Generate: move generation_*.py src files into generation/*.py (#20096)

* move generation_*.py src files into generation/*.py

* populate generation.__init__ with lazy loading

* move imports and references from generation.xxx.object to generation.object

docs/source/en/model_doc/bart.mdx

@@ -58,7 +58,7 @@ This model was contributed by [sshleifer](https://huggingface.co/sshleifer). The
 - Model predictions are intended to be identical to the original implementation when
   `forced_bos_token_id=0`. This only works, however, if the string you pass to
   [`fairseq.encode`] starts with a space.
-- [`~generation_utils.GenerationMixin.generate`] should be used for conditional generation tasks like
+- [`~generation.GenerationMixin.generate`] should be used for conditional generation tasks like
   summarization, see the example in that docstrings.
 - Models that load the *facebook/bart-large-cnn* weights will not have a `mask_token_id`, or be able to perform
   mask-filling tasks.
@@ -188,4 +188,4 @@ A list of official Hugging Face and community (indicated by 🌎) resources to h
 ## FlaxBartForCausalLM
 
 [[autodoc]] FlaxBartForCausalLM
-    - __call__
+    - __call__

docs/source/en/model_doc/donut.mdx

@@ -23,7 +23,7 @@ The abstract from the paper is the following:
 *Understanding document images (e.g., invoices) is a core but challenging task since it requires complex functions such as reading text and a holistic understanding of the document. Current Visual Document Understanding (VDU) methods outsource the task of reading text to off-the-shelf Optical Character Recognition (OCR) engines and focus on the understanding task with the OCR outputs. Although such OCR-based approaches have shown promising performance, they suffer from 1) high computational costs for using OCR; 2) inflexibility of OCR models on languages or types of document; 3) OCR error propagation to the subsequent process. To address these issues, in this paper, we introduce a novel OCR-free VDU model named Donut, which stands for Document understanding transformer. As the first step in OCR-free VDU research, we propose a simple architecture (i.e., Transformer) with a pre-training objective (i.e., cross-entropy loss). Donut is conceptually simple yet effective. Through extensive experiments and analyses, we show a simple OCR-free VDU model, Donut, achieves state-of-the-art performances on various VDU tasks in terms of both speed and accuracy. In addition, we offer a synthetic data generator that helps the model pre-training to be flexible in various languages and domains.*
 
 <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/transformers/model_doc/donut_architecture.jpg"
-alt="drawing" width="600"/> 
+alt="drawing" width="600"/>
 
 <small> Donut high-level overview. Taken from the <a href="https://arxiv.org/abs/2111.15664">original paper</a>. </small>
 
@@ -40,7 +40,7 @@ Tips:
 ## Inference
 
 Donut's [`VisionEncoderDecoder`] model accepts images as input and makes use of
-[`~generation_utils.GenerationMixin.generate`] to autoregressively generate text given the input image.
+[`~generation.GenerationMixin.generate`] to autoregressively generate text given the input image.
 
 The [`DonutFeatureExtractor`] class is responsible for preprocessing the input image and
 [`XLMRobertaTokenizer`/`XLMRobertaTokenizerFast`] decodes the generated target tokens to the target string. The
@@ -211,4 +211,4 @@ We refer to the [tutorial notebooks](https://github.com/NielsRogge/Transformers-
 ## DonutSwinModel
 
 [[autodoc]] DonutSwinModel
-    - forward
+    - forward

docs/source/en/model_doc/gptj.mdx

@@ -53,7 +53,7 @@ Tips:
 
 ### Generation
 
-The [`~generation_utils.GenerationMixin.generate`] method can be used to generate text using GPT-J
+The [`~generation.GenerationMixin.generate`] method can be used to generate text using GPT-J
 model.
 
 ```python

docs/source/en/model_doc/speech_to_text_2.mdx

@@ -38,7 +38,7 @@ Tips:
 ## Inference
 
 Speech2Text2's [`SpeechEncoderDecoderModel`] model accepts raw waveform input values from speech and
-makes use of [`~generation_utils.GenerationMixin.generate`] to translate the input speech
+makes use of [`~generation.GenerationMixin.generate`] to translate the input speech
 autoregressively to the target language.
 
 The [`Wav2Vec2FeatureExtractor`] class is responsible for preprocessing the input speech and

docs/source/en/model_doc/t5.mdx

@@ -225,7 +225,7 @@ batch) leads to very slow training on TPU.
 
 ## Inference
 
-At inference time, it is recommended to use [`~generation_utils.GenerationMixin.generate`]. This
+At inference time, it is recommended to use [`~generation.GenerationMixin.generate`]. This
 method takes care of encoding the input and feeding the encoded hidden states via cross-attention layers to the decoder
 and auto-regressively generates the decoder output. Check out [this blog post](https://huggingface.co/blog/how-to-generate) to know all the details about generating text with Transformers.
 There's also [this blog post](https://huggingface.co/blog/encoder-decoder#encoder-decoder) which explains how
@@ -244,7 +244,7 @@ Das Haus ist wunderbar.
 ```
 
 Note that T5 uses the `pad_token_id` as the `decoder_start_token_id`, so when doing generation without using
-[`~generation_utils.GenerationMixin.generate`], make sure you start it with the `pad_token_id`.
+[`~generation.GenerationMixin.generate`], make sure you start it with the `pad_token_id`.
 
 The example above only shows a single example. You can also do batched inference, like so:
 

docs/source/en/model_doc/trocr.mdx

@@ -30,7 +30,7 @@ show that the TrOCR model outperforms the current state-of-the-art models on bot
 tasks.*
 
 <img src="https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/trocr_architecture.jpg"
-alt="drawing" width="600"/> 
+alt="drawing" width="600"/>
 
 <small> TrOCR architecture. Taken from the <a href="https://arxiv.org/abs/2109.10282">original paper</a>. </small>
 
@@ -53,7 +53,7 @@ Tips:
 ## Inference
 
 TrOCR's [`VisionEncoderDecoder`] model accepts images as input and makes use of
-[`~generation_utils.GenerationMixin.generate`] to autoregressively generate text given the input image.
+[`~generation.GenerationMixin.generate`] to autoregressively generate text given the input image.
 
 The [`ViTFeatureExtractor`/`DeiTFeatureExtractor`] class is responsible for preprocessing the input image and
 [`RobertaTokenizer`/`XLMRobertaTokenizer`] decodes the generated target tokens to the target string. The
@@ -64,20 +64,20 @@ into a single instance to both extract the input features and decode the predict
 
 ``` py
 >>> from transformers import TrOCRProcessor, VisionEncoderDecoderModel
->>> import requests 
+>>> import requests
 >>> from PIL import Image
 
->>> processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten") 
+>>> processor = TrOCRProcessor.from_pretrained("microsoft/trocr-base-handwritten")
 >>> model = VisionEncoderDecoderModel.from_pretrained("microsoft/trocr-base-handwritten")
 
->>> # load image from the IAM dataset 
->>> url = "https://fki.tic.heia-fr.ch/static/img/a01-122-02.jpg" 
+>>> # load image from the IAM dataset
+>>> url = "https://fki.tic.heia-fr.ch/static/img/a01-122-02.jpg"
 >>> image = Image.open(requests.get(url, stream=True).raw).convert("RGB")
 
->>> pixel_values = processor(image, return_tensors="pt").pixel_values 
+>>> pixel_values = processor(image, return_tensors="pt").pixel_values
 >>> generated_ids = model.generate(pixel_values)
 
->>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0] 
+>>> generated_text = processor.batch_decode(generated_ids, skip_special_tokens=True)[0]
 ```
 
 See the [model hub](https://huggingface.co/models?filter=trocr) to look for TrOCR checkpoints.

docs/source/en/model_doc/whisper.mdx

@@ -24,7 +24,7 @@ The abstract from the paper is the following:
 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_utils.GenerationMixin.generate`] function for inference.
+- 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.