[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/vit_mae.md

@@ -32,7 +32,15 @@ enables us to train large models efficiently and effectively: we accelerate trai
 models that generalize well: e.g., a vanilla ViT-Huge model achieves the best accuracy (87.8%) among methods that use only ImageNet-1K data. Transfer performance in downstream
 tasks outperforms supervised pre-training and shows promising scaling behavior.*
 
-Tips:
+<img src="https://user-images.githubusercontent.com/11435359/146857310-f258c86c-fde6-48e8-9cee-badd2b21bd2c.png"
+alt="drawing" width="600"/> 
+
+<small> MAE architecture. Taken from the <a href="https://arxiv.org/abs/2111.06377">original paper.</a> </small>
+
+This model was contributed by [nielsr](https://huggingface.co/nielsr). TensorFlow version of the model was contributed by [sayakpaul](https://github.com/sayakpaul) and 
+[ariG23498](https://github.com/ariG23498) (equal contribution). The original code can be found [here](https://github.com/facebookresearch/mae). 
+
+## Usage tips
 
 - MAE (masked auto encoding) is a method for self-supervised pre-training of Vision Transformers (ViTs). The pre-training objective is relatively simple:
 by masking a large portion (75%) of the image patches, the model must reconstruct raw pixel values. One can use [`ViTMAEForPreTraining`] for this purpose.
@@ -44,14 +52,6 @@ consists of Transformer blocks) takes as input. Each mask token is a shared, lea
 sin/cos position embeddings are added both to the input of the encoder and the decoder.
 - For a visual understanding of how MAEs work you can check out this [post](https://keras.io/examples/vision/masked_image_modeling/).
 
-<img src="https://user-images.githubusercontent.com/11435359/146857310-f258c86c-fde6-48e8-9cee-badd2b21bd2c.png"
-alt="drawing" width="600"/> 
-
-<small> MAE architecture. Taken from the <a href="https://arxiv.org/abs/2111.06377">original paper.</a> </small>
-
-This model was contributed by [nielsr](https://huggingface.co/nielsr). TensorFlow version of the model was contributed by [sayakpaul](https://github.com/sayakpaul) and 
-[ariG23498](https://github.com/ariG23498) (equal contribution). The original code can be found [here](https://github.com/facebookresearch/mae). 
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with ViTMAE.
@@ -65,26 +65,31 @@ If you're interested in submitting a resource to be included here, please feel f
 
 [[autodoc]] ViTMAEConfig
 
+<frameworkcontent>
+<pt>
 
 ## ViTMAEModel
 
 [[autodoc]] ViTMAEModel
     - forward
 
-
 ## ViTMAEForPreTraining
 
 [[autodoc]] transformers.ViTMAEForPreTraining
     - forward
 
+</pt>
+<tf>
 
 ## TFViTMAEModel
 
 [[autodoc]] TFViTMAEModel
     - call
 
-
 ## TFViTMAEForPreTraining
 
 [[autodoc]] transformers.TFViTMAEForPreTraining
     - call
+
+</tf>
+</frameworkcontent>

docs/source/en/model_doc/vit_msn.md

@@ -33,7 +33,13 @@ while producing representations of a high semantic level that perform competitiv
 on ImageNet-1K, with only 5,000 annotated images, our base MSN model achieves 72.4% top-1 accuracy,
 and with 1% of ImageNet-1K labels, we achieve 75.7% top-1 accuracy, setting a new state-of-the-art for self-supervised learning on this benchmark.*
 
-Tips:
+<img src="https://i.ibb.co/W6PQMdC/Screenshot-2022-09-13-at-9-08-40-AM.png" alt="drawing" width="600"/> 
+
+<small> MSN architecture. Taken from the <a href="https://arxiv.org/abs/2204.07141">original paper.</a> </small>
+
+This model was contributed by [sayakpaul](https://huggingface.co/sayakpaul). The original code can be found [here](https://github.com/facebookresearch/msn). 
+
+## Usage tips
 
 - MSN (masked siamese networks) is a method for self-supervised pre-training of Vision Transformers (ViTs). The pre-training
 objective is to match the prototypes assigned to the unmasked views of the images to that of the masked views of the same images.
@@ -43,13 +49,6 @@ use the [`ViTMSNForImageClassification`] class which is initialized from [`ViTMS
 - MSN is particularly useful in the low-shot and extreme low-shot regimes. Notably, it achieves 75.7% top-1 accuracy with only 1% of ImageNet-1K
 labels when fine-tuned.
 
-
-<img src="https://i.ibb.co/W6PQMdC/Screenshot-2022-09-13-at-9-08-40-AM.png" alt="drawing" width="600"/> 
-
-<small> MSN architecture. Taken from the <a href="https://arxiv.org/abs/2204.07141">original paper.</a> </small>
-
-This model was contributed by [sayakpaul](https://huggingface.co/sayakpaul). The original code can be found [here](https://github.com/facebookresearch/msn). 
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with ViT MSN.
@@ -65,13 +64,11 @@ If you're interested in submitting a resource to be included here, please feel f
 
 [[autodoc]] ViTMSNConfig
 
-
 ## ViTMSNModel
 
 [[autodoc]] ViTMSNModel
     - forward
 
-
 ## ViTMSNForImageClassification
 
 [[autodoc]] ViTMSNForImageClassification

docs/source/en/model_doc/vitdet.md

@@ -21,13 +21,12 @@ The abstract from the paper is the following:
 
 *We explore the plain, non-hierarchical Vision Transformer (ViT) as a backbone network for object detection. This design enables the original ViT architecture to be fine-tuned for object detection without needing to redesign a hierarchical backbone for pre-training. With minimal adaptations for fine-tuning, our plain-backbone detector can achieve competitive results. Surprisingly, we observe: (i) it is sufficient to build a simple feature pyramid from a single-scale feature map (without the common FPN design) and (ii) it is sufficient to use window attention (without shifting) aided with very few cross-window propagation blocks. With plain ViT backbones pre-trained as Masked Autoencoders (MAE), our detector, named ViTDet, can compete with the previous leading methods that were all based on hierarchical backbones, reaching up to 61.3 AP_box on the COCO dataset using only ImageNet-1K pre-training. We hope our study will draw attention to research on plain-backbone detectors.*
 
-Tips:
-
-- For the moment, only the backbone is available.
-
 This model was contributed by [nielsr](https://huggingface.co/nielsr).
 The original code can be found [here](https://github.com/facebookresearch/detectron2/tree/main/projects/ViTDet).
 
+Tips:
+
+- At the moment, only the backbone is available.
 
 ## VitDetConfig
 

docs/source/en/model_doc/vitmatte.md

@@ -21,10 +21,6 @@ The abstract from the paper is the following:
 
 *Recently, plain vision Transformers (ViTs) have shown impressive performance on various computer vision tasks, thanks to their strong modeling capacity and large-scale pretraining. However, they have not yet conquered the problem of image matting. We hypothesize that image matting could also be boosted by ViTs and present a new efficient and robust ViT-based matting system, named ViTMatte. Our method utilizes (i) a hybrid attention mechanism combined with a convolution neck to help ViTs achieve an excellent performance-computation trade-off in matting tasks. (ii) Additionally, we introduce the detail capture module, which just consists of simple lightweight convolutions to complement the detailed information required by matting. To the best of our knowledge, ViTMatte is the first work to unleash the potential of ViT on image matting with concise adaptation. It inherits many superior properties from ViT to matting, including various pretraining strategies, concise architecture design, and flexible inference strategies. We evaluate ViTMatte on Composition-1k and Distinctions-646, the most commonly used benchmark for image matting, our method achieves state-of-the-art performance and outperforms prior matting works by a large margin.*
 
-Tips:
-
-- The model expects both the image and trimap (concatenated) as input. One can use [`ViTMatteImageProcessor`] for this purpose.
-
 This model was contributed by [nielsr](https://huggingface.co/nielsr).
 The original code can be found [here](https://github.com/hustvl/ViTMatte).
 
@@ -39,6 +35,10 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 
 - A demo notebook regarding inference with [`VitMatteForImageMatting`], including background replacement, can be found [here](https://github.com/NielsRogge/Transformers-Tutorials/tree/master/ViTMatte).
 
+<Tip>
+
+The model expects both the image and trimap (concatenated) as input. Use [`ViTMatteImageProcessor`] for this purpose.
+</Tip>
 
 ## VitMatteConfig
 

docs/source/en/model_doc/vits.md

@@ -16,7 +16,6 @@ specific language governing permissions and limitations under the License.
 
 The VITS model was proposed in [Conditional Variational Autoencoder with Adversarial Learning for End-to-End Text-to-Speech](https://arxiv.org/abs/2106.06103) by Jaehyeon Kim, Jungil Kong, Juhee Son.
 
-
 VITS (**V**ariational **I**nference with adversarial learning for end-to-end **T**ext-to-**S**peech) is an end-to-end 
 speech synthesis model that predicts a speech waveform conditional on an input text sequence. It is a conditional variational 
 autoencoder (VAE) comprised of a posterior encoder, decoder, and conditional prior.
@@ -42,7 +41,7 @@ as these checkpoints use the same architecture and a slightly modified tokenizer
 
 This model was contributed by [Matthijs](https://huggingface.co/Matthijs) and [sanchit-gandhi](https://huggingface.co/sanchit-gandhi). The original code can be found [here](https://github.com/jaywalnut310/vits).
 
-## Model Usage
+## Usage examples
 
 Both the VITS and MMS-TTS checkpoints can be used with the same API. Since the flow-based model is non-deterministic, it 
 is good practice to set a seed to ensure reproducibility of the outputs. For languages with a Roman alphabet, 

docs/source/en/model_doc/vivit.md

@@ -21,7 +21,6 @@ The abstract from the paper is the following:
 
 *We present pure-transformer based models for video classification, drawing upon the recent success of such models in image classification. Our model extracts spatio-temporal tokens from the input video, which are then encoded by a series of transformer layers. In order to handle the long sequences of tokens encountered in video, we propose several, efficient variants of our model which factorise the spatial- and temporal-dimensions of the input. Although transformer-based models are known to only be effective when large training datasets are available, we show how we can effectively regularise the model during training and leverage pretrained image models to be able to train on comparatively small datasets. We conduct thorough ablation studies, and achieve state-of-the-art results on multiple video classification benchmarks including Kinetics 400 and 600, Epic Kitchens, Something-Something v2 and Moments in Time, outperforming prior methods based on deep 3D convolutional networks.*
 
-
 This model was contributed by [jegormeister](https://huggingface.co/jegormeister). The original code (written in JAX) can be found [here](https://github.com/google-research/scenic/tree/main/scenic/projects/vivit).
 
 ## VivitConfig

docs/source/en/model_doc/wav2vec2-conformer.md

@@ -24,7 +24,10 @@ The official results of the model can be found in Table 3 and Table 4 of the pap
 
 The Wav2Vec2-Conformer weights were released by the Meta AI team within the [Fairseq library](https://github.com/pytorch/fairseq/blob/main/examples/wav2vec/README.md#pre-trained-models).
 
-Tips:
+This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten).
+The original code can be found [here](https://github.com/pytorch/fairseq/tree/main/examples/wav2vec).
+
+## Usage tips
 
 - Wav2Vec2-Conformer follows the same architecture as Wav2Vec2, but replaces the *Attention*-block with a *Conformer*-block
   as introduced in [Conformer: Convolution-augmented Transformer for Speech Recognition](https://arxiv.org/abs/2005.08100).
@@ -34,10 +37,7 @@ an improved word error rate.
 - Wav2Vec2-Conformer can use either no relative position embeddings, Transformer-XL-like position embeddings, or
   rotary position embeddings by setting the correct `config.position_embeddings_type`.
 
-This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten).
-The original code can be found [here](https://github.com/pytorch/fairseq/tree/main/examples/wav2vec).
-
-## Documentation resources
+## Resources
 
 - [Audio classification task guide](../tasks/audio_classification)
 - [Automatic speech recognition task guide](../tasks/asr)

docs/source/en/model_doc/wav2vec2.md

@@ -31,14 +31,14 @@ of the art on the 100 hour subset while using 100 times less labeled data. Using
 pre-training on 53k hours of unlabeled data still achieves 4.8/8.2 WER. This demonstrates the feasibility of speech
 recognition with limited amounts of labeled data.*
 
-Tips:
+This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten).
+
+## Usage tips
 
 - Wav2Vec2 is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
 - Wav2Vec2 model was trained using connectionist temporal classification (CTC) so the model output has to be decoded
   using [`Wav2Vec2CTCTokenizer`].
 
-This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten).
-
 ## Resources
 
 A list of official Hugging Face and community (indicated by 🌎) resources to help you get started with Wav2Vec2. If you're interested in submitting a resource to be included here, please feel free to open a Pull Request and we'll review it! The resource should ideally demonstrate something new instead of duplicating an existing resource.
@@ -167,6 +167,9 @@ Otherwise, [`~Wav2Vec2ProcessorWithLM.batch_decode`] performance will be slower
 
 [[autodoc]] models.wav2vec2.modeling_flax_wav2vec2.FlaxWav2Vec2ForPreTrainingOutput
 
+<frameworkcontent>
+<pt>
+
 ## Wav2Vec2Model
 
 [[autodoc]] Wav2Vec2Model
@@ -198,6 +201,9 @@ Otherwise, [`~Wav2Vec2ProcessorWithLM.batch_decode`] performance will be slower
 [[autodoc]] Wav2Vec2ForPreTraining
     - forward
 
+</pt>
+<tf>
+
 ## TFWav2Vec2Model
 
 [[autodoc]] TFWav2Vec2Model
@@ -213,6 +219,9 @@ Otherwise, [`~Wav2Vec2ProcessorWithLM.batch_decode`] performance will be slower
 [[autodoc]] TFWav2Vec2ForCTC
     - call
 
+</tf>
+<jax>
+
 ## FlaxWav2Vec2Model
 
 [[autodoc]] FlaxWav2Vec2Model
@@ -227,3 +236,6 @@ Otherwise, [`~Wav2Vec2ProcessorWithLM.batch_decode`] performance will be slower
 
 [[autodoc]] FlaxWav2Vec2ForPreTraining
     - __call__
+
+</jax>
+</frameworkcontent>

docs/source/en/model_doc/wav2vec2_phoneme.md

@@ -31,7 +31,13 @@ mapping phonemes of the training languages to the target language using articula
 this simple method significantly outperforms prior work which introduced task-specific architectures and used only part
 of a monolingually pretrained model.*
 
-Tips:
+Relevant checkpoints can be found under https://huggingface.co/models?other=phoneme-recognition.
+
+This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten)
+
+The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
+
+## Usage tips
 
 - Wav2Vec2Phoneme uses the exact same architecture as Wav2Vec2
 - Wav2Vec2Phoneme is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
@@ -39,17 +45,16 @@ Tips:
   decoded using [`Wav2Vec2PhonemeCTCTokenizer`].
 - Wav2Vec2Phoneme can be fine-tuned on multiple language at once and decode unseen languages in a single forward pass
   to a sequence of phonemes
-- By default the model outputs a sequence of phonemes. In order to transform the phonemes to a sequence of words one
+- By default, the model outputs a sequence of phonemes. In order to transform the phonemes to a sequence of words one
   should make use of a dictionary and language model.
 
-Relevant checkpoints can be found under https://huggingface.co/models?other=phoneme-recognition.
 
-This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten)
-
-The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
+<Tip>
 
-Wav2Vec2Phoneme's architecture is based on the Wav2Vec2 model, so one can refer to [`Wav2Vec2`]'s documentation page except for the tokenizer.
+Wav2Vec2Phoneme's architecture is based on the Wav2Vec2 model, for API reference, check out [`Wav2Vec2`](wav2vec2)'s documentation page 
+except for the tokenizer.
 
+</Tip>
 
 ## Wav2Vec2PhonemeCTCTokenizer
 

docs/source/en/model_doc/wavlm.md

@@ -35,7 +35,12 @@ additional overlapped utterances are created unsupervisely and incorporated duri
 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.*
 
-Tips:
+Relevant checkpoints can be found under https://huggingface.co/models?other=wavlm.
+
+This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten). The Authors' code can be
+found [here](https://github.com/microsoft/unilm/tree/master/wavlm).
+
+## Usage tips
 
 - WavLM is a speech model that accepts a float array corresponding to the raw waveform of the speech signal. Please use
   [`Wav2Vec2Processor`] for the feature extraction.
@@ -43,12 +48,7 @@ Tips:
   using [`Wav2Vec2CTCTokenizer`].
 - WavLM performs especially well on speaker verification, speaker identification, and speaker diarization tasks.
 
-Relevant checkpoints can be found under https://huggingface.co/models?other=wavlm.
-
-This model was contributed by [patrickvonplaten](https://huggingface.co/patrickvonplaten). The Authors' code can be
-found [here](https://github.com/microsoft/unilm/tree/master/wavlm).
-
-## Documentation resources
+## Resources
 
 - [Audio classification task guide](../tasks/audio_classification)
 - [Automatic speech recognition task guide](../tasks/asr)

docs/source/en/model_doc/whisper.md

@@ -24,18 +24,16 @@ The abstract from the paper is the following:
 
 *We study the capabilities of speech processing systems trained simply to predict large amounts of transcripts of audio on the internet. When scaled to 680,000 hours of multilingual and multitask supervision, the resulting models generalize well to standard benchmarks and are often competitive with prior fully supervised results but in a zeroshot transfer setting without the need for any finetuning. When compared to humans, the models approach their accuracy and robustness. We are releasing models and inference code to serve as a foundation for further work on robust speech processing.*
 
+This model was contributed by [Arthur Zucker](https://huggingface.co/ArthurZ). The Tensorflow version of this model was contributed by [amyeroberts](https://huggingface.co/amyeroberts).
+The original code can be found [here](https://github.com/openai/whisper).
 
-Tips:
+## Usage tips
 
 - The model usually performs well without requiring any finetuning.
 - The architecture follows a classic encoder-decoder architecture, which means that it relies on the [`~generation.GenerationMixin.generate`] function for inference.
 - Inference is currently only implemented for short-form i.e. audio is pre-segmented into <=30s segments. Long-form (including timestamps) will be implemented in a future release.
 - One can use [`WhisperProcessor`] to prepare audio for the model, and decode the predicted ID's back into text.
 
-This model was contributed by [Arthur Zucker](https://huggingface.co/ArthurZ). The Tensorflow version of this model was contributed by [amyeroberts](https://huggingface.co/amyeroberts).
-The original code can be found [here](https://github.com/openai/whisper).
-
-
 ## WhisperConfig
 
 [[autodoc]] WhisperConfig
@@ -76,6 +74,9 @@ The original code can be found [here](https://github.com/openai/whisper).
     - batch_decode
     - decode
 
+<frameworkcontent>
+<pt>
+
 ## WhisperModel
 
 [[autodoc]] WhisperModel
@@ -98,6 +99,8 @@ The original code can be found [here](https://github.com/openai/whisper).
 [[autodoc]] WhisperForAudioClassification
     - forward
 
+</pt>
+<tf>
 
 ## TFWhisperModel
 
@@ -109,6 +112,8 @@ The original code can be found [here](https://github.com/openai/whisper).
 [[autodoc]] TFWhisperForConditionalGeneration
     - call
 
+</tf>
+<jax>
 
 ## FlaxWhisperModel
 
@@ -125,3 +130,6 @@ The original code can be found [here](https://github.com/openai/whisper).
 [[autodoc]] FlaxWhisperForAudioClassification
     - __call__
 
+</jax>
+</frameworkcontent>
+

docs/source/en/model_doc/xglm.md

@@ -42,7 +42,7 @@ in social value tasks such as hate speech detection in five languages and find i
 
 This model was contributed by [Suraj](https://huggingface.co/valhalla). The original code can be found [here](https://github.com/pytorch/fairseq/tree/main/examples/xglm).
 
-## Documentation resources
+## Resources
 
 - [Causal language modeling task guide](../tasks/language_modeling)
 
@@ -62,6 +62,9 @@ This model was contributed by [Suraj](https://huggingface.co/valhalla). The orig
 
 [[autodoc]] XGLMTokenizerFast
 
+<frameworkcontent>
+<pt>
+
 ## XGLMModel
 
 [[autodoc]] XGLMModel
@@ -72,6 +75,9 @@ This model was contributed by [Suraj](https://huggingface.co/valhalla). The orig
 [[autodoc]] XGLMForCausalLM
     - forward
 
+</pt>
+<tf>
+
 ## TFXGLMModel
 
 [[autodoc]] TFXGLMModel
@@ -82,6 +88,9 @@ This model was contributed by [Suraj](https://huggingface.co/valhalla). The orig
 [[autodoc]] TFXGLMForCausalLM
     - call
 
+</tf>
+<jax>
+
 ## FlaxXGLMModel
 
 [[autodoc]] FlaxXGLMModel
@@ -90,4 +99,7 @@ This model was contributed by [Suraj](https://huggingface.co/valhalla). The orig
 ## FlaxXGLMForCausalLM
 
 [[autodoc]] FlaxXGLMForCausalLM
-    - __call__
+    - __call__
\ No newline at end of file
+
+</jax>
+</frameworkcontent>
\ No newline at end of file

docs/source/en/model_doc/xlm-prophetnet.md

@@ -36,7 +36,7 @@ Zhang, Ming Zhou on 13 Jan, 2020.
 
 XLM-ProphetNet is an encoder-decoder model and can predict n-future tokens for "ngram" language modeling instead of
 just the next token. Its architecture is identical to ProhpetNet, but the model was trained on the multi-lingual
-"wiki100" Wikipedia dump.
+"wiki100" Wikipedia dump. XLM-ProphetNet's model architecture and pretraining objective is same as ProphetNet, but XLM-ProphetNet was pre-trained on the cross-lingual dataset XGLUE.
 
 The abstract from the paper is the following:
 
@@ -52,11 +52,7 @@ state-of-the-art results on all these datasets compared to the models using the
 
 The Authors' code can be found [here](https://github.com/microsoft/ProphetNet).
 
-Tips:
-
-- XLM-ProphetNet's model architecture and pretraining objective is same as ProphetNet, but XLM-ProphetNet was pre-trained on the cross-lingual dataset XGLUE.
-
-## Documentation resources
+## Resources
 
 - [Causal language modeling task guide](../tasks/language_modeling)
 - [Translation task guide](../tasks/translation)

docs/source/en/model_doc/xlm-roberta-xl.md

@@ -24,15 +24,15 @@ The abstract from the paper is the following:
 
 *Recent work has demonstrated the effectiveness of cross-lingual language model pretraining for cross-lingual understanding. In this study, we present the results of two larger multilingual masked language models, with 3.5B and 10.7B parameters. Our two new models dubbed XLM-R XL and XLM-R XXL outperform XLM-R by 1.8% and 2.4% average accuracy on XNLI. Our model also outperforms the RoBERTa-Large model on several English tasks of the GLUE benchmark by 0.3% on average while handling 99 more languages. This suggests pretrained models with larger capacity may obtain both strong performance on high-resource languages while greatly improving low-resource languages. We make our code and models publicly available.*
 
-Tips:
+This model was contributed by [Soonhwan-Kwon](https://github.com/Soonhwan-Kwon) and [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/xlmr).
 
-- XLM-RoBERTa-XL is a multilingual model trained on 100 different languages. Unlike some XLM multilingual models, it does
-  not require `lang` tensors to understand which language is used, and should be able to determine the correct
-  language from the input ids.
+## Usage tips
 
-This model was contributed by [Soonhwan-Kwon](https://github.com/Soonhwan-Kwon) and [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/xlmr).
+XLM-RoBERTa-XL is a multilingual model trained on 100 different languages. Unlike some XLM multilingual models, it does 
+not require `lang` tensors to understand which language is used, and should be able to determine the correct 
+language from the input ids.
 
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)

docs/source/en/model_doc/xlm-roberta.md

@@ -46,16 +46,14 @@ languages at scale. Finally, we show, for the first time, the possibility of mul
 per-language performance; XLM-Ris very competitive with strong monolingual models on the GLUE and XNLI benchmarks. We
 will make XLM-R code, data, and models publicly available.*
 
-Tips:
+This model was contributed by [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/xlmr).
+
+## Usage tips
 
 - XLM-RoBERTa is a multilingual model trained on 100 different languages. Unlike some XLM multilingual models, it does
   not require `lang` tensors to understand which language is used, and should be able to determine the correct
   language from the input ids.
 - Uses RoBERTa tricks on the XLM approach, but does not use the translation language modeling objective. It only uses masked language modeling on sentences coming from one language.
-- This implementation is the same as RoBERTa. Refer to the [documentation of RoBERTa](roberta) for usage examples
-  as well as the information relative to the inputs and outputs.
-
-This model was contributed by [stefan-it](https://huggingface.co/stefan-it). The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/examples/xlmr).
 
 ## Resources
 
@@ -110,6 +108,11 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 
 - A blog post on how to [Deploy Serverless XLM RoBERTa on AWS Lambda](https://www.philschmid.de/multilingual-serverless-xlm-roberta-with-huggingface).
 
+<Tip> 
+
+This implementation is the same as RoBERTa. Refer to the [documentation of RoBERTa](roberta) for usage examples as well as the information relative to the inputs and outputs.
+</Tip>
+
 ## XLMRobertaConfig
 
 [[autodoc]] XLMRobertaConfig
@@ -126,6 +129,9 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 
 [[autodoc]] XLMRobertaTokenizerFast
 
+<frameworkcontent>
+<pt>
+
 ## XLMRobertaModel
 
 [[autodoc]] XLMRobertaModel
@@ -161,6 +167,9 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 [[autodoc]] XLMRobertaForQuestionAnswering
     - forward
 
+</pt>
+<tf>
+
 ## TFXLMRobertaModel
 
 [[autodoc]] TFXLMRobertaModel
@@ -196,6 +205,9 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 [[autodoc]] TFXLMRobertaForQuestionAnswering
     - call
 
+</tf>
+<jax>
+
 ## FlaxXLMRobertaModel
 
 [[autodoc]] FlaxXLMRobertaModel
@@ -230,3 +242,6 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 
 [[autodoc]] FlaxXLMRobertaForQuestionAnswering
     - __call__
+
+</jax>
+</frameworkcontent>
\ No newline at end of file

docs/source/en/model_doc/xlm-v.md

@@ -35,7 +35,10 @@ a multilingual language model with a one million token vocabulary. XLM-V outperf
 tested on ranging from natural language inference (XNLI), question answering (MLQA, XQuAD, TyDiQA), and
 named entity recognition (WikiAnn) to low-resource tasks (Americas NLI, MasakhaNER).*
 
-Tips:
+This model was contributed by [stefan-it](https://huggingface.co/stefan-it), including detailed experiments with XLM-V on downstream tasks.
+The experiments repository can be found [here](https://github.com/stefan-it/xlm-v-experiments).
+
+## Usage tips
 
 - XLM-V is compatible with the XLM-RoBERTa model architecture, only model weights from [`fairseq`](https://github.com/facebookresearch/fairseq)
   library had to be converted.
@@ -43,5 +46,7 @@ Tips:
 
 A XLM-V (base size) model is available under the [`facebook/xlm-v-base`](https://huggingface.co/facebook/xlm-v-base) identifier.
 
-This model was contributed by [stefan-it](https://huggingface.co/stefan-it), including detailed experiments with XLM-V on downstream tasks.
-The experiments repository can be found [here](https://github.com/stefan-it/xlm-v-experiments).
+<Tip>
+
+XLM-V architecture is the same as XLM-RoBERTa, refer to [XLM-RoBERTa documentation](xlm-roberta) for API reference, and examples.
+</Tip>
\ No newline at end of file

docs/source/en/model_doc/xlm.md

@@ -46,7 +46,9 @@ obtain 34.3 BLEU on WMT'16 German-English, improving the previous state of the a
 machine translation, we obtain a new state of the art of 38.5 BLEU on WMT'16 Romanian-English, outperforming the
 previous best approach by more than 4 BLEU. Our code and pretrained models will be made publicly available.*
 
-Tips:
+This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The original code can be found [here](https://github.com/facebookresearch/XLM/).
+
+## Usage tips
 
 - XLM has many different checkpoints, which were trained using different objectives: CLM, MLM or TLM. Make sure to
   select the correct objective for your task (e.g. MLM checkpoints are not suitable for generation).
@@ -57,9 +59,7 @@ Tips:
     * Masked language modeling (MLM) which is like RoBERTa. One of the languages is selected for each training sample, and the model input is a sentence of 256 tokens, that may span over several documents in one of those languages, with dynamic masking of the tokens.
     * A combination of MLM and translation language modeling (TLM). This consists of concatenating a sentence in two different languages, with random masking. To predict one of the masked tokens, the model can use both, the surrounding context in language 1 and the context given by language 2.
 
-This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The original code can be found [here](https://github.com/facebookresearch/XLM/).
-
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)
@@ -84,6 +84,9 @@ This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The o
 
 [[autodoc]] models.xlm.modeling_xlm.XLMForQuestionAnsweringOutput
 
+<frameworkcontent>
+<pt>
+
 ## XLMModel
 
 [[autodoc]] XLMModel
@@ -119,6 +122,9 @@ This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The o
 [[autodoc]] XLMForQuestionAnswering
     - forward
 
+</pt>
+<tf>
+
 ## TFXLMModel
 
 [[autodoc]] TFXLMModel
@@ -148,3 +154,8 @@ This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The o
 
 [[autodoc]] TFXLMForQuestionAnsweringSimple
     - call
+
+</tf>
+</frameworkcontent>
+
+

docs/source/en/model_doc/xlnet.md

@@ -44,7 +44,9 @@ formulation. Furthermore, XLNet integrates ideas from Transformer-XL, the state-
 pretraining. Empirically, under comparable experiment settings, XLNet outperforms BERT on 20 tasks, often by a large
 margin, including question answering, natural language inference, sentiment analysis, and document ranking.*
 
-Tips:
+This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The original code can be found [here](https://github.com/zihangdai/xlnet/).
+
+## Usage tips
 
 - The specific attention pattern can be controlled at training and test time using the `perm_mask` input.
 - Due to the difficulty of training a fully auto-regressive model over various factorization order, XLNet is pretrained
@@ -56,9 +58,7 @@ Tips:
 - XLNet is not a traditional autoregressive model but uses a training strategy that builds on that. It permutes the tokens in the sentence, then allows the model to use the last n tokens to predict the token n+1. Since this is all done with a mask, the sentence is actually fed in the model in the right order, but instead of masking the first n tokens for n+1, XLNet uses a mask that hides the previous tokens in some given permutation of 1,…,sequence length.
 - XLNet also uses the same recurrence mechanism as Transformer-XL to build long-term dependencies.
 
-This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The original code can be found [here](https://github.com/zihangdai/xlnet/).
-
-## Documentation resources
+## Resources
 
 - [Text classification task guide](../tasks/sequence_classification)
 - [Token classification task guide](../tasks/token_classification)
@@ -110,6 +110,9 @@ This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The o
 
 [[autodoc]] models.xlnet.modeling_tf_xlnet.TFXLNetForQuestionAnsweringSimpleOutput
 
+<frameworkcontent>
+<pt>
+
 ## XLNetModel
 
 [[autodoc]] XLNetModel
@@ -145,6 +148,9 @@ This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The o
 [[autodoc]] XLNetForQuestionAnswering
     - forward
 
+</pt>
+<tf>
+
 ## TFXLNetModel
 
 [[autodoc]] TFXLNetModel
@@ -174,3 +180,6 @@ This model was contributed by [thomwolf](https://huggingface.co/thomwolf). The o
 
 [[autodoc]] TFXLNetForQuestionAnsweringSimple
     - call
+
+</tf>
+</frameworkcontent>
\ No newline at end of file

docs/source/en/model_doc/xls_r.md

@@ -34,14 +34,18 @@ language identification. Moreover, we show that with sufficient model size, cros
 English-only pretraining when translating English speech into other languages, a setting which favors monolingual
 pretraining. We hope XLS-R can help to improve speech processing tasks for many more languages of the world.*
 
-Tips:
+Relevant checkpoints can be found under https://huggingface.co/models?other=xls_r.
+
+The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
+
+## Usage tips
 
 - XLS-R is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
 - XLS-R model was trained using connectionist temporal classification (CTC) so the model output has to be decoded using
   [`Wav2Vec2CTCTokenizer`].
 
-Relevant checkpoints can be found under https://huggingface.co/models?other=xls_r.
+<Tip>
 
-XLS-R's architecture is based on the Wav2Vec2 model, so one can refer to [Wav2Vec2's documentation page](wav2vec2).
+XLS-R's architecture is based on the Wav2Vec2 model, refer to [Wav2Vec2's documentation page](wav2vec2) for API reference.
 
-The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
+</Tip>
\ No newline at end of file

docs/source/en/model_doc/xlsr_wav2vec2.md

@@ -34,12 +34,16 @@ individual models. Analysis shows that the latent discrete speech representation
 increased sharing for related languages. We hope to catalyze research in low-resource speech understanding by releasing
 XLSR-53, a large model pretrained in 53 languages.*
 
-Tips:
+The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
+
+## Usage tips
 
 - XLSR-Wav2Vec2 is a speech model that accepts a float array corresponding to the raw waveform of the speech signal.
 - XLSR-Wav2Vec2 model was trained using connectionist temporal classification (CTC) so the model output has to be
   decoded using [`Wav2Vec2CTCTokenizer`].
 
+<Tip>
+
 XLSR-Wav2Vec2's architecture is based on the Wav2Vec2 model, so one can refer to [Wav2Vec2's documentation page](wav2vec2).
 
-The original code can be found [here](https://github.com/pytorch/fairseq/tree/master/fairseq/models/wav2vec).
+</Tip>