utterances are created unsupervisedly and incorporated during model training. Lastly, we scale up
 the training dataset from 60k hours to 94k hours. WavLM Large achieves state-of-the-art performance on the SUPERB
 benchmark, and brings significant improvements for various speech processing tasks on their representative benchmarks.*
 

docs/source/en/perf_infer_cpu.md

@@ -67,7 +67,7 @@ python run_qa.py \
 
 <Tip warning={true}>
 
-For PyTorch >= 1.14.0, JIT-mode could benefit any model for prediction and evaluaion since the dict input is supported in `jit.trace`.
+For PyTorch >= 1.14.0, JIT-mode could benefit any model for prediction and evaluation since the dict input is supported in `jit.trace`.
 
 For PyTorch < 1.14.0, JIT-mode could benefit a model if its forward parameter order matches the tuple input order in `jit.trace`, such as a question-answering model. If the forward parameter order does not match the tuple input order in `jit.trace`, like a text classification model, `jit.trace` will fail and we are capturing this with the exception here to make it fallback. Logging is used to notify users.
 

docs/source/en/pr_checks.md

@@ -166,7 +166,7 @@ Note that instead of applying this to a whole class, you can apply it to the rel
 # Copied from transformers.models.bert.modeling_bert.BertPreTrainedModel._init_weights
 ```
 
-Sometimes the copy is exactly the same except for names: for instance in `RobertaAttention`, we use `RobertaSelfAttention` insted of `BertSelfAttention` but other than that, the code is exactly the same. This is why `# Copied from` supports simple string replacements with the follwoing syntax: `Copied from xxx with foo->bar`. This means the code is copied with all instances of `foo` being replaced by `bar`. You can see how it used [here](https://github.com/huggingface/transformers/blob/2bd7a27a671fd1d98059124024f580f8f5c0f3b5/src/transformers/models/roberta/modeling_roberta.py#L304C1-L304C86) in `RobertaAttention` with the comment:
+Sometimes the copy is exactly the same except for names: for instance in `RobertaAttention`, we use `RobertaSelfAttention` insted of `BertSelfAttention` but other than that, the code is exactly the same. This is why `# Copied from` supports simple string replacements with the following syntax: `Copied from xxx with foo->bar`. This means the code is copied with all instances of `foo` being replaced by `bar`. You can see how it used [here](https://github.com/huggingface/transformers/blob/2bd7a27a671fd1d98059124024f580f8f5c0f3b5/src/transformers/models/roberta/modeling_roberta.py#L304C1-L304C86) in `RobertaAttention` with the comment:
 
 ```py
 # Copied from transformers.models.bert.modeling_bert.BertAttention with Bert->Roberta

docs/source/en/quantization.md

@@ -36,7 +36,7 @@ Try AWQ quantization with this [notebook](https://colab.research.google.com/driv
 
 [Activation-aware Weight Quantization (AWQ)](https://hf.co/papers/2306.00978) doesn't quantize all the weights in a model, and instead, it preserves a small percentage of weights that are important for LLM performance. This significantly reduces quantization loss such that you can run models in 4-bit precision without experiencing any performance degradation.
 
-There are several libraries for quantizing models with the AWQ algorithm, such as [llm-awq](https://github.com/mit-han-lab/llm-awq), [autoawq](https://github.com/casper-hansen/AutoAWQ) or [optimum-intel](https://huggingface.co/docs/optimum/main/en/intel/optimization_inc). Transformers supports loading models quantized with the llm-awq and autoawq libraries. This guide will show you how to load models quantized with autoawq, but the processs is similar for llm-awq quantized models.
+There are several libraries for quantizing models with the AWQ algorithm, such as [llm-awq](https://github.com/mit-han-lab/llm-awq), [autoawq](https://github.com/casper-hansen/AutoAWQ) or [optimum-intel](https://huggingface.co/docs/optimum/main/en/intel/optimization_inc). Transformers supports loading models quantized with the llm-awq and autoawq libraries. This guide will show you how to load models quantized with autoawq, but the process is similar for llm-awq quantized models.
 
 Make sure you have autoawq installed:
 
@@ -214,7 +214,7 @@ quantized_model = AutoModelForCausalLM.from_pretrained(model_id, device_map="aut
 
 <Tip warning={true}>
 
-Depending on your hardware, it can take some time to quantize a model from scratch. It can take ~5 minutes to quantize the [faceboook/opt-350m]() model on a free-tier Google Colab GPU, but it'll take ~4 hours to quantize a 175B parameter model on a NVIDIA A100. Before you quantize a model, it is a good idea to check the Hub if a GPTQ-quantized version of the model already exists.
+Depending on your hardware, it can take some time to quantize a model from scratch. It can take ~5 minutes to quantize the [facebook/opt-350m]() model on a free-tier Google Colab GPU, but it'll take ~4 hours to quantize a 175B parameter model on a NVIDIA A100. Before you quantize a model, it is a good idea to check the Hub if a GPTQ-quantized version of the model already exists.
 
 </Tip>
 
@@ -583,7 +583,7 @@ The speed and throughput of fused and unfused modules were also tested with the
 <div class="flex gap-4">
   <div>
     <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/fused_forward_memory_plot.png" alt="generate throughput per batch size" />
-    <figcaption class="mt-2 text-center text-sm text-gray-500">foward peak memory/batch size</figcaption>
+    <figcaption class="mt-2 text-center text-sm text-gray-500">forward peak memory/batch size</figcaption>
   </div>
   <div>
     <img class="rounded-xl" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/quantization/fused_generate_throughput_plot.png" alt="forward latency per batch size" />

docs/source/en/tasks/idefics.md

@@ -42,7 +42,7 @@ In this guide, you'll learn how to:
   - [Prompted image captioning](#prompted-image-captioning)
   - [Few-shot prompting](#few-shot-prompting)
   - [Visual question answering](#visual-question-answering)
-  - [Image classificaiton](#image-classification)
+  - [Image classification](#image-classification)
   - [Image-guided text generation](#image-guided-text-generation)
 - [Run inference in batch mode](#running-inference-in-batch-mode)
 - [Run IDEFICS instruct for conversational use](#idefics-instruct-for-conversational-use)

docs/source/en/tasks/masked_language_modeling.md

@@ -108,8 +108,7 @@ For masked language modeling, the next step is to load a DistilRoBERTa tokenizer
 >>> tokenizer = AutoTokenizer.from_pretrained("distilroberta-base")
 ```
 
-You'll notice from the example above, the `text` field is actually nested inside `answers`. This means you'll need to e
-xtract the `text` subfield from its nested structure with the [`flatten`](https://huggingface.co/docs/datasets/process#flatten) method:
+You'll notice from the example above, the `text` field is actually nested inside `answers`. This means you'll need to extract the `text` subfield from its nested structure with the [`flatten`](https://huggingface.co/docs/datasets/process#flatten) method:
 
 ```py
 >>> eli5 = eli5.flatten()

docs/source/en/testing.md

@@ -976,7 +976,7 @@ Some decorators like `@parameterized` rewrite test names, therefore `@slow` and
 `@require_*` have to be listed last for them to work correctly. Here is an example of the correct usage:
 
 ```python no-style
-@parameteriz ed.expand(...)
+@parameterized.expand(...)
 @slow
 def test_integration_foo():
 ```

docs/source/en/tokenizer_summary.md

@@ -143,7 +143,7 @@ Byte-Pair Encoding (BPE) was introduced in [Neural Machine Translation of Rare W
 al., 2015)](https://arxiv.org/abs/1508.07909). BPE relies on a pre-tokenizer that splits the training data into
 words. Pretokenization can be as simple as space tokenization, e.g. [GPT-2](model_doc/gpt2), [RoBERTa](model_doc/roberta). More advanced pre-tokenization include rule-based tokenization, e.g. [XLM](model_doc/xlm),
 [FlauBERT](model_doc/flaubert) which uses Moses for most languages, or [GPT](model_doc/gpt) which uses
-Spacy and ftfy, to count the frequency of each word in the training corpus.
+spaCy and ftfy, to count the frequency of each word in the training corpus.
 
 After pre-tokenization, a set of unique words has been created and the frequency with which each word occurred in the
 training data has been determined. Next, BPE creates a base vocabulary consisting of all symbols that occur in the set

docs/source/it/community.md

@@ -17,7 +17,7 @@ Questa pagina raggruppa le risorse sviluppate dalla comunità riguardo 🤗 Tran
 | Notebook     |      Descrizione      |      Autore      |      |
 |:----------|:-------------|:-------------|------:|
 | [Fine-tuning di un Transformer pre-addestrato, al fine di generare testi di canzoni](https://github.com/AlekseyKorshuk/huggingartists) | Come generare testi di canzoni nello stile del vostro artista preferito attraverso il fine-tuning di un modello GPT-2. |  [Aleksey Korshuk](https://github.com/AlekseyKorshuk) | [![Aprilo in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/AlekseyKorshuk/huggingartists/blob/master/huggingartists-demo.ipynb) |
-| [Addestramento di T5 in Tensorflow 2 ](https://github.com/snapthat/TF-T5-text-to-text) | Come addestrare T5 per qualsiasi attività usando Tensorflow 2. Questo notebook mostra come risolvere l'attività di "Question Answering" usando Tensorflow 2 e SQUAD. | [Muhammad Harris](https://github.com/HarrisDePerceptron) |[![Aprilo in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/snapthat/TF-T5-text-to-text/blob/master/snapthatT5/notebooks/TF-T5-Datasets%20Training.ipynb) |
+| [Addestramento di T5 in Tensorflow 2](https://github.com/snapthat/TF-T5-text-to-text) | Come addestrare T5 per qualsiasi attività usando Tensorflow 2. Questo notebook mostra come risolvere l'attività di "Question Answering" usando Tensorflow 2 e SQUAD. | [Muhammad Harris](https://github.com/HarrisDePerceptron) |[![Aprilo in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/snapthat/TF-T5-text-to-text/blob/master/snapthatT5/notebooks/TF-T5-Datasets%20Training.ipynb) |
 | [Addestramento di T5 con TPU](https://github.com/patil-suraj/exploring-T5/blob/master/T5_on_TPU.ipynb)  | Come addestrare T5 su SQUAD con Transformers e NLP. | [Suraj Patil](https://github.com/patil-suraj) |[![Aprilo in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/patil-suraj/exploring-T5/blob/master/T5_on_TPU.ipynb#scrollTo=QLGiFCDqvuil) |
 | [Fine-tuning di T5 per la classificazione e scelta multipla](https://github.com/patil-suraj/exploring-T5/blob/master/t5_fine_tuning.ipynb)  | Come effettuare il fine-tuning di T5 per le attività di classificazione a scelta multipla - usando un formato testo-a-testo - con PyTorch Lightning. |  [Suraj Patil](https://github.com/patil-suraj) | [![Aprilo in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/patil-suraj/exploring-T5/blob/master/t5_fine_tuning.ipynb) |
 | [Fine-tuning di DialoGPT su nuovi dataset e lingue](https://github.com/ncoop57/i-am-a-nerd/blob/master/_notebooks/2020-05-12-chatbot-part-1.ipynb)  | Come effettuare il fine-tuning di un modello DialoGPT su un nuovo dataset per chatbots conversazionali open-dialog. |  [Nathan Cooper](https://github.com/ncoop57) | [![Aprilo in Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/ncoop57/i-am-a-nerd/blob/master/_notebooks/2020-05-12-chatbot-part-1.ipynb) |

docs/source/ja/add_new_model.md

@@ -430,7 +430,7 @@ def _init_weights(self, module):
 ```py
 def _init_weights(self, module):
     """Initialize the weights"""
-    if isinstnace(module, Wav2Vec2ForPreTraining):
+    if isinstance(module, Wav2Vec2ForPreTraining):
         module.project_hid.reset_parameters()
         module.project_q.reset_parameters()
         module.project_hid._is_hf_initialized = True

docs/source/ja/main_classes/deepspeed.md

@@ -2135,7 +2135,7 @@ train_batch_size = 1 * world_size
 # - if using `offload_param` you can manually finetune stage3_param_persistence_threshold to control
 # - which params should remain on gpus - the larger the value the smaller the offload size
 #
-# For indepth info on Deepspeed config see
+# For in-depth info on Deepspeed config see
 # https://huggingface.co/docs/transformers/main/main_classes/deepspeed
 
 # keeping the same format as json for consistency, except it uses lower case for true/false

docs/source/ja/testing.md

@@ -904,7 +904,7 @@ RUN_SLOW=1 pytest tests
 
 
 ```python no-style
-@parameteriz ed.expand(...)
+@parameterized.expand(...)
 @slow
 def test_integration_foo():
 ```

docs/source/ko/add_new_model.md

@@ -369,7 +369,7 @@ def _init_weights(self, module):
 ```py
 def _init_weights(self, module):
     """Initialize the weights"""
-    if isinstnace(module, Wav2Vec2ForPreTraining):
+    if isinstance(module, Wav2Vec2ForPreTraining):
         module.project_hid.reset_parameters()
         module.project_q.reset_parameters()
         module.project_hid._is_hf_initialized = True

docs/source/zh/main_classes/deepspeed.md

@@ -1982,7 +1982,7 @@ train_batch_size = 1 * world_size
 # - if using `offload_param` you can manually finetune stage3_param_persistence_threshold to control
 # - which params should remain on gpus - the larger the value the smaller the offload size
 #
-# For indepth info on Deepspeed config see
+# For in-depth info on Deepspeed config see
 # https://huggingface.co/docs/transformers/main/main_classes/deepspeed
 
 # keeping the same format as json for consistency, except it uses lower case for true/false

examples/flax/language-modeling/README.md

@@ -449,7 +449,7 @@ are 8 TPU cores on 4 chips (each chips has 2 cores), while "8 GPU" are 8 GPU chi
 
 For comparison one can run the same pre-training with PyTorch/XLA on TPU. To set up PyTorch/XLA on Cloud TPU VMs, please 
 refer to [this](https://cloud.google.com/tpu/docs/pytorch-xla-ug-tpu-vm) guide.
-Having created the tokenzier and configuration in `norwegian-roberta-base`, we create the following symbolic links:
+Having created the tokenizer and configuration in `norwegian-roberta-base`, we create the following symbolic links:
 
 ```bash
 ln -s ~/transformers/examples/pytorch/language-modeling/run_mlm.py ./
@@ -499,7 +499,7 @@ python3 xla_spawn.py --num_cores ${NUM_TPUS} run_mlm.py --output_dir="./runs" \
 
 For comparison you can run the same pre-training with PyTorch on GPU. Note that we have to make use of `gradient_accumulation` 
 because the maximum batch size that fits on a single V100 GPU is 32 instead of 128.
-Having created the tokenzier and configuration in `norwegian-roberta-base`, we create the following symbolic links:
+Having created the tokenizer and configuration in `norwegian-roberta-base`, we create the following symbolic links:
 
 ```bash
 ln -s ~/transformers/examples/pytorch/language-modeling/run_mlm.py ./

examples/flax/question-answering/run_qa.py

@@ -674,7 +674,7 @@ def main():
             raise ValueError("--do_train requires a train dataset")
         train_dataset = raw_datasets["train"]
         if data_args.max_train_samples is not None:
-            # We will select sample from whole data if agument is specified
+            # We will select sample from whole data if argument is specified
             max_train_samples = min(len(train_dataset), data_args.max_train_samples)
             train_dataset = train_dataset.select(range(max_train_samples))
         # Create train feature from dataset

examples/flax/speech-recognition/run_flax_speech_recognition_seq2seq.py

@@ -62,7 +62,7 @@ from transformers.utils.versions import require_version
 # Will error if the minimal version of Transformers is not installed. Remove at your own risk.
 check_min_version("4.38.0.dev0")
 
-require_version("datasets>=2.14.0", "To fix: pip install -r examples/flax/speech-recogintion/requirements.txt")
+require_version("datasets>=2.14.0", "To fix: pip install -r examples/flax/speech-recognition/requirements.txt")
 
 logger = logging.getLogger(__name__)
 

examples/flax/vision/run_image_classification.py

@@ -330,7 +330,7 @@ def main():
 
     # Initialize datasets and pre-processing transforms
     # We use torchvision here for faster pre-processing
-    # Note that here we are using some default pre-processing, for maximum accuray
+    # Note that here we are using some default pre-processing, for maximum accuracy
     # one should tune this part and carefully select what transformations to use.
     normalize = transforms.Normalize(mean=[0.5, 0.5, 0.5], std=[0.5, 0.5, 0.5])
     train_dataset = torchvision.datasets.ImageFolder(

examples/legacy/question-answering/run_squad.py

@@ -148,7 +148,7 @@ def train(args, train_dataset, model, tokenizer):
     # Check if continuing training from a checkpoint
     if os.path.exists(args.model_name_or_path):
         try:
-            # set global_step to gobal_step of last saved checkpoint from model path
+            # set global_step to global_step of last saved checkpoint from model path
             checkpoint_suffix = args.model_name_or_path.split("-")[-1].split("/")[0]
             global_step = int(checkpoint_suffix)
             epochs_trained = global_step // (len(train_dataloader) // args.gradient_accumulation_steps)
@@ -166,7 +166,7 @@ def train(args, train_dataset, model, tokenizer):
     train_iterator = trange(
         epochs_trained, int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0]
     )
-    # Added here for reproductibility
+    # Added here for reproducibility
     set_seed(args)
 
     for _ in train_iterator:
@@ -705,7 +705,7 @@ def main():
     if args.local_rank == -1 or args.no_cuda:
         device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
         args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
-    else:  # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
+    else:  # Initializes the distributed backend which will take care of synchronizing nodes/GPUs
         torch.cuda.set_device(args.local_rank)
         device = torch.device("cuda", args.local_rank)
         torch.distributed.init_process_group(backend="nccl")

examples/legacy/run_swag.py

@@ -338,7 +338,7 @@ def train(args, train_dataset, model, tokenizer):
     tr_loss, logging_loss = 0.0, 0.0
     model.zero_grad()
     train_iterator = trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0])
-    set_seed(args)  # Added here for reproductibility
+    set_seed(args)  # Added here for reproducibility
     for _ in train_iterator:
         epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
         for step, batch in enumerate(epoch_iterator):
@@ -538,7 +538,7 @@ def main():
         default=1,
         help="Number of updates steps to accumulate before performing a backward/update pass.",
     )
-    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight deay if we apply some.")
+    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight decay if we apply some.")
     parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.")
     parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
     parser.add_argument(
@@ -612,7 +612,7 @@ def main():
     if args.local_rank == -1 or args.no_cuda:
         device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
         args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
-    else:  # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
+    else:  # Initializes the distributed backend which will take care of synchronizing nodes/GPUs
         torch.cuda.set_device(args.local_rank)
         device = torch.device("cuda", args.local_rank)
         torch.distributed.init_process_group(backend="nccl")

examples/legacy/seq2seq/README.md

@@ -321,7 +321,7 @@ For example,
 ./save_len_file.py Helsinki-NLP/opus-mt-en-ro  wmt_en_ro
 ./dynamic_bs_example.sh --max_tokens_per_batch=2000 --output_dir benchmark_dynamic_bs
 ```
-splits `wmt_en_ro/train` into 11,197 uneven lengthed batches and can finish 1 epoch in 8 minutes on a v100.
+splits `wmt_en_ro/train` into 11,197 uneven length batches and can finish 1 epoch in 8 minutes on a v100.
 
 For comparison,
 ```bash

examples/legacy/seq2seq/run_distributed_eval.py

@@ -154,7 +154,7 @@ def run_generate():
     parser.add_argument("--src_lang", type=str, default=None, required=False)
     parser.add_argument("--tgt_lang", type=str, default=None, required=False)
     parser.add_argument(
-        "--prefix", type=str, required=False, default=None, help="will be added to the begininng of src examples"
+        "--prefix", type=str, required=False, default=None, help="will be added to the beginning of src examples"
     )
     parser.add_argument("--fp16", action="store_true")
     parser.add_argument("--debug", action="store_true")

examples/legacy/seq2seq/run_eval.py

@@ -107,7 +107,7 @@ def run_generate(verbose=True):
     parser.add_argument("--score_path", type=str, required=False, default="metrics.json", help="where to save metrics")
     parser.add_argument("--device", type=str, required=False, default=DEFAULT_DEVICE, help="cuda, cuda:1, cpu etc.")
     parser.add_argument(
-        "--prefix", type=str, required=False, default=None, help="will be added to the begininng of src examples"
+        "--prefix", type=str, required=False, default=None, help="will be added to the beginning of src examples"
     )
     parser.add_argument("--task", type=str, default="summarization", help="used for task_specific_params + metrics")
     parser.add_argument("--bs", type=int, default=8, required=False, help="batch size")

examples/legacy/seq2seq/seq2seq_trainer.py

@@ -65,7 +65,7 @@ class Seq2SeqTrainer(Trainer):
 
         if self.args.label_smoothing != 0 or (self.data_args is not None and self.data_args.ignore_pad_token_for_loss):
             assert self.config.pad_token_id is not None, (
-                "Make sure that `config.pad_token_id` is correcly defined when ignoring `pad_token` for loss"
+                "Make sure that `config.pad_token_id` is correctly defined when ignoring `pad_token` for loss"
                 " calculation or doing label smoothing."
             )
 

examples/legacy/seq2seq/seq2seq_training_args.py

@@ -31,7 +31,7 @@ class Seq2SeqTrainingArguments(TrainingArguments):
         label_smoothing (:obj:`float`, `optional`, defaults to 0):
             The label smoothing epsilon to apply (if not zero).
         sortish_sampler (:obj:`bool`, `optional`, defaults to :obj:`False`):
-            Whether to SortishSamler or not. It sorts the inputs according to lengths in-order to minimizing the padding size.
+            Whether to SortishSampler or not. It sorts the inputs according to lengths in-order to minimizing the padding size.
         predict_with_generate (:obj:`bool`, `optional`, defaults to :obj:`False`):
             Whether to use generate to calculate generative metrics (ROUGE, BLEU).
     """
@@ -39,7 +39,7 @@ class Seq2SeqTrainingArguments(TrainingArguments):
     label_smoothing: Optional[float] = field(
         default=0.0, metadata={"help": "The label smoothing epsilon to apply (if not zero)."}
     )
-    sortish_sampler: bool = field(default=False, metadata={"help": "Whether to SortishSamler or not."})
+    sortish_sampler: bool = field(default=False, metadata={"help": "Whether to SortishSampler or not."})
     predict_with_generate: bool = field(
         default=False, metadata={"help": "Whether to use generate to calculate generative metrics (ROUGE, BLEU)."}
     )

examples/pytorch/contrastive-image-text/run_clip.py

@@ -289,7 +289,7 @@ def main():
     )
     logger.info(f"Training/evaluation parameters {training_args}")
 
-    # 3. Detecting last checkpoint and eventualy continue from last checkpoint
+    # 3. Detecting last checkpoint and eventually continue from last checkpoint
     last_checkpoint = None
     if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
         last_checkpoint = get_last_checkpoint(training_args.output_dir)
@@ -528,7 +528,7 @@ def main():
         # Transform images on the fly as doing it on the whole dataset takes too much time.
         test_dataset.set_transform(transform_images)
 
-    # 8. Initalize our trainer
+    # 8. Initialize our trainer
     trainer = Trainer(
         model=model,
         args=training_args,

examples/pytorch/image-classification/README.md

@@ -114,10 +114,10 @@ from datasets import load_dataset
 # example 1: local folder
 dataset = load_dataset("imagefolder", data_dir="path_to_your_folder")
 
-# example 2: local files (suppoted formats are tar, gzip, zip, xz, rar, zstd)
+# example 2: local files (supported formats are tar, gzip, zip, xz, rar, zstd)
 dataset = load_dataset("imagefolder", data_files="path_to_zip_file")
 
-# example 3: remote files (suppoted formats are tar, gzip, zip, xz, rar, zstd)
+# example 3: remote files (supported formats are tar, gzip, zip, xz, rar, zstd)
 dataset = load_dataset("imagefolder", data_files="https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip")
 
 # example 4: providing several splits

examples/pytorch/image-classification/run_image_classification.py

@@ -404,7 +404,7 @@ def main():
         # Set the validation transforms
         dataset["validation"].set_transform(val_transforms)
 
-    # Initalize our trainer
+    # Initialize our trainer
     trainer = Trainer(
         model=model,
         args=training_args,

examples/pytorch/image-pretraining/README.md

@@ -25,7 +25,7 @@ NOTE: If you encounter problems/have suggestions for improvement, open an issue
 
 ## SimMIM
 
-The `run_mim.py` script can be used to pre-train any Transformer-based vision model in the library (concretly, any model supported by the `AutoModelForMaskedImageModeling` API) for masked image modeling as proposed in [SimMIM: A Simple Framework for Masked Image Modeling](https://arxiv.org/abs/2111.09886) using PyTorch.
+The `run_mim.py` script can be used to pre-train any Transformer-based vision model in the library (concretely, any model supported by the `AutoModelForMaskedImageModeling` API) for masked image modeling as proposed in [SimMIM: A Simple Framework for Masked Image Modeling](https://arxiv.org/abs/2111.09886) using PyTorch.
 
 <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/simmim_architecture.jpg"
 alt="drawing" width="300"/> 

examples/pytorch/multiple-choice/run_swag_no_trainer.py

@@ -90,7 +90,7 @@ def parse_args():
         default=128,
         help=(
             "The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
-            " sequences shorter will be padded if `--pad_to_max_lengh` is passed."
+            " sequences shorter will be padded if `--pad_to_max_length` is passed."
         ),
     )
     parser.add_argument(

examples/pytorch/question-answering/run_qa.py

@@ -378,7 +378,7 @@ def main():
         )
 
     # Preprocessing the datasets.
-    # Preprocessing is slighlty different for training and evaluation.
+    # Preprocessing is slightly different for training and evaluation.
     if training_args.do_train:
         column_names = raw_datasets["train"].column_names
     elif training_args.do_eval:

examples/pytorch/question-answering/run_qa_beam_search.py

@@ -354,7 +354,7 @@ def main():
     )
 
     # Preprocessing the datasets.
-    # Preprocessing is slighlty different for training and evaluation.
+    # Preprocessing is slightly different for training and evaluation.
     if training_args.do_train:
         column_names = raw_datasets["train"].column_names
     elif training_args.do_eval:

examples/pytorch/question-answering/run_qa_beam_search_no_trainer.py

@@ -119,7 +119,7 @@ def parse_args():
         default=384,
         help=(
             "The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
-            " sequences shorter will be padded if `--pad_to_max_lengh` is passed."
+            " sequences shorter will be padded if `--pad_to_max_length` is passed."
         ),
     )
     parser.add_argument(
@@ -385,7 +385,7 @@ def main():
     )
 
     # Preprocessing the datasets.
-    # Preprocessing is slighlty different for training and evaluation.
+    # Preprocessing is slightly different for training and evaluation.
     column_names = raw_datasets["train"].column_names
 
     question_column_name = "question" if "question" in column_names else column_names[0]
@@ -508,7 +508,7 @@ def main():
         raise ValueError("--do_train requires a train dataset")
     train_dataset = raw_datasets["train"]
     if args.max_train_samples is not None:
-        # We will select sample from whole data if agument is specified
+        # We will select sample from whole data if argument is specified
         train_dataset = train_dataset.select(range(args.max_train_samples))
     # Create train feature from dataset
     with accelerator.main_process_first():
@@ -877,7 +877,7 @@ def main():
                     commit_message=f"Training in progress epoch {epoch}", blocking=False, auto_lfs_prune=True
                 )
 
-    # intialize all lists to collect the batches
+    # initialize all lists to collect the batches
     all_start_top_log_probs = []
     all_start_top_index = []
     all_end_top_log_probs = []
@@ -936,7 +936,7 @@ def main():
     logger.info(f"Evaluation metrics: {eval_metric}")
 
     if args.do_predict:
-        # intialize all lists to collect the batches
+        # initialize all lists to collect the batches
 
         all_start_top_log_probs = []
         all_start_top_index = []

examples/pytorch/question-answering/run_qa_no_trainer.py

@@ -123,7 +123,7 @@ def parse_args():
         default=384,
         help=(
             "The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
-            " sequences shorter will be padded if `--pad_to_max_lengh` is passed."
+            " sequences shorter will be padded if `--pad_to_max_length` is passed."
         ),
     )
     parser.add_argument(
@@ -460,7 +460,7 @@ def main():
         model = AutoModelForQuestionAnswering.from_config(config, trust_remote_code=args.trust_remote_code)
 
     # Preprocessing the datasets.
-    # Preprocessing is slighlty different for training and evaluation.
+    # Preprocessing is slightly different for training and evaluation.
 
     column_names = raw_datasets["train"].column_names
 
@@ -561,7 +561,7 @@ def main():
         raise ValueError("--do_train requires a train dataset")
     train_dataset = raw_datasets["train"]
     if args.max_train_samples is not None:
-        # We will select sample from whole data if agument is specified
+        # We will select sample from whole data if argument is specified
         train_dataset = train_dataset.select(range(args.max_train_samples))
 
     # Create train feature from dataset

examples/pytorch/question-answering/run_seq2seq_qa.py

@@ -559,7 +559,7 @@ def main():
             raise ValueError("--do_train requires a train dataset")
         train_dataset = raw_datasets["train"]
         if data_args.max_train_samples is not None:
-            # We will select sample from whole data if agument is specified
+            # We will select sample from whole data if argument is specified
             max_train_samples = min(len(train_dataset), data_args.max_train_samples)
             train_dataset = train_dataset.select(range(max_train_samples))
         # Create train feature from dataset

examples/pytorch/semantic-segmentation/run_semantic_segmentation.py

@@ -503,7 +503,7 @@ def main():
         # Set the validation transforms
         dataset["validation"].set_transform(preprocess_val)
 
-    # Initalize our trainer
+    # Initialize our trainer
     trainer = Trainer(
         model=model,
         args=training_args,

examples/pytorch/speech-recognition/README.md

@@ -446,7 +446,7 @@ A very common use case is to leverage a pretrained speech encoder model,
 
 By pairing a pretrained speech model with a pretrained text model, the warm-started model has prior knowledge of both the source audio and target text domains. However, the cross-attention weights between the encoder and decoder are randomly initialised. Thus, the model requires fine-tuning to learn the cross-attention weights and align the encoder mapping with that of the decoder. We can perform this very fine-tuning procedure using the example script.
 
-As an example, let's instantiate a *Wav2Vec2-2-Bart* model with the `SpeechEnocderDecoderModel` framework. First create an empty repo on `hf.co`:
+As an example, let's instantiate a *Wav2Vec2-2-Bart* model with the `SpeechEncoderDecoderModel` framework. First create an empty repo on `hf.co`:
 
 ```bash
 huggingface-cli repo create wav2vec2-2-bart-base
@@ -506,7 +506,7 @@ Having warm-started the speech-encoder-decoder model under `<your-user-name>/wav
 In the script [`run_speech_recognition_seq2seq`], we load the warm-started model, 
 feature extractor, and tokenizer, process a speech recognition dataset, 
 and subsequently make use of the [`Seq2SeqTrainer`](https://huggingface.co/docs/transformers/main/en/main_classes/trainer#transformers.Seq2SeqTrainer) to train our system.
-Note that it is important to align the target transcriptions with the decoder's vocabulary. For example, the [`Librispeech`](https://huggingface.co/datasets/librispeech_asr) dataset only contains captilized letters in the transcriptions,
+Note that it is important to align the target transcriptions with the decoder's vocabulary. For example, the [`Librispeech`](https://huggingface.co/datasets/librispeech_asr) dataset only contains capitalized letters in the transcriptions,
 whereas BART was pretrained mostly on normalized text. Thus, it is recommended to add the argument 
 `--do_lower_case` to the fine-tuning script when using a warm-started `SpeechEncoderDecoderModel`. 
 The model is fine-tuned on the standard cross-entropy language modeling

examples/pytorch/speech-recognition/run_speech_recognition_ctc_adapter.py

@@ -146,7 +146,7 @@ class DataTrainingArguments:
                 " should be trained on in ISO 693-3 code, e.g. `tur` for Turkish"
                 " Wav2Vec2's MMS ISO codes can be looked up here: https://dl.fbaipublicfiles.com/mms/misc/language_coverage_mms.html"
                 " If you are not training the adapter layers on a language, simply choose"
-                " another accronym that fits your data."
+                " another acronym that fits your data."
             )
         },
     )

examples/pytorch/text-classification/README.md

@@ -129,7 +129,7 @@ python run_classification.py \
     --num_train_epochs 15 \
     --output_dir /tmp/${dataset}_${subset}/ 
 ```
- It results in a Micro F1 score of around 0.82 without any text and label filtering. Note that you have to explictly remove the "unused" split from the dataset, since it is not used for classification.
+ It results in a Micro F1 score of around 0.82 without any text and label filtering. Note that you have to explicitly remove the "unused" split from the dataset, since it is not used for classification.
 
 ### Mixed precision training
 

examples/pytorch/text-classification/run_classification.py

@@ -83,7 +83,7 @@ class DataTrainingArguments:
         metadata={
             "help": (
                 "The name of the text column in the input dataset or a CSV/JSON file. "
-                'If not specified, will use the "sentence" column for single/multi-label classifcation task.'
+                'If not specified, will use the "sentence" column for single/multi-label classification task.'
             )
         },
     )
@@ -121,7 +121,7 @@ class DataTrainingArguments:
         metadata={
             "help": (
                 "The name of the label column in the input dataset or a CSV/JSON file. "
-                'If not specified, will use the "label" column for single/multi-label classifcation task'
+                'If not specified, will use the "label" column for single/multi-label classification task'
             )
         },
     )
@@ -260,7 +260,7 @@ class ModelArguments:
 
 
 def get_label_list(raw_dataset, split="train") -> List[str]:
-    """Get the list of labels from a mutli-label dataset"""
+    """Get the list of labels from a multi-label dataset"""
 
     if isinstance(raw_dataset[split]["label"][0], list):
         label_list = [label for sample in raw_dataset[split]["label"] for label in sample]
@@ -343,7 +343,7 @@ def main():
 
     # Get the datasets: you can either provide your own CSV/JSON training and evaluation files, or specify a dataset name
     # to load from huggingface/datasets. In ether case, you can specify a the key of the column(s) containing the text and
-    # the key of the column containing the label. If multiple columns are specified for the text, they will be joined togather
+    # the key of the column containing the label. If multiple columns are specified for the text, they will be joined together
     # for the actual text value.
     # In distributed training, the load_dataset function guarantee that only one local process can concurrently
     # download the dataset.

examples/pytorch/text-generation/README.md

@@ -18,7 +18,7 @@ limitations under the License.
 
 Based on the script [`run_generation.py`](https://github.com/huggingface/transformers/blob/main/examples/pytorch/text-generation/run_generation.py).
 
-Conditional text generation using the auto-regressive models of the library: GPT, GPT-2, GPTJ, Transformer-XL, XLNet, CTRL, BLOOM, LLAMA, OPT.
+Conditional text generation using the auto-regressive models of the library: GPT, GPT-2, GPT-J, Transformer-XL, XLNet, CTRL, BLOOM, LLAMA, OPT.
 A similar script is used for our official demo [Write With Transfomer](https://transformer.huggingface.co), where you
 can try out the different models available in the library.
 

examples/pytorch/translation/run_translation_no_trainer.py

@@ -175,7 +175,7 @@ def parse_args():
         default=128,
         help=(
             "The maximum total input sequence length after tokenization. Sequences longer than this will be truncated,"
-            " sequences shorter will be padded if `--pad_to_max_lengh` is passed."
+            " sequences shorter will be padded if `--pad_to_max_length` is passed."
         ),
     )
     parser.add_argument(

examples/research_projects/bert-loses-patience/run_glue_with_pabee.py

@@ -148,7 +148,7 @@ def train(args, train_dataset, model, tokenizer):
     steps_trained_in_current_epoch = 0
     # Check if continuing training from a checkpoint
     if os.path.exists(args.model_name_or_path):
-        # set global_step to gobal_step of last saved checkpoint from model path
+        # set global_step to global_step of last saved checkpoint from model path
         global_step = int(args.model_name_or_path.split("-")[-1].split("/")[0])
         epochs_trained = global_step // (len(train_dataloader) // args.gradient_accumulation_steps)
         steps_trained_in_current_epoch = global_step % (len(train_dataloader) // args.gradient_accumulation_steps)
@@ -169,7 +169,7 @@ def train(args, train_dataset, model, tokenizer):
         desc="Epoch",
         disable=args.local_rank not in [-1, 0],
     )
-    set_seed(args)  # Added here for reproductibility
+    set_seed(args)  # Added here for reproducibility
     for _ in train_iterator:
         epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
         for step, batch in enumerate(epoch_iterator):
@@ -614,7 +614,7 @@ def main():
     if args.local_rank == -1 or args.no_cuda:
         device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
         args.n_gpu = torch.cuda.device_count()
-    else:  # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
+    else:  # Initializes the distributed backend which will take care of synchronizing nodes/GPUs
         torch.cuda.set_device(args.local_rank)
         device = torch.device("cuda", args.local_rank)
         torch.distributed.init_process_group(backend="nccl")

examples/research_projects/codeparrot/scripts/preprocessing.py

@@ -60,7 +60,7 @@ def is_autogenerated(example, scan_width=5):
 def is_config_or_test(example, scan_width=5, coeff=0.05):
     """Check if file is a configuration file or a unit test by :
     1- looking for keywords in the first few lines of the file.
-    2- counting number of occurence of the words 'config' and 'test' with respect to number of lines.
+    2- counting number of occurrence of the words 'config' and 'test' with respect to number of lines.
     """
 
     keywords = ["unit tests", "test file", "configuration file"]

examples/research_projects/deebert/run_glue_deebert.py

@@ -162,7 +162,7 @@ def train(args, train_dataset, model, tokenizer, train_highway=False):
     tr_loss, logging_loss = 0.0, 0.0
     model.zero_grad()
     train_iterator = trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0])
-    set_seed(args)  # Added here for reproductibility (even between python 2 and 3)
+    set_seed(args)  # Added here for reproducibility (even between python 2 and 3)
     for _ in train_iterator:
         epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
         for step, batch in enumerate(epoch_iterator):
@@ -491,7 +491,7 @@ def main():
         help="Number of updates steps to accumulate before performing a backward/update pass.",
     )
     parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.")
-    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight deay if we apply some.")
+    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight decay if we apply some.")
     parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.")
     parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
     parser.add_argument(
@@ -566,7 +566,7 @@ def main():
     if args.local_rank == -1 or args.no_cuda:
         device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
         args.n_gpu = torch.cuda.device_count()
-    else:  # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
+    else:  # Initializes the distributed backend which will take care of synchronizing nodes/GPUs
         torch.cuda.set_device(args.local_rank)
         device = torch.device("cuda", args.local_rank)
         torch.distributed.init_process_group(backend="nccl")

examples/research_projects/distillation/run_squad_w_distillation.py

@@ -165,7 +165,7 @@ def train(args, train_dataset, model, tokenizer, teacher=None):
     # Check if continuing training from a checkpoint
     if os.path.exists(args.model_name_or_path):
         try:
-            # set global_step to gobal_step of last saved checkpoint from model path
+            # set global_step to global_step of last saved checkpoint from model path
             checkpoint_suffix = args.model_name_or_path.split("-")[-1].split("/")[0]
             global_step = int(checkpoint_suffix)
             epochs_trained = global_step // (len(train_dataloader) // args.gradient_accumulation_steps)
@@ -183,7 +183,7 @@ def train(args, train_dataset, model, tokenizer, teacher=None):
     train_iterator = trange(
         epochs_trained, int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0]
     )
-    # Added here for reproductibility
+    # Added here for reproducibility
     set_seed(args)
 
     for _ in train_iterator:
@@ -731,7 +731,7 @@ def main():
     if args.local_rank == -1 or args.no_cuda:
         device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
         args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
-    else:  # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
+    else:  # Initializes the distributed backend which will take care of synchronizing nodes/GPUs
         torch.cuda.set_device(args.local_rank)
         device = torch.device("cuda", args.local_rank)
         torch.distributed.init_process_group(backend="nccl")

examples/research_projects/mm-imdb/run_mmimdb.py

@@ -134,7 +134,7 @@ def train(args, train_dataset, model, tokenizer, criterion):
     best_f1, n_no_improve = 0, 0
     model.zero_grad()
     train_iterator = trange(int(args.num_train_epochs), desc="Epoch", disable=args.local_rank not in [-1, 0])
-    set_seed(args)  # Added here for reproductibility
+    set_seed(args)  # Added here for reproducibility
     for _ in train_iterator:
         epoch_iterator = tqdm(train_dataloader, desc="Iteration", disable=args.local_rank not in [-1, 0])
         for step, batch in enumerate(epoch_iterator):
@@ -384,7 +384,7 @@ def main():
         help="Number of updates steps to accumulate before performing a backward/update pass.",
     )
     parser.add_argument("--learning_rate", default=5e-5, type=float, help="The initial learning rate for Adam.")
-    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight deay if we apply some.")
+    parser.add_argument("--weight_decay", default=0.0, type=float, help="Weight decay if we apply some.")
     parser.add_argument("--adam_epsilon", default=1e-8, type=float, help="Epsilon for Adam optimizer.")
     parser.add_argument("--max_grad_norm", default=1.0, type=float, help="Max gradient norm.")
     parser.add_argument(
@@ -460,7 +460,7 @@ def main():
     if args.local_rank == -1 or args.no_cuda:
         device = torch.device("cuda" if torch.cuda.is_available() and not args.no_cuda else "cpu")
         args.n_gpu = 0 if args.no_cuda else torch.cuda.device_count()
-    else:  # Initializes the distributed backend which will take care of sychronizing nodes/GPUs
+    else:  # Initializes the distributed backend which will take care of synchronizing nodes/GPUs
         torch.cuda.set_device(args.local_rank)
         device = torch.device("cuda", args.local_rank)
         torch.distributed.init_process_group(backend="nccl")

examples/research_projects/quantization-qdqbert/evaluate-hf-trt-qa.py

@@ -275,7 +275,7 @@ else:
 # https://huggingface.co/docs/datasets/loading_datasets.
 
 # Preprocessing the datasets.
-# Preprocessing is slighlty different for training and evaluation.
+# Preprocessing is slightly different for training and evaluation.
 
 column_names = raw_datasets["validation"].column_names
 

examples/research_projects/quantization-qdqbert/run_quant_qa.py

@@ -349,7 +349,7 @@ def main():
         )
 
     # Preprocessing the datasets.
-    # Preprocessing is slighlty different for training and evaluation.
+    # Preprocessing is slightly different for training and evaluation.
     if training_args.do_train or model_args.do_calib:
         column_names = raw_datasets["train"].column_names
     elif training_args.do_eval or model_args.save_onnx:
@@ -448,7 +448,7 @@ def main():
             raise ValueError("--do_train requires a train dataset")
         train_dataset = raw_datasets["train"]
         if data_args.max_train_samples is not None:
-            # We will select sample from whole data if agument is specified
+            # We will select sample from whole data if argument is specified
             max_train_samples = min(len(train_dataset), data_args.max_train_samples)
             train_dataset = train_dataset.select(range(max_train_samples))
         # Create train feature from dataset

examples/research_projects/seq2seq-distillation/README.md

@@ -239,7 +239,7 @@ For example,
 ./save_len_file.py Helsinki-NLP/opus-mt-en-ro  wmt_en_ro
 ./dynamic_bs_example.sh --max_tokens_per_batch=2000 --output_dir benchmark_dynamic_bs
 ```
-splits `wmt_en_ro/train` into 11,197 uneven lengthed batches and can finish 1 epoch in 8 minutes on a v100.
+splits `wmt_en_ro/train` into 11,197 uneven length batches and can finish 1 epoch in 8 minutes on a v100.
 
 For comparison,
 ```bash

examples/research_projects/seq2seq-distillation/run_eval.py

@@ -94,7 +94,7 @@ def run_generate(verbose=True):
     parser.add_argument("--score_path", type=str, required=False, default="metrics.json", help="where to save metrics")
     parser.add_argument("--device", type=str, required=False, default=DEFAULT_DEVICE, help="cuda, cuda:1, cpu etc.")
     parser.add_argument(
-        "--prefix", type=str, required=False, default=None, help="will be added to the begininng of src examples"
+        "--prefix", type=str, required=False, default=None, help="will be added to the beginning of src examples"
     )
     parser.add_argument("--task", type=str, default="summarization", help="used for task_specific_params + metrics")
     parser.add_argument("--bs", type=int, default=8, required=False, help="batch size")

examples/research_projects/wav2vec2/run_common_voice.py

@@ -69,12 +69,12 @@ class ModelArguments:
     hidden_dropout: Optional[float] = field(
         default=0.1,
         metadata={
-            "help": "The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler."
+            "help": "The dropout probability for all fully connected layers in the embeddings, encoder, and pooler."
         },
     )
     feat_proj_dropout: Optional[float] = field(
         default=0.1,
-        metadata={"help": "The dropout probabilitiy for all 1D convolutional layers in feature extractor."},
+        metadata={"help": "The dropout probability for all 1D convolutional layers in feature extractor."},
     )
     mask_time_prob: Optional[float] = field(
         default=0.05,

examples/tensorflow/contrastive-image-text/run_clip.py

@@ -311,7 +311,7 @@ def main():
     # Log on each process the small summary:
     logger.info(f"Training/evaluation parameters {training_args}")
 
-    # 3. Detecting last checkpoint and eventualy continue from last checkpoint
+    # 3. Detecting last checkpoint and eventually continue from last checkpoint
     last_checkpoint = None
     if os.path.isdir(training_args.output_dir) and training_args.do_train and not training_args.overwrite_output_dir:
         if last_checkpoint is None and len(os.listdir(training_args.output_dir)) > 0:

examples/tensorflow/image-classification/README.md

@@ -107,10 +107,10 @@ from datasets import load_dataset
 # example 1: local folder
 dataset = load_dataset("imagefolder", data_dir="path_to_your_folder")
 
-# example 2: local files (suppoted formats are tar, gzip, zip, xz, rar, zstd)
+# example 2: local files (supported formats are tar, gzip, zip, xz, rar, zstd)
 dataset = load_dataset("imagefolder", data_files="path_to_zip_file")
 
-# example 3: remote files (suppoted formats are tar, gzip, zip, xz, rar, zstd)
+# example 3: remote files (supported formats are tar, gzip, zip, xz, rar, zstd)
 dataset = load_dataset("imagefolder", data_files="https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip")
 
 # example 4: providing several splits

examples/tensorflow/language-modeling-tpu/train_unigram.py

@@ -109,7 +109,7 @@ def main(args):
     tokenizer.decoder = decoders.Metaspace()
 
     if args.export_to_hub:
-        logger.info("Exporting the trained tokenzier to Hub.")
+        logger.info("Exporting the trained tokenizer to Hub.")
         new_tokenizer = AlbertTokenizerFast(tokenizer_object=tokenizer)
         new_tokenizer.push_to_hub("unigram-tokenizer-dataset")
 

examples/tensorflow/question-answering/run_qa.py

@@ -512,7 +512,7 @@ def main():
             raise ValueError("--do_train requires a train dataset")
         train_dataset = datasets["train"]
         if data_args.max_train_samples is not None:
-            # We will select sample from whole data if agument is specified
+            # We will select sample from whole data if argument is specified
             max_train_samples = min(len(train_dataset), data_args.max_train_samples)
             train_dataset = train_dataset.select(range(max_train_samples))
         # Create train feature from dataset

src/transformers/modeling_flax_utils.py

@@ -211,7 +211,7 @@ class FlaxPreTrainedModel(PushToHubMixin, FlaxGenerationMixin):
         self.input_shape = input_shape
         self.generation_config = GenerationConfig.from_model_config(config) if self.can_generate() else None
 
-        # To check if the model was intialized automatically.
+        # To check if the model was initialized automatically.
         self._is_initialized = _do_init
 
         if _do_init:

src/transformers/modeling_utils.py

@@ -3736,7 +3736,7 @@ class PreTrainedModel(nn.Module, ModuleUtilsMixin, GenerationMixin, PushToHubMix
                     else:
                         hf_quantizer.create_quantized_param(model, value, key, "cpu", state_dict)
 
-        # retrieve unintialized modules and initialize before maybe overriding that with the pretrained weights.
+        # retrieve uninitialized modules and initialize before maybe overriding that with the pretrained weights.
         if _fast_init:
             if not ignore_mismatched_sizes:
                 if remove_prefix_from_model:

src/transformers/models/big_bird/configuration_big_bird.py

@@ -58,7 +58,7 @@ class BigBirdConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 4096):

src/transformers/models/biogpt/configuration_biogpt.py

@@ -53,7 +53,7 @@ class BioGptConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 1024):

src/transformers/models/canine/configuration_canine.py

@@ -50,7 +50,7 @@ class CanineConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoders, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoders, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 16384):

src/transformers/models/clap/configuration_clap.py

@@ -202,7 +202,7 @@ class ClapAudioConfig(PretrainedConfig):
             Whether or not to enable patch fusion. This is the main contribution of the authors, and should give the
             best results.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the encoder.
+            The dropout probability for all fully connected layers in the encoder.
         fusion_type (`[type]`, *optional*):
             Fusion type used for the patch fusion.
         patch_embed_input_channels (`int`, *optional*, defaults to 1):

src/transformers/models/convbert/configuration_convbert.py

@@ -61,7 +61,7 @@ class ConvBertConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 512):

src/transformers/models/cpmant/configuration_cpmant.py

@@ -51,7 +51,7 @@ class CpmAntConfig(PretrainedConfig):
         num_hidden_layers (`int`, *optional*, defaults to 48):
             Number of layers of the Transformer encoder.
         dropout_p (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder.
+            The dropout probability for all fully connected layers in the embeddings, encoder.
         position_bias_num_buckets (`int`, *optional*, defaults to 512):
             The number of position_bias buckets.
         position_bias_max_distance (`int`, *optional*, defaults to 2048):

src/transformers/models/deit/configuration_deit.py

@@ -59,7 +59,7 @@ class DeiTConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         initializer_range (`float`, *optional*, defaults to 0.02):

src/transformers/models/deprecated/mctct/configuration_mctct.py

@@ -64,7 +64,7 @@ class MCTCTConfig(PretrainedConfig):
         initializer_range (`float`, *optional*, defaults to 0.02):
             The standard deviation of the truncated_normal_initializer for initializing all weight matrices.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.3):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.3):
             The dropout ratio for the attention probabilities.
         pad_token_id (`int`, *optional*, defaults to 1):

src/transformers/models/distilbert/modeling_flax_distilbert.py

@@ -146,7 +146,7 @@ class FlaxEmbeddings(nn.Module):
             position_embeds = self.position_embeddings(position_ids.astype("i4"))
         else:
             position_embeds = self.pos_encoding[:, :seq_length, :]
-            # explictly cast the positions here, since self.embed_positions are not registered as parameters
+            # explicitly cast the positions here, since self.embed_positions are not registered as parameters
             position_embeds = position_embeds.astype(inputs_embeds.dtype)
 
         # Sum all embeddings

src/transformers/models/dpt/configuration_dpt.py

@@ -54,7 +54,7 @@ class DPTConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         initializer_range (`float`, *optional*, defaults to 0.02):

src/transformers/models/esm/modeling_esmfold.py

@@ -1915,7 +1915,7 @@ class EsmFoldingTrunk(nn.Module):
         # This parameter means the axial attention will be computed
         # in a chunked manner. This should make the memory used more or less O(L) instead of O(L^2).
         # It's equivalent to running a for loop over chunks of the dimension we're iterative over,
-        # where the chunk_size is the size of the chunks, so 128 would mean to parse 128-lengthed chunks.
+        # where the chunk_size is the size of the chunks, so 128 would mean to parse 128-length chunks.
         self.chunk_size = chunk_size
 
     def forward(self, seq_feats, pair_feats, true_aa, residx, mask, no_recycles):

src/transformers/models/flava/configuration_flava.py

@@ -53,7 +53,7 @@ class FlavaImageConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         initializer_range (`float`, *optional*, defaults to 0.02):
@@ -188,7 +188,7 @@ class FlavaTextConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         initializer_range (`float`, *optional*, defaults to 0.02):
@@ -302,7 +302,7 @@ class FlavaMultimodalConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         initializer_range (`float`, *optional*, defaults to 0.02):

src/transformers/models/fnet/configuration_fnet.py

@@ -52,7 +52,7 @@ class FNetConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         max_position_embeddings (`int`, *optional*, defaults to 512):
             The maximum sequence length that this model might ever be used with. Typically set this to something large
             just in case (e.g., 512 or 1024 or 2048).

src/transformers/models/gpt_neo/configuration_gpt_neo.py

@@ -70,7 +70,7 @@ class GPTNeoConfig(PretrainedConfig):
         resid_dropout (`float`, *optional*, defaults to 0.0):
             Residual dropout used in the attention pattern.
         embed_dropout (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_dropout (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         classifier_dropout (`float`, *optional*, defaults to 0.1):

src/transformers/models/groupvit/configuration_groupvit.py

@@ -177,7 +177,7 @@ class GroupViTVisionConfig(PretrainedConfig):
         layer_norm_eps (`float`, *optional*, defaults to 1e-5):
             The epsilon used by the layer normalization layers.
         dropout (`float`, *optional*, defaults to 0.0):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_dropout (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         initializer_range (`float`, *optional*, defaults to 0.02):

src/transformers/models/hubert/configuration_hubert.py

@@ -63,7 +63,7 @@ class HubertConfig(PretrainedConfig):
         attention_dropout(`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         final_dropout (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for the final projection layer of [`Wav2Vec2ForCTC`].
+            The dropout probability for the final projection layer of [`Wav2Vec2ForCTC`].
         layerdrop (`float`, *optional*, defaults to 0.1):
             The LayerDrop probability. See the [LayerDrop paper](see https://arxiv.org/abs/1909.11556) for more
             details.

src/transformers/models/layoutlmv2/configuration_layoutlmv2.py

@@ -57,7 +57,7 @@ class LayoutLMv2Config(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 512):

src/transformers/models/layoutlmv3/configuration_layoutlmv3.py

@@ -63,7 +63,7 @@ class LayoutLMv3Config(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 512):

src/transformers/models/marian/modeling_flax_marian.py

@@ -711,7 +711,7 @@ class FlaxMarianEncoder(nn.Module):
         inputs_embeds = self.embed_tokens(input_ids) * self.embed_scale
 
         positions = jnp.take(self.embed_positions, position_ids, axis=0)
-        # explictly cast the positions here, since self.embed_positions are not registered as parameters
+        # explicitly cast the positions here, since self.embed_positions are not registered as parameters
         positions = positions.astype(inputs_embeds.dtype)
 
         hidden_states = inputs_embeds + positions
@@ -771,7 +771,7 @@ class FlaxMarianDecoder(nn.Module):
 
         # embed positions
         positions = jnp.take(self.embed_positions, position_ids, axis=0)
-        # explictly cast the positions here, since self.embed_positions are not registered as parameters
+        # explicitly cast the positions here, since self.embed_positions are not registered as parameters
         positions = positions.astype(inputs_embeds.dtype)
 
         hidden_states = inputs_embeds + positions

src/transformers/models/mobilevit/configuration_mobilevit.py

@@ -77,7 +77,7 @@ class MobileViTConfig(PretrainedConfig):
         output_stride (`int`, *optional*, defaults to 32):
             The ratio of the spatial resolution of the output to the resolution of the input image.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the Transformer encoder.
+            The dropout probability for all fully connected layers in the Transformer encoder.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.0):
             The dropout ratio for the attention probabilities.
         classifier_dropout_prob (`float`, *optional*, defaults to 0.1):

src/transformers/models/mra/configuration_mra.py

@@ -52,7 +52,7 @@ class MraConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 512):

src/transformers/models/nystromformer/configuration_nystromformer.py

@@ -52,7 +52,7 @@ class NystromformerConfig(PretrainedConfig):
             The non-linear activation function (function or string) in the encoder and pooler. If string, `"gelu"`,
             `"relu"`, `"selu"` and `"gelu_new"` are supported.
         hidden_dropout_prob (`float`, *optional*, defaults to 0.1):
-            The dropout probabilitiy for all fully connected layers in the embeddings, encoder, and pooler.
+            The dropout probability for all fully connected layers in the embeddings, encoder, and pooler.
         attention_probs_dropout_prob (`float`, *optional*, defaults to 0.1):
             The dropout ratio for the attention probabilities.
         max_position_embeddings (`int`, *optional*, defaults to 512):

src/transformers/models/pegasus/modeling_flax_pegasus.py

@@ -707,7 +707,7 @@ class FlaxPegasusEncoder(nn.Module):
 
         # embed positions
         embed_pos = jnp.take(self.embed_positions, position_ids, axis=0)
-        # explictly cast the positions here, since self.embed_positions are not registered as parameters
+        # explicitly cast the positions here, since self.embed_positions are not registered as parameters
         embed_pos = embed_pos.astype(inputs_embeds.dtype)
 
         hidden_states = inputs_embeds + embed_pos
@@ -778,7 +778,7 @@ class FlaxPegasusDecoder(nn.Module):
 
         # embed positions
         positions = jnp.take(self.embed_positions, position_ids, axis=0)
-        # explictly cast the positions here, since self.embed_positions are not registered as parameters
+        # explicitly cast the positions here, since self.embed_positions are not registered as parameters
         positions = positions.astype(inputs_embeds.dtype)
 
         hidden_states = inputs_embeds + positions