[Docs] Model_doc structure/clarity improvements (#26876)

* first batch of structure improvements for model_docs

* second batch of structure improvements for model_docs

* more structure improvements for model_docs

* more structure improvements for model_docs

* structure improvements for cv model_docs

* more structural refactoring

* addressed feedback about image processors

docs/source/en/model_doc/mluke.md

@@ -37,6 +37,10 @@ representations into the input allows us to extract more language-agnostic featu
 multilingual cloze prompt task with the mLAMA dataset. We show that entity-based prompt elicits correct factual
 knowledge more likely than using only word representations.*
 
+This model was contributed by [ryo0634](https://huggingface.co/ryo0634). The original code can be found [here](https://github.com/studio-ousia/luke).
+
+## Usage tips
+
 One can directly plug in the weights of mLUKE into a LUKE model, like so:
 
 ```python
@@ -53,10 +57,12 @@ from transformers import MLukeTokenizer
 tokenizer = MLukeTokenizer.from_pretrained("studio-ousia/mluke-base")
 ```
 
+<Tip>
+
 As mLUKE's architecture is equivalent to that of LUKE, one can refer to [LUKE's documentation page](luke) for all
 tips, code examples and notebooks.
 
-This model was contributed by [ryo0634](https://huggingface.co/ryo0634). The original code can be found [here](https://github.com/studio-ousia/luke).
+</Tip>
 
 ## MLukeTokenizer
 

docs/source/en/model_doc/mms.md

@@ -306,7 +306,6 @@ with torch.no_grad():
    outputs = model(**inputs)
 ```
 
-
 ### Language Identification (LID)
 
 Different LID models are available based on the number of languages they can recognize - [126](https://huggingface.co/facebook/mms-lid-126), [256](https://huggingface.co/facebook/mms-lid-256), [512](https://huggingface.co/facebook/mms-lid-512), [1024](https://huggingface.co/facebook/mms-lid-1024), [2048](https://huggingface.co/facebook/mms-lid-2048), [4017](https://huggingface.co/facebook/mms-lid-4017). 
@@ -378,4 +377,13 @@ processor.id2label.values()
 
 ### Audio Pretrained Models
 
-Pretrained models are available for two different sizes - [300M](https://huggingface.co/facebook/mms-300m) , [1Bil](https://huggingface.co/facebook/mms-1b). The architecture is based on the Wav2Vec2 model, so one can refer to [Wav2Vec2's documentation page](wav2vec2) for further details on how to finetune with models for various downstream tasks.
+Pretrained models are available for two different sizes - [300M](https://huggingface.co/facebook/mms-300m) , 
+[1Bil](https://huggingface.co/facebook/mms-1b). 
+
+<Tip>
+
+The MMS for ASR architecture is based on the Wav2Vec2 model, refer to [Wav2Vec2's documentation page](wav2vec2) for further 
+details on how to finetune with models for various downstream tasks.
+
+MMS-TTS uses the same model architecture as VITS, refer to [VITS's documentation page](vits) for API reference.
+</Tip>

docs/source/en/model_doc/mobilebert.md

@@ -37,7 +37,9 @@ natural language inference tasks of GLUE, MobileBERT achieves a GLUEscore o 77.7
 latency on a Pixel 4 phone. On the SQuAD v1.1/v2.0 question answering task, MobileBERT achieves a dev F1 score of
 90.0/79.2 (1.5/2.1 higher than BERT_BASE).*
 
-Tips:
+This model was contributed by [vshampor](https://huggingface.co/vshampor). The original code can be found [here](https://github.com/google-research/mobilebert).
+
+## Usage tips
 
 - MobileBERT is a model with absolute position embeddings so it's usually advised to pad the inputs on the right rather
   than the left.
@@ -45,9 +47,8 @@ Tips:
   efficient at predicting masked tokens and at NLU in general, but is not optimal for text generation. Models trained
   with a causal language modeling (CLM) objective are better in that regard.
 
-This model was contributed by [vshampor](https://huggingface.co/vshampor). The original code can be found [here](https://github.com/google-research/mobilebert).
 
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)
@@ -73,6 +74,9 @@ This model was contributed by [vshampor](https://huggingface.co/vshampor). The o
 
 [[autodoc]] models.mobilebert.modeling_tf_mobilebert.TFMobileBertForPreTrainingOutput
 
+<frameworkcontent>
+<pt>
+
 ## MobileBertModel
 
 [[autodoc]] MobileBertModel
@@ -113,6 +117,9 @@ This model was contributed by [vshampor](https://huggingface.co/vshampor). The o
 [[autodoc]] MobileBertForQuestionAnswering
     - forward
 
+</pt>
+<tf>
+
 ## TFMobileBertModel
 
 [[autodoc]] TFMobileBertModel
@@ -152,3 +159,6 @@ This model was contributed by [vshampor](https://huggingface.co/vshampor). The o
 
 [[autodoc]] TFMobileBertForQuestionAnswering
     - call
+
+</tf>
+</frameworkcontent>

docs/source/en/model_doc/mobilenet_v1.md

@@ -24,7 +24,9 @@ The abstract from the paper is the following:
 
 *We present a class of efficient models called MobileNets for mobile and embedded vision applications. MobileNets are based on a streamlined architecture that uses depth-wise separable convolutions to build light weight deep neural networks. We introduce two simple global hyper-parameters that efficiently trade off between latency and accuracy. These hyper-parameters allow the model builder to choose the right sized model for their application based on the constraints of the problem. We present extensive experiments on resource and accuracy tradeoffs and show strong performance compared to other popular models on ImageNet classification. We then demonstrate the effectiveness of MobileNets across a wide range of applications and use cases including object detection, finegrain classification, face attributes and large scale geo-localization.*
 
-Tips:
+This model was contributed by [matthijs](https://huggingface.co/Matthijs). The original code and weights can be found [here](https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet_v1.md).
+
+## Usage tips
 
 - The checkpoints are named **mobilenet\_v1\_*depth*\_*size***, for example **mobilenet\_v1\_1.0\_224**, where **1.0** is the depth multiplier (sometimes also referred to as "alpha" or the width multiplier) and **224** is the resolution of the input images the model was trained on.
 
@@ -46,8 +48,6 @@ Unsupported features:
 
 - It's common to extract the output from the pointwise layers at indices 5, 11, 12, 13 for downstream purposes. Using `output_hidden_states=True` returns the output from all intermediate layers. There is currently no way to limit this to specific layers.
 
-This model was contributed by [matthijs](https://huggingface.co/Matthijs). The original code and weights can be found [here](https://github.com/tensorflow/models/blob/master/research/slim/nets/mobilenet_v1.md).
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with MobileNetV1.

docs/source/en/model_doc/mobilenet_v2.md

@@ -26,7 +26,9 @@ The abstract from the paper is the following:
 
 *The MobileNetV2 architecture is based on an inverted residual structure where the input and output of the residual block are thin bottleneck layers opposite to traditional residual models which use expanded representations in the input an MobileNetV2 uses lightweight depthwise convolutions to filter features in the intermediate expansion layer. Additionally, we find that it is important to remove non-linearities in the narrow layers in order to maintain representational power. We demonstrate that this improves performance and provide an intuition that led to this design. Finally, our approach allows decoupling of the input/output domains from the expressiveness of the transformation, which provides a convenient framework for further analysis. We measure our performance on Imagenet classification, COCO object detection, VOC image segmentation. We evaluate the trade-offs between accuracy, and number of operations measured by multiply-adds (MAdd), as well as the number of parameters.*
 
-Tips:
+This model was contributed by [matthijs](https://huggingface.co/Matthijs). The original code and weights can be found [here for the main model](https://github.com/tensorflow/models/tree/master/research/slim/nets/mobilenet) and [here for DeepLabV3+](https://github.com/tensorflow/models/tree/master/research/deeplab).
+
+## Usage tips
 
 - The checkpoints are named **mobilenet\_v2\_*depth*\_*size***, for example **mobilenet\_v2\_1.0\_224**, where **1.0** is the depth multiplier (sometimes also referred to as "alpha" or the width multiplier) and **224** is the resolution of the input images the model was trained on.
 
@@ -50,8 +52,6 @@ Unsupported features:
 
 - The DeepLabV3+ segmentation head does not use the final convolution layer from the backbone, but this layer gets computed anyway. There is currently no way to tell [`MobileNetV2Model`] up to which layer it should run.
 
-This model was contributed by [matthijs](https://huggingface.co/Matthijs). The original code and weights can be found [here for the main model](https://github.com/tensorflow/models/tree/master/research/slim/nets/mobilenet) and [here for DeepLabV3+](https://github.com/tensorflow/models/tree/master/research/deeplab).
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with MobileNetV2.

docs/source/en/model_doc/mobilevit.md

@@ -24,7 +24,9 @@ The abstract from the paper is the following:
 
 *Light-weight convolutional neural networks (CNNs) are the de-facto for mobile vision tasks. Their spatial inductive biases allow them to learn representations with fewer parameters across different vision tasks. However, these networks are spatially local. To learn global representations, self-attention-based vision trans-formers (ViTs) have been adopted. Unlike CNNs, ViTs are heavy-weight. In this paper, we ask the following question: is it possible to combine the strengths of CNNs and ViTs to build a light-weight and low latency network for mobile vision tasks? Towards this end, we introduce MobileViT, a light-weight and general-purpose vision transformer for mobile devices. MobileViT presents a different perspective for the global processing of information with transformers, i.e., transformers as convolutions. Our results show that MobileViT significantly outperforms CNN- and ViT-based networks across different tasks and datasets. On the ImageNet-1k dataset, MobileViT achieves top-1 accuracy of 78.4% with about 6 million parameters, which is 3.2% and 6.2% more accurate than MobileNetv3 (CNN-based) and DeIT (ViT-based) for a similar number of parameters. On the MS-COCO object detection task, MobileViT is 5.7% more accurate than MobileNetv3 for a similar number of parameters.*
 
-Tips:
+This model was contributed by [matthijs](https://huggingface.co/Matthijs). The TensorFlow version of the model was contributed by [sayakpaul](https://huggingface.co/sayakpaul). The original code and weights can be found [here](https://github.com/apple/ml-cvnets).
+
+## Usage tips
 
 - MobileViT is more like a CNN than a Transformer model. It does not work on sequence data but on batches of images. Unlike ViT, there are no embeddings. The backbone model outputs a feature map. You can follow [this tutorial](https://keras.io/examples/vision/mobilevit) for a lightweight introduction.
 - One can use [`MobileViTImageProcessor`] to prepare images for the model. Note that if you do your own preprocessing, the pretrained checkpoints expect images to be in BGR pixel order (not RGB).
@@ -58,9 +60,6 @@ with open(tflite_filename, "wb") as f:
   The resulting model will be just **about an MB** making it a good fit for mobile applications where resources and network
   bandwidth can be constrained.
 
-
-This model was contributed by [matthijs](https://huggingface.co/Matthijs). The TensorFlow version of the model was contributed by [sayakpaul](https://huggingface.co/sayakpaul). The original code and weights can be found [here](https://github.com/apple/ml-cvnets).
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with MobileViT.
@@ -91,6 +90,9 @@ If you're interested in submitting a resource to be included here, please feel f
     - preprocess
     - post_process_semantic_segmentation
 
+<frameworkcontent>
+<pt>
+
 ## MobileViTModel
 
 [[autodoc]] MobileViTModel
@@ -106,6 +108,9 @@ If you're interested in submitting a resource to be included here, please feel f
 [[autodoc]] MobileViTForSemanticSegmentation
     - forward
 
+</pt>
+<tf>
+
 ## TFMobileViTModel
 
 [[autodoc]] TFMobileViTModel
@@ -120,3 +125,6 @@ If you're interested in submitting a resource to be included here, please feel f
 
 [[autodoc]] TFMobileViTForSemanticSegmentation
     - call
+
+</tf>
+</frameworkcontent>
\ No newline at end of file

docs/source/en/model_doc/mobilevitv2.md

@@ -26,17 +26,16 @@ The abstract from the paper is the following:
 
 *Mobile vision transformers (MobileViT) can achieve state-of-the-art performance across several mobile vision tasks, including classification and detection. Though these models have fewer parameters, they have high latency as compared to convolutional neural network-based models. The main efficiency bottleneck in MobileViT is the multi-headed self-attention (MHA) in transformers, which requires O(k2) time complexity with respect to the number of tokens (or patches) k. Moreover, MHA requires costly operations (e.g., batch-wise matrix multiplication) for computing self-attention, impacting latency on resource-constrained devices. This paper introduces a separable self-attention method with linear complexity, i.e. O(k). A simple yet effective characteristic of the proposed method is that it uses element-wise operations for computing self-attention, making it a good choice for resource-constrained devices. The improved model, MobileViTV2, is state-of-the-art on several mobile vision tasks, including ImageNet object classification and MS-COCO object detection. With about three million parameters, MobileViTV2 achieves a top-1 accuracy of 75.6% on the ImageNet dataset, outperforming MobileViT by about 1% while running 3.2× faster on a mobile device.*
 
-Tips:
+This model was contributed by [shehan97](https://huggingface.co/shehan97).
+The original code can be found [here](https://github.com/apple/ml-cvnets).
+
+## Usage tips
 
 - MobileViTV2 is more like a CNN than a Transformer model. It does not work on sequence data but on batches of images. Unlike ViT, there are no embeddings. The backbone model outputs a feature map.
 - One can use [`MobileViTImageProcessor`] to prepare images for the model. Note that if you do your own preprocessing, the pretrained checkpoints expect images to be in BGR pixel order (not RGB).
 - The available image classification checkpoints are pre-trained on [ImageNet-1k](https://huggingface.co/datasets/imagenet-1k) (also referred to as ILSVRC 2012, a collection of 1.3 million images and 1,000 classes).
 - The segmentation model uses a [DeepLabV3](https://arxiv.org/abs/1706.05587) head. The available semantic segmentation checkpoints are pre-trained on [PASCAL VOC](http://host.robots.ox.ac.uk/pascal/VOC/).
 
-This model was contributed by [shehan97](https://huggingface.co/shehan97).
-The original code can be found [here](https://github.com/apple/ml-cvnets).
-
-
 ## MobileViTV2Config
 
 [[autodoc]] MobileViTV2Config

docs/source/en/model_doc/mpnet.md

@@ -37,14 +37,14 @@ down-streaming tasks (GLUE, SQuAD, etc). Experimental results show that MPNet ou
 margin, and achieves better results on these tasks compared with previous state-of-the-art pre-trained methods (e.g.,
 BERT, XLNet, RoBERTa) under the same model setting.*
 
-Tips:
+The original code can be found [here](https://github.com/microsoft/MPNet).
 
-- MPNet doesn't have `token_type_ids`, you don't need to indicate which token belongs to which segment. just
-  separate your segments with the separation token `tokenizer.sep_token` (or `[sep]`).
+## Usage tips
 
-The original code can be found [here](https://github.com/microsoft/MPNet).
+MPNet doesn't have `token_type_ids`, you don't need to indicate which token belongs to which segment. Just 
+separate your segments with the separation token `tokenizer.sep_token` (or `[sep]`).
 
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)
@@ -68,6 +68,9 @@ The original code can be found [here](https://github.com/microsoft/MPNet).
 
 [[autodoc]] MPNetTokenizerFast
 
+<frameworkcontent>
+<pt>
+
 ## MPNetModel
 
 [[autodoc]] MPNetModel
@@ -98,6 +101,9 @@ The original code can be found [here](https://github.com/microsoft/MPNet).
 [[autodoc]] MPNetForQuestionAnswering
     - forward
 
+</pt>
+<tf>
+
 ## TFMPNetModel
 
 [[autodoc]] TFMPNetModel
@@ -127,3 +133,6 @@ The original code can be found [here](https://github.com/microsoft/MPNet).
 
 [[autodoc]] TFMPNetForQuestionAnswering
     - call
+
+</tf>
+</frameworkcontent>

docs/source/en/model_doc/mpt.md

@@ -30,13 +30,14 @@ The original code is available at the  [`llm-foundry`](https://github.com/mosaic
 
 Read more about it [in the release blogpost](https://www.mosaicml.com/blog/mpt-7b)
 
-Tips:
+## Usage tips
 
 - Learn more about some techniques behind training of the model [in this section of llm-foundry repository](https://github.com/mosaicml/llm-foundry/blob/main/TUTORIAL.md#faqs)
 - If you want to use the advanced version of the model (triton kernels, direct flash attention integration), you can still use the original model implementation by adding `trust_remote_code=True` when calling `from_pretrained`.
 
-- [Fine-tuning Notebook](https://colab.research.google.com/drive/1HCpQkLL7UXW8xJUJJ29X7QAeNJKO0frZ?usp=sharing) on how to fine-tune MPT-7B on a free Google Colab instance to turn the model into a Chatbot.
+## Resources
 
+- [Fine-tuning Notebook](https://colab.research.google.com/drive/1HCpQkLL7UXW8xJUJJ29X7QAeNJKO0frZ?usp=sharing) on how to fine-tune MPT-7B on a free Google Colab instance to turn the model into a Chatbot.
 
 ## MptConfig
 

docs/source/en/model_doc/mra.md

@@ -27,24 +27,20 @@ The abstract from the paper is the following:
 This model was contributed by [novice03](https://huggingface.co/novice03).
 The original code can be found [here](https://github.com/mlpen/mra-attention).
 
-
 ## MraConfig
 
 [[autodoc]] MraConfig
 
-
 ## MraModel
 
 [[autodoc]] MraModel
     - forward
 
-
 ## MraForMaskedLM
 
 [[autodoc]] MraForMaskedLM
     - forward
 
-
 ## MraForSequenceClassification
 
 [[autodoc]] MraForSequenceClassification
@@ -55,13 +51,11 @@ The original code can be found [here](https://github.com/mlpen/mra-attention).
 [[autodoc]] MraForMultipleChoice
     - forward
 
-
 ## MraForTokenClassification
 
 [[autodoc]] MraForTokenClassification
     - forward
 
-
 ## MraForQuestionAnswering
 
 [[autodoc]] MraForQuestionAnswering

docs/source/en/model_doc/mt5.md

@@ -60,7 +60,7 @@ Google has released the following variants:
 This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten). The original code can be
 found [here](https://github.com/google-research/multilingual-t5).
 
-## Documentation resources
+## Resources
 
 - [Translation task guide](../tasks/translation)
 - [Summarization task guide](../tasks/summarization)
@@ -82,6 +82,8 @@ See [`T5Tokenizer`] for all details.
 
 See [`T5TokenizerFast`] for all details.
 
+<frameworkcontent>
+<pt>
 
 ## MT5Model
 
@@ -103,6 +105,9 @@ See [`T5TokenizerFast`] for all details.
 
 [[autodoc]] MT5ForQuestionAnswering
 
+</pt>
+<tf>
+
 ## TFMT5Model
 
 [[autodoc]] TFMT5Model
@@ -115,6 +120,9 @@ See [`T5TokenizerFast`] for all details.
 
 [[autodoc]] TFMT5EncoderModel
 
+</tf>
+<jax>
+
 ## FlaxMT5Model
 
 [[autodoc]] FlaxMT5Model
@@ -126,3 +134,6 @@ See [`T5TokenizerFast`] for all details.
 ## FlaxMT5EncoderModel
 
 [[autodoc]] FlaxMT5EncoderModel
+
+</jax>
+</frameworkcontent>

docs/source/en/model_doc/mvp.md

@@ -28,15 +28,17 @@ According to the abstract,
 - MVP also has task-specific soft prompts to stimulate the model's capacity in performing a certain task.
 - MVP is specially designed for natural language generation and can be adapted to a wide range of generation tasks, including but not limited to summarization, data-to-text generation, open-ended dialogue system, story generation, question answering, question generation, task-oriented dialogue system, commonsense generation, paraphrase generation, text style transfer, and text simplification. Our model can also be adapted to natural language understanding tasks such as sequence classification and (extractive) question answering.
 
-Tips:
+This model was contributed by [Tianyi Tang](https://huggingface.co/StevenTang). The detailed information and instructions can be found [here](https://github.com/RUCAIBox/MVP).
+
+## Usage tips
+
 - We have released a series of models [here](https://huggingface.co/models?filter=mvp), including MVP, MVP with task-specific prompts, and multi-task pre-trained variants.
 - If you want to use a model without prompts (standard Transformer), you can load it through `MvpForConditionalGeneration.from_pretrained('RUCAIBox/mvp')`.
 - If you want to use a model with task-specific prompts, such as summarization, you can load it through `MvpForConditionalGeneration.from_pretrained('RUCAIBox/mvp-summarization')`.
 - Our model supports lightweight prompt tuning following [Prefix-tuning](https://arxiv.org/abs/2101.00190) with method `set_lightweight_tuning()`.
 
-This model was contributed by [Tianyi Tang](https://huggingface.co/StevenTang). The detailed information and instructions can be found [here](https://github.com/RUCAIBox/MVP).
+## Usage examples
 
-## Examples
 For summarization, it is an example to use MVP and MVP with summarization-specific prompts.
 
 ```python
@@ -104,7 +106,7 @@ For lightweight tuning, *i.e.*, fixing the model and only tuning prompts, you ca
 >>> model.set_lightweight_tuning()
 ```
 
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Question answering task guide](../tasks/question_answering)

docs/source/en/model_doc/nat.md

@@ -36,7 +36,18 @@ that boosts image classification and downstream vision performance. Experimental
 NAT-Tiny reaches 83.2% top-1 accuracy on ImageNet, 51.4% mAP on MS-COCO and 48.4% mIoU on ADE20K, which is 1.9%
 ImageNet accuracy, 1.0% COCO mAP, and 2.6% ADE20K mIoU improvement over a Swin model with similar size. *
 
-Tips:
+<img
+src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/neighborhood-attention-pattern.jpg"
+alt="drawing" width="600"/>
+
+<small> Neighborhood Attention compared to other attention patterns.
+Taken from the <a href="https://arxiv.org/abs/2204.07143">original paper</a>.</small>
+
+This model was contributed by [Ali Hassani](https://huggingface.co/alihassanijr).
+The original code can be found [here](https://github.com/SHI-Labs/Neighborhood-Attention-Transformer).
+
+## Usage tips
+
 - One can use the [`AutoImageProcessor`] API to prepare images for the model.
 - NAT can be used as a *backbone*. When `output_hidden_states = True`,
 it will output both `hidden_states` and `reshaped_hidden_states`.
@@ -50,16 +61,6 @@ or build on your system by running `pip install natten`.
 Note that the latter will likely take time to compile. NATTEN does not support Windows devices yet.
 - Patch size of 4 is only supported at the moment.
 
-<img
-src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/neighborhood-attention-pattern.jpg"
-alt="drawing" width="600"/>
-
-<small> Neighborhood Attention compared to other attention patterns.
-Taken from the <a href="https://arxiv.org/abs/2204.07143">original paper</a>.</small>
-
-This model was contributed by [Ali Hassani](https://huggingface.co/alihassanijr).
-The original code can be found [here](https://github.com/SHI-Labs/Neighborhood-Attention-Transformer).
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with NAT.
@@ -75,7 +76,6 @@ If you're interested in submitting a resource to be included here, please feel f
 
 [[autodoc]] NatConfig
 
-
 ## NatModel
 
 [[autodoc]] NatModel

docs/source/en/model_doc/nezha.md

@@ -35,7 +35,7 @@ and natural language inference (XNLI).*
 
 This model was contributed by [sijunhe](https://huggingface.co/sijunhe). The original code can be found [here](https://github.com/huawei-noah/Pretrained-Language-Model/tree/master/NEZHA-PyTorch).
 
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)

docs/source/en/model_doc/nllb-moe.md

@@ -37,22 +37,24 @@ improvements to counteract overfitting while training on thousands of tasks. Cri
 a human-translated benchmark, Flores-200, and combined human evaluation with a novel toxicity benchmark covering all languages in Flores-200 to assess translation safety.
 Our model achieves an improvement of 44% BLEU relative to the previous state-of-the-art, laying important groundwork towards realizing a universal translation system.*
 
-Tips:
+This model was contributed by [Arthur Zucker](https://huggingface.co/ArtZucker).
+The original code can be found [here](https://github.com/facebookresearch/fairseq).
+
+## Usage tips
 
 - M2M100ForConditionalGeneration is the base model for both NLLB and NLLB MoE
 - The NLLB-MoE is very similar to the NLLB model, but it's feed forward layer is based on the implementation of SwitchTransformers.
 - The tokenizer is the same as the NLLB models.
 
-This model was contributed by [Arthur Zucker](https://huggingface.co/ArtZucker).
-The original code can be found [here](https://github.com/facebookresearch/fairseq).
-
 ## Implementation differences with SwitchTransformers
+
 The biggest difference is the way the tokens are routed. NLLB-MoE uses a `top-2-gate` which means that for each input, only the top two experts are selected based on the 
 highest predicted probabilities from the gating network, and the remaining experts are ignored. In `SwitchTransformers`, only the top-1 probabilities are computed, 
 which means that tokens have less probability of being forwarded. Moreover, if a token is not routed to any expert, `SwitchTransformers` still adds its unmodified hidden 
 states (kind of like a residual connection) while they are masked in `NLLB`'s top-2 routing mechanism. 
 
 ## Generating with NLLB-MoE
+
 The available checkpoints require around 350GB of storage. Make sure to use `accelerate` if you do not have enough RAM on your machine.
 
 While generating the target text set the `forced_bos_token_id` to the target language id. The following
@@ -99,7 +101,7 @@ See example below for a translation from romanian to german:
 >>> tokenizer.batch_decode(translated_tokens, skip_special_tokens=True)[0]
 ```
 
-## Documentation resources
+## Resources
 
 - [Translation task guide](../tasks/translation)
 - [Summarization task guide](../tasks/summarization)

docs/source/en/model_doc/nllb.md

@@ -16,8 +16,9 @@ rendered properly in your Markdown viewer.
 
 # NLLB
 
-**DISCLAIMER:** The default behaviour for the tokenizer has recently been fixed (and thus changed)!
+## Updated tokenizer behavior 
 
+**DISCLAIMER:** The default behaviour for the tokenizer was fixed and thus changed in April 2023.
 The previous version adds `[self.eos_token_id, self.cur_lang_code]` at the end of the token sequence for both target and source tokenization. This is wrong as the NLLB paper mentions (page 48, 6.1.1. Model Architecture) :
 
 *Note that we prefix the source sequence with the source language, as opposed to the target
@@ -56,7 +57,7 @@ Enabling the old behaviour can be done as follows:
 
 For more details, feel free to check the linked [PR](https://github.com/huggingface/transformers/pull/22313) and [Issue](https://github.com/huggingface/transformers/issues/19943).
 
-## Overview of NLLB
+## Overview
 
 The NLLB model was presented in [No Language Left Behind: Scaling Human-Centered Machine Translation](https://arxiv.org/abs/2207.04672) by Marta R. Costa-jussà, James Cross, Onur Çelebi,
 Maha Elbayad, Kenneth Heafield, Kevin Heffernan, Elahe Kalbassi, Janice Lam, Daniel Licht, Jean Maillard, Anna Sun, Skyler Wang, Guillaume Wenzek, Al Youngblood, Bapi Akula,
@@ -131,7 +132,7 @@ See example below for a translation from romanian to german:
 UN-Chef sagt, es gibt keine militärische Lösung in Syrien
 ```
 
-## Documentation resources
+## Resources
 
 - [Translation task guide](../tasks/translation)
 - [Summarization task guide](../tasks/summarization)

docs/source/en/model_doc/nougat.md

@@ -33,7 +33,7 @@ alt="drawing" width="600"/>
 This model was contributed by [nielsr](https://huggingface.co/nielsr). The original code can be found
 [here](https://github.com/facebookresearch/nougat).
 
-Tips:
+## Usage tips
 
 - The quickest way to get started with Nougat is by checking the [tutorial
   notebooks](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/Nougat), which show how to use the model
@@ -89,6 +89,12 @@ into a single instance to both extract the input features and decode the predict
 
 See the [model hub](https://huggingface.co/models?filter=nougat) to look for Nougat checkpoints.
 
+<Tip>
+
+The model is identical to [Donut](donut) in terms of architecture.
+
+</Tip>
+
 ## NougatImageProcessor
 
 [[autodoc]] NougatImageProcessor

docs/source/en/model_doc/nystromformer.md

@@ -37,7 +37,7 @@ favorably relative to other efficient self-attention methods. Our code is availa
 
 This model was contributed by [novice03](https://huggingface.co/novice03). The original code can be found [here](https://github.com/mlpen/Nystromformer).
 
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)

docs/source/en/model_doc/oneformer.md

@@ -26,7 +26,14 @@ The abstract from the paper is the following:
 
 *Universal Image Segmentation is not a new concept. Past attempts to unify image segmentation in the last decades include scene parsing, panoptic segmentation, and, more recently, new panoptic architectures. However, such panoptic architectures do not truly unify image segmentation because they need to be trained individually on the semantic, instance, or panoptic segmentation to achieve the best performance. Ideally, a truly universal framework should be trained only once and achieve SOTA performance across all three image segmentation tasks. To that end, we propose OneFormer, a universal image segmentation framework that unifies segmentation with a multi-task train-once design. We first propose a task-conditioned joint training strategy that enables training on ground truths of each domain (semantic, instance, and panoptic segmentation) within a single multi-task training process. Secondly, we introduce a task token to condition our model on the task at hand, making our model task-dynamic to support multi-task training and inference. Thirdly, we propose using a query-text contrastive loss during training to establish better inter-task and inter-class distinctions. Notably, our single OneFormer model outperforms specialized Mask2Former models across all three segmentation tasks on ADE20k, CityScapes, and COCO, despite the latter being trained on each of the three tasks individually with three times the resources. With new ConvNeXt and DiNAT backbones, we observe even more performance improvement. We believe OneFormer is a significant step towards making image segmentation more universal and accessible.*
 
-Tips:
+The figure below illustrates the architecture of OneFormer. Taken from the [original paper](https://arxiv.org/abs/2211.06220).
+
+<img width="600" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/oneformer_architecture.png"/>
+
+This model was contributed by [Jitesh Jain](https://huggingface.co/praeclarumjj3). The original code can be found [here](https://github.com/SHI-Labs/OneFormer).
+
+## Usage tips
+
 -  OneFormer requires two inputs during inference: *image* and *task token*. 
 - During training, OneFormer only uses panoptic annotations.
 - If you want to train the model in a distributed environment across multiple nodes, then one should update the
@@ -35,12 +42,6 @@ Tips:
 - One can use [`OneFormerProcessor`] to prepare input images and task inputs for the model and optional targets for the model. [`OneformerProcessor`] wraps [`OneFormerImageProcessor`] and [`CLIPTokenizer`] into a single instance to both prepare the images and encode the task inputs.
 - To get the final segmentation, depending on the task, you can call [`~OneFormerProcessor.post_process_semantic_segmentation`] or [`~OneFormerImageProcessor.post_process_instance_segmentation`] or [`~OneFormerImageProcessor.post_process_panoptic_segmentation`]. All three tasks can be solved using [`OneFormerForUniversalSegmentation`] output, panoptic segmentation accepts an optional `label_ids_to_fuse` argument to fuse instances of the target object/s (e.g. sky) together.
 
-The figure below illustrates the architecture of OneFormer. Taken from the [original paper](https://arxiv.org/abs/2211.06220).
-
-<img width="600" src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/oneformer_architecture.png"/>
-
-This model was contributed by [Jitesh Jain](https://huggingface.co/praeclarumjj3). The original code can be found [here](https://github.com/SHI-Labs/OneFormer).
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with OneFormer.

docs/source/en/model_doc/open-llama.md

@@ -18,7 +18,7 @@ rendered properly in your Markdown viewer.
 
 <Tip warning={true}>
 
-This model is in maintenance mode only, so we won't accept any new PRs changing its code.
+This model is in maintenance mode only, we don't accept any new PRs changing its code.
 
 If you run into any issues running this model, please reinstall the last version that supported this model: v4.31.0.
 You can do so by running the following command: `pip install -U transformers==4.31.0`.