o Zh @sdZddlZddlZddlZddlZddlZddlmZddZddZ dd Z d d Z d Z d Z ddZddZdiddZdZdZdZdZdZdZdZdZdZdZdZd Zd!Zd"Zd#Zeeeeeeeeeeeeeed$Z d%Z!d&Z"eZ#eZ$d'Z%d(Z&d)Z'd)Z(d)Z)d)Z*d)Z+d)Z,eZ-d)Z.d)Z/d)Z0d)Z1eZ2d)Z3d)Z4d)Z5eZ6d)Z7eZ8d)Z9eZ:eZ;eZed+e!fd,e"fd-efd2e%fd3e0fd4e&fd5e'fd6e=fd7e2fd8e(fd9e5fd:e3fd;e)fde9fd?e+fd@e6fdAe7fdBe,fdCe-fdDe8fdEe:fdFe4fdGe/fdHe.fgZ?egdIZ@dJZAdKZBdLZCdMZDdNZEdOZFdPZGdQZHdRZIdSZJdTZKeCeBeAeFeDeGeEeHeIeJeKdU ZLdVZMdWZNdVZOdXZPdYZQdZZRd[ZSeOeNeMeRePeQeSd\ZTd]d^ZUddddd_d`dadddddddb dcddZVdjdedfZWdgdhZXdS)kz3 Doc utilities: Utilities related to documentation N) OrderedDictcCs@t|rdSt|}|d}t|t|}d|S)z^Return the indentation level of the start of the docstring of a class or function (or method).r)inspectisclass getsource splitlineslenlstrip)funcsource first_lineZfunction_def_levelr E/var/www/auris/lib/python3.10/site-packages/transformers/utils/doc.pyget_docstring_indentation_levels   rcfdd}|S)Ncs$d|jdur |jnd|_|SN)join__doc__fndocstrr rdocstring_decorator& z1add_start_docstrings..docstring_decoratorr rrr rradd_start_docstrings% rcr)Ncsd|jddd}d|d}t||jdur|jnd}ztdd |D}t|t|}Wn tyB}Ynw}|d kr_fd d D}t t |d }d ||}|||_|S)Nz[`.rz`]z The aa forward method, overrides the `__call__` special method. Although the recipe for forward pass needs to be defined within this function, one should call the [`Module`] instance afterwards instead of this since the former takes care of running the pre and post processing steps while the latter silently ignores them. rcss |] }|dkr|VqdS)rN)strip.0liner r r >szUadd_start_docstrings_to_model_forward..docstring_decorator..rcs"g|] }tt|dqS) )textwrapindentdedent)r!docZcorrect_indentationr r Gs"zVadd_start_docstrings_to_model_forward..docstring_decorator..r$) __qualname__splitrrnextrrr StopIterationr%r&r'r)r class_nameintroZ current_docZfirst_non_emptyZdoc_indentationdocs docstringrr)rr.s"     zBadd_start_docstrings_to_model_forward..docstring_decoratorr rr rr%add_start_docstrings_to_model_forward-s  r3cr)Ncs$|jdur|jndd|_|Sr)rrrrr rrRrz/add_end_docstrings..docstring_decoratorr rr rradd_end_docstringsQrr4a: Returns: [`{full_output_type}`] or `tuple(torch.FloatTensor)`: A [`{full_output_type}`] or a tuple of `torch.FloatTensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([`{config_class}`]) and inputs. a* Returns: [`{full_output_type}`] or `tuple(tf.Tensor)`: A [`{full_output_type}`] or a tuple of `tf.Tensor` (if `return_dict=False` is passed or when `config.return_dict=False`) comprising various elements depending on the configuration ([`{config_class}`]) and inputs. cCs$td|}|dur dS|dS)z.Returns the indentation in the first line of tz^(\s*)\SNrr)researchgroups)tr6r r r _get_indentks r9cCst|}g}d}|dD](}t||kr*t|dkr$||dd|d}q ||ddd7}q ||ddtt|D]}tdd||||<td d ||||<qEd|S) z,Convert output_args_doc to display properly.r rNz^(\s+)(\S+)(\s+)z \1- **\2**\3z :\s*\n\s*(\S)z -- \1)r9r,rappendranger5subr)Zoutput_args_docr&blocksZ current_blockr"ir r r_convert_output_args_docqs    rBTc s|j}d}|durU|d}d}|t|kr4td||dur4|d7}|t|kr4td||dus |t|krJd||dd}t|}n |rUtd|jd|rr|j d|j}|j d rht nt } | j ||d } nt|}d |d } |dur| d 7} | } |dur| |7} |dur| d}d}t||dkr|d7}t||dkstt||} | |krd|| fdd|D}d|} | S)zH Prepares the return part of the docstring using `output_type`. Nr:rz^\s*(Args|Parameters):\s*$z@No `Args` or `Parameters` section is found in the docstring of `zH`. Make sure it has docstring and contain either `Args` or `Parameters`.rTF)full_output_type config_classz Returns: ``z: r$cs(g|]}t|dkr|n|qS)r)rr Zto_addr rr*s(z._prepare_output_docstrings..)rr,rr5r6rrB ValueError__name__ __module__ startswithTF_RETURN_INTRODUCTIONPT_RETURN_INTRODUCTIONformatstrr9) output_typerF min_indentZ add_introZoutput_docstringZparams_docstringlinesrArEr0resultr&r rHr_prepare_output_docstringssL         rUaJ This example uses a random model as the real ones are all very big. To get proper results, you should use {real_checkpoint} instead of {fake_checkpoint}. If you get out-of-memory when loading that checkpoint, you can try adding `device_map="auto"` in the `from_pretrained` call. a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import torch >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer( ... "HuggingFace is a company based in Paris and New York", add_special_tokens=False, return_tensors="pt" ... ) >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_token_class_ids = logits.argmax(-1) >>> # Note that tokens are classified rather then input words which means that >>> # there might be more predicted token classes than words. >>> # Multiple token classes might account for the same word >>> predicted_tokens_classes = [model.config.id2label[t.item()] for t in predicted_token_class_ids[0]] >>> predicted_tokens_classes {expected_output} >>> labels = predicted_token_class_ids >>> loss = model(**inputs, labels=labels).loss >>> round(loss.item(), 2) {expected_loss} ``` a_ Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import torch >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet" >>> inputs = tokenizer(question, text, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**inputs) >>> answer_start_index = outputs.start_logits.argmax() >>> answer_end_index = outputs.end_logits.argmax() >>> predict_answer_tokens = inputs.input_ids[0, answer_start_index : answer_end_index + 1] >>> tokenizer.decode(predict_answer_tokens, skip_special_tokens=True) {expected_output} >>> # target is "nice puppet" >>> target_start_index = torch.tensor([{qa_target_start_index}]) >>> target_end_index = torch.tensor([{qa_target_end_index}]) >>> outputs = model(**inputs, start_positions=target_start_index, end_positions=target_end_index) >>> loss = outputs.loss >>> round(loss.item(), 2) {expected_loss} ``` a Example of single-label classification: ```python >>> import torch >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_class_id = logits.argmax().item() >>> model.config.id2label[predicted_class_id] {expected_output} >>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)` >>> num_labels = len(model.config.id2label) >>> model = {model_class}.from_pretrained("{checkpoint}", num_labels=num_labels) >>> labels = torch.tensor([1]) >>> loss = model(**inputs, labels=labels).loss >>> round(loss.item(), 2) {expected_loss} ``` Example of multi-label classification: ```python >>> import torch >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}", problem_type="multi_label_classification") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_class_ids = torch.arange(0, logits.shape[-1])[torch.sigmoid(logits).squeeze(dim=0) > 0.5] >>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)` >>> num_labels = len(model.config.id2label) >>> model = {model_class}.from_pretrained( ... "{checkpoint}", num_labels=num_labels, problem_type="multi_label_classification" ... ) >>> labels = torch.sum( ... torch.nn.functional.one_hot(predicted_class_ids[None, :].clone(), num_classes=num_labels), dim=1 ... ).to(torch.float) >>> loss = model(**inputs, labels=labels).loss ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import torch >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("The capital of France is {mask}.", return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> # retrieve index of {mask} >>> mask_token_index = (inputs.input_ids == tokenizer.mask_token_id)[0].nonzero(as_tuple=True)[0] >>> predicted_token_id = logits[0, mask_token_index].argmax(axis=-1) >>> tokenizer.decode(predicted_token_id) {expected_output} >>> labels = tokenizer("The capital of France is Paris.", return_tensors="pt")["input_ids"] >>> # mask labels of non-{mask} tokens >>> labels = torch.where(inputs.input_ids == tokenizer.mask_token_id, labels, -100) >>> outputs = model(**inputs, labels=labels) >>> round(outputs.loss.item(), 2) {expected_loss} ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import torch >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import torch >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced." >>> choice0 = "It is eaten with a fork and a knife." >>> choice1 = "It is eaten while held in the hand." >>> labels = torch.tensor(0).unsqueeze(0) # choice0 is correct (according to Wikipedia ;)), batch size 1 >>> encoding = tokenizer([prompt, prompt], [choice0, choice1], return_tensors="pt", padding=True) >>> outputs = model(**{{k: v.unsqueeze(0) for k, v in encoding.items()}}, labels=labels) # batch size is 1 >>> # the linear classifier still needs to be trained >>> loss = outputs.loss >>> logits = outputs.logits ``` a Example: ```python >>> import torch >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="pt") >>> outputs = model(**inputs, labels=inputs["input_ids"]) >>> loss = outputs.loss >>> logits = outputs.logits ``` aY Example: ```python >>> from transformers import AutoProcessor, {model_class} >>> import torch >>> from datasets import load_dataset >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = processor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state >>> list(last_hidden_states.shape) {expected_output} ``` au Example: ```python >>> from transformers import AutoProcessor, {model_class} >>> from datasets import load_dataset >>> import torch >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = processor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_ids = torch.argmax(logits, dim=-1) >>> # transcribe speech >>> transcription = processor.batch_decode(predicted_ids) >>> transcription[0] {expected_output} >>> inputs["labels"] = processor(text=dataset[0]["text"], return_tensors="pt").input_ids >>> # compute loss >>> loss = model(**inputs).loss >>> round(loss.item(), 2) {expected_loss} ``` a Example: ```python >>> from transformers import AutoFeatureExtractor, {model_class} >>> from datasets import load_dataset >>> import torch >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> feature_extractor = AutoFeatureExtractor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = feature_extractor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> predicted_class_ids = torch.argmax(logits, dim=-1).item() >>> predicted_label = model.config.id2label[predicted_class_ids] >>> predicted_label {expected_output} >>> # compute loss - target_label is e.g. "down" >>> target_label = model.config.id2label[0] >>> inputs["labels"] = torch.tensor([model.config.label2id[target_label]]) >>> loss = model(**inputs).loss >>> round(loss.item(), 2) {expected_loss} ``` a Example: ```python >>> from transformers import AutoFeatureExtractor, {model_class} >>> from datasets import load_dataset >>> import torch >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> feature_extractor = AutoFeatureExtractor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = feature_extractor(dataset[0]["audio"]["array"], return_tensors="pt", sampling_rate=sampling_rate) >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> probabilities = torch.sigmoid(logits[0]) >>> # labels is a one-hot array of shape (num_frames, num_speakers) >>> labels = (probabilities > 0.5).long() >>> labels[0].tolist() {expected_output} ``` a. Example: ```python >>> from transformers import AutoFeatureExtractor, {model_class} >>> from datasets import load_dataset >>> import torch >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> feature_extractor = AutoFeatureExtractor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = feature_extractor( ... [d["array"] for d in dataset[:2]["audio"]], sampling_rate=sampling_rate, return_tensors="pt", padding=True ... ) >>> with torch.no_grad(): ... embeddings = model(**inputs).embeddings >>> embeddings = torch.nn.functional.normalize(embeddings, dim=-1).cpu() >>> # the resulting embeddings can be used for cosine similarity-based retrieval >>> cosine_sim = torch.nn.CosineSimilarity(dim=-1) >>> similarity = cosine_sim(embeddings[0], embeddings[1]) >>> threshold = 0.7 # the optimal threshold is dataset-dependent >>> if similarity < threshold: ... print("Speakers are not the same!") >>> round(similarity.item(), 2) {expected_output} ``` a Example: ```python >>> from transformers import AutoImageProcessor, {model_class} >>> import torch >>> from datasets import load_dataset >>> dataset = load_dataset("huggingface/cats-image", trust_remote_code=True) >>> image = dataset["test"]["image"][0] >>> image_processor = AutoImageProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = image_processor(image, return_tensors="pt") >>> with torch.no_grad(): ... outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state >>> list(last_hidden_states.shape) {expected_output} ``` a Example: ```python >>> from transformers import AutoImageProcessor, {model_class} >>> import torch >>> from datasets import load_dataset >>> dataset = load_dataset("huggingface/cats-image", trust_remote_code=True) >>> image = dataset["test"]["image"][0] >>> image_processor = AutoImageProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = image_processor(image, return_tensors="pt") >>> with torch.no_grad(): ... logits = model(**inputs).logits >>> # model predicts one of the 1000 ImageNet classes >>> predicted_label = logits.argmax(-1).item() >>> print(model.config.id2label[predicted_label]) {expected_output} ``` )SequenceClassificationQuestionAnsweringTokenClassificationMultipleChoiceMaskedLMLMHead BaseModelSpeechBaseModelCTCAudioClassificationAudioFrameClassification AudioXVectorVisionBaseModelImageClassificationa Example: ```python >>> from transformers import AutoProcessor, {model_class}, SpeechT5HifiGan >>> model = {model_class}.from_pretrained("{checkpoint}") >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> vocoder = SpeechT5HifiGan.from_pretrained("microsoft/speecht5_hifigan") >>> inputs = processor(text="Hello, my dog is cute", return_tensors="pt") >>> # generate speech >>> speech = model.generate(inputs["input_ids"], speaker_embeddings=speaker_embeddings, vocoder=vocoder) ``` az Example: ```python >>> from transformers import AutoProcessor, {model_class} >>> model = {model_class}.from_pretrained("{checkpoint}") >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> inputs = processor(text="Hello, my dog is cute", return_tensors="pt") >>> # generate speech >>> speech = model(inputs["input_ids"]) ``` a Example: ```python >>> from PIL import Image >>> import requests >>> from transformers import AutoProcessor, {model_class} >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg" >>> image = Image.open(requests.get(url, stream=True).raw) >>> inputs = processor(images=image, return_tensors="pt") >>> outputs = model(**inputs) ``` a Example: ```python >>> from transformers import AutoImageProcessor, {model_class} >>> import torch >>> from PIL import Image >>> import requests >>> url = "http://images.cocodataset.org/val2017/000000039769.jpg" >>> image = Image.open(requests.get(url, stream=True).raw) >>> processor = AutoImageProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> device = torch.device("cuda" if torch.cuda.is_available() else "cpu") >>> model.to(device) >>> # prepare image for the model >>> inputs = processor(images=image, return_tensors="pt").to(device) >>> with torch.no_grad(): ... outputs = model(**inputs) >>> # interpolate to original size >>> post_processed_output = processor.post_process_depth_estimation( ... outputs, [(image.height, image.width)], ... ) >>> predicted_depth = post_processed_output[0]["predicted_depth"] ``` z% Example: ```python ``` a Example: ```python >>> from PIL import Image >>> import requests >>> from transformers import AutoProcessor, {model_class} >>> model = {model_class}.from_pretrained("{checkpoint}") >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> messages = [ ... {{ ... "role": "user", "content": [ ... {{"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg"}}, ... {{"type": "text", "text": "Where is the cat standing?"}}, ... ] ... }}, ... ] >>> inputs = processor.apply_chat_template( ... messages, ... tokenizer=True, ... return_dict=True, ... return_tensors="pt", ... add_generation_prompt=True ... ) >>> # Generate >>> generate_ids = model.generate(**inputs) >>> processor.batch_decode(generate_ids, skip_special_tokens=True)[0] ``` text-to-audio-spectrogramtext-to-audio-waveformautomatic-speech-recognitionaudio-frame-classificationaudio-classification audio-xvectorimage-text-to-text image-to-textvisual-question-answeringdepth-estimationvideo-classificationzero-shot-image-classificationimage-classificationzero-shot-object-detectionobject-detectionimage-segmentationimage-to-imageimage-feature-extractiontext-generationtable-question-answeringdocument-question-answeringquestion-answeringtext2text-generationnext-sentence-predictionmultiple-choicetext-classificationtoken-classification fill-maskmask-generation pretraining))Z+MODEL_FOR_TEXT_TO_SPECTROGRAM_MAPPING_NAMESrd)Z(MODEL_FOR_TEXT_TO_WAVEFORM_MAPPING_NAMESre)Z(MODEL_FOR_SPEECH_SEQ_2_SEQ_MAPPING_NAMESrf)ZMODEL_FOR_CTC_MAPPING_NAMESrf)Z2MODEL_FOR_AUDIO_FRAME_CLASSIFICATION_MAPPING_NAMESrg)Z,MODEL_FOR_AUDIO_CLASSIFICATION_MAPPING_NAMESrh)Z%MODEL_FOR_AUDIO_XVECTOR_MAPPING_NAMESri)Z*MODEL_FOR_IMAGE_TEXT_TO_TEXT_MAPPING_NAMESrj)Z$MODEL_FOR_VISION_2_SEQ_MAPPING_NAMESrk)Z1MODEL_FOR_VISUAL_QUESTION_ANSWERING_MAPPING_NAMESrl)Z(MODEL_FOR_DEPTH_ESTIMATION_MAPPING_NAMESrm)Z,MODEL_FOR_VIDEO_CLASSIFICATION_MAPPING_NAMESrn)Z6MODEL_FOR_ZERO_SHOT_IMAGE_CLASSIFICATION_MAPPING_NAMESro)Z,MODEL_FOR_IMAGE_CLASSIFICATION_MAPPING_NAMESrp)Z2MODEL_FOR_ZERO_SHOT_OBJECT_DETECTION_MAPPING_NAMESrq)Z(MODEL_FOR_OBJECT_DETECTION_MAPPING_NAMESrr)Z*MODEL_FOR_IMAGE_SEGMENTATION_MAPPING_NAMESrs)Z&MODEL_FOR_IMAGE_TO_IMAGE_MAPPING_NAMESrt)ZMODEL_FOR_IMAGE_MAPPING_NAMESru)Z!MODEL_FOR_CAUSAL_LM_MAPPING_NAMESrv)Z0MODEL_FOR_TABLE_QUESTION_ANSWERING_MAPPING_NAMESrw)Z3MODEL_FOR_DOCUMENT_QUESTION_ANSWERING_MAPPING_NAMESrx)Z*MODEL_FOR_QUESTION_ANSWERING_MAPPING_NAMESry)Z,MODEL_FOR_SEQ_TO_SEQ_CAUSAL_LM_MAPPING_NAMESrz)Z0MODEL_FOR_NEXT_SENTENCE_PREDICTION_MAPPING_NAMESr{)Z'MODEL_FOR_MULTIPLE_CHOICE_MAPPING_NAMESr|)Z/MODEL_FOR_SEQUENCE_CLASSIFICATION_MAPPING_NAMESr})Z,MODEL_FOR_TOKEN_CLASSIFICATION_MAPPING_NAMESr~)Z!MODEL_FOR_MASKED_LM_MAPPING_NAMESr)Z'MODEL_FOR_MASK_GENERATION_MAPPING_NAMESr)Z#MODEL_FOR_PRETRAINING_MAPPING_NAMESraI Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer( ... "HuggingFace is a company based in Paris and New York", add_special_tokens=False, return_tensors="tf" ... ) >>> logits = model(**inputs).logits >>> predicted_token_class_ids = tf.math.argmax(logits, axis=-1) >>> # Note that tokens are classified rather then input words which means that >>> # there might be more predicted token classes than words. >>> # Multiple token classes might account for the same word >>> predicted_tokens_classes = [model.config.id2label[t] for t in predicted_token_class_ids[0].numpy().tolist()] >>> predicted_tokens_classes {expected_output} ``` ```python >>> labels = predicted_token_class_ids >>> loss = tf.math.reduce_mean(model(**inputs, labels=labels).loss) >>> round(float(loss), 2) {expected_loss} ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet" >>> inputs = tokenizer(question, text, return_tensors="tf") >>> outputs = model(**inputs) >>> answer_start_index = int(tf.math.argmax(outputs.start_logits, axis=-1)[0]) >>> answer_end_index = int(tf.math.argmax(outputs.end_logits, axis=-1)[0]) >>> predict_answer_tokens = inputs.input_ids[0, answer_start_index : answer_end_index + 1] >>> tokenizer.decode(predict_answer_tokens) {expected_output} ``` ```python >>> # target is "nice puppet" >>> target_start_index = tf.constant([{qa_target_start_index}]) >>> target_end_index = tf.constant([{qa_target_end_index}]) >>> outputs = model(**inputs, start_positions=target_start_index, end_positions=target_end_index) >>> loss = tf.math.reduce_mean(outputs.loss) >>> round(float(loss), 2) {expected_loss} ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf") >>> logits = model(**inputs).logits >>> predicted_class_id = int(tf.math.argmax(logits, axis=-1)[0]) >>> model.config.id2label[predicted_class_id] {expected_output} ``` ```python >>> # To train a model on `num_labels` classes, you can pass `num_labels=num_labels` to `.from_pretrained(...)` >>> num_labels = len(model.config.id2label) >>> model = {model_class}.from_pretrained("{checkpoint}", num_labels=num_labels) >>> labels = tf.constant(1) >>> loss = model(**inputs, labels=labels).loss >>> round(float(loss), 2) {expected_loss} ``` a4 Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("The capital of France is {mask}.", return_tensors="tf") >>> logits = model(**inputs).logits >>> # retrieve index of {mask} >>> mask_token_index = tf.where((inputs.input_ids == tokenizer.mask_token_id)[0]) >>> selected_logits = tf.gather_nd(logits[0], indices=mask_token_index) >>> predicted_token_id = tf.math.argmax(selected_logits, axis=-1) >>> tokenizer.decode(predicted_token_id) {expected_output} ``` ```python >>> labels = tokenizer("The capital of France is Paris.", return_tensors="tf")["input_ids"] >>> # mask labels of non-{mask} tokens >>> labels = tf.where(inputs.input_ids == tokenizer.mask_token_id, labels, -100) >>> outputs = model(**inputs, labels=labels) >>> round(float(outputs.loss), 2) {expected_loss} ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf") >>> outputs = model(inputs) >>> last_hidden_states = outputs.last_hidden_state ``` a# Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced." >>> choice0 = "It is eaten with a fork and a knife." >>> choice1 = "It is eaten while held in the hand." >>> encoding = tokenizer([prompt, prompt], [choice0, choice1], return_tensors="tf", padding=True) >>> inputs = {{k: tf.expand_dims(v, 0) for k, v in encoding.items()}} >>> outputs = model(inputs) # batch size is 1 >>> # the linear classifier still needs to be trained >>> logits = outputs.logits ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> import tensorflow as tf >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="tf") >>> outputs = model(inputs) >>> logits = outputs.logits ``` a" Example: ```python >>> from transformers import AutoProcessor, {model_class} >>> from datasets import load_dataset >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = processor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="tf") >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state >>> list(last_hidden_states.shape) {expected_output} ``` aw Example: ```python >>> from transformers import AutoProcessor, {model_class} >>> from datasets import load_dataset >>> import tensorflow as tf >>> dataset = load_dataset("hf-internal-testing/librispeech_asr_demo", "clean", split="validation", trust_remote_code=True) >>> dataset = dataset.sort("id") >>> sampling_rate = dataset.features["audio"].sampling_rate >>> processor = AutoProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> # audio file is decoded on the fly >>> inputs = processor(dataset[0]["audio"]["array"], sampling_rate=sampling_rate, return_tensors="tf") >>> logits = model(**inputs).logits >>> predicted_ids = tf.math.argmax(logits, axis=-1) >>> # transcribe speech >>> transcription = processor.batch_decode(predicted_ids) >>> transcription[0] {expected_output} ``` ```python >>> inputs["labels"] = processor(text=dataset[0]["text"], return_tensors="tf").input_ids >>> # compute loss >>> loss = model(**inputs).loss >>> round(float(loss), 2) {expected_loss} ``` aq Example: ```python >>> from transformers import AutoImageProcessor, {model_class} >>> from datasets import load_dataset >>> dataset = load_dataset("huggingface/cats-image", trust_remote_code=True) >>> image = dataset["test"]["image"][0] >>> image_processor = AutoImageProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = image_processor(image, return_tensors="tf") >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state >>> list(last_hidden_states.shape) {expected_output} ``` a Example: ```python >>> from transformers import AutoImageProcessor, {model_class} >>> import tensorflow as tf >>> from datasets import load_dataset >>> dataset = load_dataset("huggingface/cats-image", trust_remote_code=True) >>> image = dataset["test"]["image"][0] >>> image_processor = AutoImageProcessor.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = image_processor(image, return_tensors="tf") >>> logits = model(**inputs).logits >>> # model predicts one of the 1000 ImageNet classes >>> predicted_label = int(tf.math.argmax(logits, axis=-1)) >>> print(model.config.id2label[predicted_label]) {expected_output} ``` ) rVrWrXrYrZr[r\r]r^rbrcar Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="jax") >>> outputs = model(**inputs) >>> logits = outputs.logits ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> question, text = "Who was Jim Henson?", "Jim Henson was a nice puppet" >>> inputs = tokenizer(question, text, return_tensors="jax") >>> outputs = model(**inputs) >>> start_scores = outputs.start_logits >>> end_scores = outputs.end_logits ``` a} Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("The capital of France is {mask}.", return_tensors="jax") >>> outputs = model(**inputs) >>> logits = outputs.logits ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="jax") >>> outputs = model(**inputs) >>> last_hidden_states = outputs.last_hidden_state ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> prompt = "In Italy, pizza served in formal settings, such as at a restaurant, is presented unsliced." >>> choice0 = "It is eaten with a fork and a knife." >>> choice1 = "It is eaten while held in the hand." >>> encoding = tokenizer([prompt, prompt], [choice0, choice1], return_tensors="jax", padding=True) >>> outputs = model(**{{k: v[None, :] for k, v in encoding.items()}}) >>> logits = outputs.logits ``` a Example: ```python >>> from transformers import AutoTokenizer, {model_class} >>> tokenizer = AutoTokenizer.from_pretrained("{checkpoint}") >>> model = {model_class}.from_pretrained("{checkpoint}") >>> inputs = tokenizer("Hello, my dog is cute", return_tensors="np") >>> outputs = model(**inputs) >>> # retrieve logts for next token >>> next_token_logits = outputs.logits[:, -1] ``` )rVrWrXrYrZr\r[cKsB|D]\}}|dur qd|d}td|dd|}q|S)zo Removes the lines testing an output with the doctest syntax in a code sample when it's set to `None`. N{}z\n([^\n]+)\n\s+z\nr:)itemsr5r?)r2kwargskeyvalueZdoc_keyr r rfilter_outputs_from_examples  rz[MASK]) processor_class checkpointrQrFmaskqa_target_start_indexqa_target_end_index model_clsmodalityexpected_output expected_lossreal_checkpointrevisionc s* fdd}|S)Nc sfdur |jddn}|dddkrt}n |dddkr$t}nt}|   dd }d |vs.docstring_decoratorr )rQrFrr rrreplace_return_docstringssrcCs6tj|j|j|j|j|jd}t||}|j |_ |S)zReturns a copy of a function f.)nameZargdefsclosure) types FunctionType__code__ __globals__rJ __defaults__ __closure__ functoolsupdate_wrapper__kwdefaults__)fgr r r copy_func(s r)NT)NN)Yrrrr5r%r collectionsrrrr3r4rNrMr9rBrUrZPT_TOKEN_CLASSIFICATION_SAMPLEZPT_QUESTION_ANSWERING_SAMPLEZ!PT_SEQUENCE_CLASSIFICATION_SAMPLEZPT_MASKED_LM_SAMPLEZPT_BASE_MODEL_SAMPLEZPT_MULTIPLE_CHOICE_SAMPLEZPT_CAUSAL_LM_SAMPLEZPT_SPEECH_BASE_MODEL_SAMPLEZPT_SPEECH_CTC_SAMPLEZPT_SPEECH_SEQ_CLASS_SAMPLEZPT_SPEECH_FRAME_CLASS_SAMPLEZPT_SPEECH_XVECTOR_SAMPLEZPT_VISION_BASE_MODEL_SAMPLEZPT_VISION_SEQ_CLASS_SAMPLErZ TEXT_TO_AUDIO_SPECTROGRAM_SAMPLEZTEXT_TO_AUDIO_WAVEFORM_SAMPLEZ!AUDIO_FRAME_CLASSIFICATION_SAMPLEZAUDIO_XVECTOR_SAMPLEZIMAGE_TO_TEXT_SAMPLEZDEPTH_ESTIMATION_SAMPLEZVIDEO_CLASSIFICATION_SAMPLEZ!ZERO_SHOT_OBJECT_DETECTION_SAMPLEZIMAGE_TO_IMAGE_SAMPLEZIMAGE_FEATURE_EXTRACTION_SAMPLEZ"DOCUMENT_QUESTION_ANSWERING_SAMPLEZNEXT_SENTENCE_PREDICTION_SAMPLEZMULTIPLE_CHOICE_SAMPLEZPRETRAINING_SAMPLEZMASK_GENERATION_SAMPLEZ VISUAL_QUESTION_ANSWERING_SAMPLEZTEXT_GENERATION_SAMPLEZIMAGE_CLASSIFICATION_SAMPLEZIMAGE_SEGMENTATION_SAMPLEZFILL_MASK_SAMPLEZOBJECT_DETECTION_SAMPLEZQUESTION_ANSWERING_SAMPLEZTEXT2TEXT_GENERATION_SAMPLEZTEXT_CLASSIFICATION_SAMPLEZTABLE_QUESTION_ANSWERING_SAMPLEZTOKEN_CLASSIFICATION_SAMPLEZAUDIO_CLASSIFICATION_SAMPLEZ#AUTOMATIC_SPEECH_RECOGNITION_SAMPLEZ%ZERO_SHOT_IMAGE_CLASSIFICATION_SAMPLEZ$IMAGE_TEXT_TO_TEXT_GENERATION_SAMPLEZ#PIPELINE_TASKS_TO_SAMPLE_DOCSTRINGSZMODELS_TO_PIPELINEZTF_TOKEN_CLASSIFICATION_SAMPLEZTF_QUESTION_ANSWERING_SAMPLEZ!TF_SEQUENCE_CLASSIFICATION_SAMPLEZTF_MASKED_LM_SAMPLEZTF_BASE_MODEL_SAMPLEZTF_MULTIPLE_CHOICE_SAMPLEZTF_CAUSAL_LM_SAMPLEZTF_SPEECH_BASE_MODEL_SAMPLEZTF_SPEECH_CTC_SAMPLEZTF_VISION_BASE_MODEL_SAMPLEZTF_VISION_SEQ_CLASS_SAMPLErZ FLAX_TOKEN_CLASSIFICATION_SAMPLEZFLAX_QUESTION_ANSWERING_SAMPLEZ#FLAX_SEQUENCE_CLASSIFICATION_SAMPLEZFLAX_MASKED_LM_SAMPLEZFLAX_BASE_MODEL_SAMPLEZFLAX_MULTIPLE_CHOICE_SAMPLEZFLAX_CAUSAL_LM_SAMPLErrrrrr r r rsN   $   5 !": #$#!"%(!#!$  c