transformers_doc/task_summary.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# Transformers installation\n",
    "! pip install transformers datasets\n",
    "# To install from source instead of the last release, comment the command above and uncomment the following one.\n",
    "# ! pip install git+https://github.com/huggingface/transformers.git"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Summary of the tasks"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This page shows the most frequent use-cases when using the library. The models available allow for many different\n",
    "configurations and a great versatility in use-cases. The most simple ones are presented here, showcasing usage for\n",
    "tasks such as question answering, sequence classification, named entity recognition and others.\n",
    "\n",
    "These examples leverage auto-models, which are classes that will instantiate a model according to a given checkpoint,\n",
    "automatically selecting the correct model architecture. Please check the [AutoModel](https://huggingface.co/docs/transformers/main/en/model_doc/auto#transformers.AutoModel) documentation\n",
    "for more information. Feel free to modify the code to be more specific and adapt it to your specific use-case.\n",
    "\n",
    "In order for a model to perform well on a task, it must be loaded from a checkpoint corresponding to that task. These\n",
    "checkpoints are usually pre-trained on a large corpus of data and fine-tuned on a specific task. This means the\n",
    "following:\n",
    "\n",
    "- Not all models were fine-tuned on all tasks. If you want to fine-tune a model on a specific task, you can leverage\n",
    "  one of the *run_$TASK.py* scripts in the [examples](https://github.com/huggingface/transformers/tree/main/examples) directory.\n",
    "- Fine-tuned models were fine-tuned on a specific dataset. This dataset may or may not overlap with your use-case and\n",
    "  domain. As mentioned previously, you may leverage the [examples](https://github.com/huggingface/transformers/tree/main/examples) scripts to fine-tune your model, or you may\n",
    "  create your own training script.\n",
    "\n",
    "In order to do an inference on a task, several mechanisms are made available by the library:\n",
    "\n",
    "- Pipelines: very easy-to-use abstractions, which require as little as two lines of code.\n",
    "- Direct model use: Less abstractions, but more flexibility and power via a direct access to a tokenizer\n",
    "  (PyTorch/TensorFlow) and full inference capacity.\n",
    "\n",
    "Both approaches are showcased here.\n",
    "\n",
    "<Tip>\n",
    "\n",
    "All tasks presented here leverage pre-trained checkpoints that were fine-tuned on specific tasks. Loading a\n",
    "checkpoint that was not fine-tuned on a specific task would load only the base transformer layers and not the\n",
    "additional head that is used for the task, initializing the weights of that head randomly.\n",
    "\n",
    "This would produce random output.\n",
    "\n",
    "</Tip>"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Sequence Classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sequence classification is the task of classifying sequences according to a given number of classes. An example of\n",
    "sequence classification is the GLUE dataset, which is entirely based on that task. If you would like to fine-tune a\n",
    "model on a GLUE sequence classification task, you may leverage the [run_glue.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/text-classification/run_glue.py), [run_tf_glue.py](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/text-classification/run_tf_glue.py), [run_tf_text_classification.py](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/text-classification/run_tf_text_classification.py) or [run_xnli.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/text-classification/run_xnli.py) scripts.\n",
    "\n",
    "Here is an example of using pipelines to do sentiment analysis: identifying if a sequence is positive or negative. It\n",
    "leverages a fine-tuned model on sst2, which is a GLUE task.\n",
    "\n",
    "This returns a label (\"POSITIVE\" or \"NEGATIVE\") alongside a score, as follows:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "label: NEGATIVE, with score: 0.9991"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "classifier = pipeline(\"sentiment-analysis\")\n",
    "\n",
    "result = classifier(\"I hate you\")[0]\n",
    "print(f\"label: {result['label']}, with score: {round(result['score'], 4)}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "label: POSITIVE, with score: 0.9999"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = classifier(\"I love you\")[0]\n",
    "print(f\"label: {result['label']}, with score: {round(result['score'], 4)}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is an example of doing a sequence classification using a model to determine if two sequences are paraphrases of\n",
    "each other. The process is the following:\n",
    "\n",
    "1. Instantiate a tokenizer and a model from the checkpoint name. The model is identified as a BERT model and loads it\n",
    "   with the weights stored in the checkpoint.\n",
    "2. Build a sequence from the two sentences, with the correct model-specific separators, token type ids and attention\n",
    "   masks (which will be created automatically by the tokenizer).\n",
    "3. Pass this sequence through the model so that it is classified in one of the two available classes: 0 (not a\n",
    "   paraphrase) and 1 (is a paraphrase).\n",
    "4. Compute the softmax of the result to get probabilities over the classes.\n",
    "5. Print the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "not paraphrase: 10%\n",
       "is paraphrase: 90%"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoTokenizer, AutoModelForSequenceClassification\n",
    "import torch\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"bert-base-cased-finetuned-mrpc\")\n",
    "model = AutoModelForSequenceClassification.from_pretrained(\"bert-base-cased-finetuned-mrpc\")\n",
    "\n",
    "classes = [\"not paraphrase\", \"is paraphrase\"]\n",
    "\n",
    "sequence_0 = \"The company HuggingFace is based in New York City\"\n",
    "sequence_1 = \"Apples are especially bad for your health\"\n",
    "sequence_2 = \"HuggingFace's headquarters are situated in Manhattan\"\n",
    "\n",
    "# The tokenizer will automatically add any model specific separators (i.e. <CLS> and <SEP>) and tokens to\n",
    "# the sequence, as well as compute the attention masks.\n",
    "paraphrase = tokenizer(sequence_0, sequence_2, return_tensors=\"pt\")\n",
    "not_paraphrase = tokenizer(sequence_0, sequence_1, return_tensors=\"pt\")\n",
    "\n",
    "paraphrase_classification_logits = model(**paraphrase).logits\n",
    "not_paraphrase_classification_logits = model(**not_paraphrase).logits\n",
    "\n",
    "paraphrase_results = torch.softmax(paraphrase_classification_logits, dim=1).tolist()[0]\n",
    "not_paraphrase_results = torch.softmax(not_paraphrase_classification_logits, dim=1).tolist()[0]\n",
    "\n",
    "# Should be paraphrase\n",
    "for i in range(len(classes)):\n",
    "    print(f\"{classes[i]}: {int(round(paraphrase_results[i] * 100))}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "not paraphrase: 94%\n",
       "is paraphrase: 6%"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Should not be paraphrase\n",
    "for i in range(len(classes)):\n",
    "    print(f\"{classes[i]}: {int(round(not_paraphrase_results[i] * 100))}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "not paraphrase: 10%\n",
       "is paraphrase: 90%"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoTokenizer, TFAutoModelForSequenceClassification\n",
    "import tensorflow as tf\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"bert-base-cased-finetuned-mrpc\")\n",
    "model = TFAutoModelForSequenceClassification.from_pretrained(\"bert-base-cased-finetuned-mrpc\")\n",
    "\n",
    "classes = [\"not paraphrase\", \"is paraphrase\"]\n",
    "\n",
    "sequence_0 = \"The company HuggingFace is based in New York City\"\n",
    "sequence_1 = \"Apples are especially bad for your health\"\n",
    "sequence_2 = \"HuggingFace's headquarters are situated in Manhattan\"\n",
    "\n",
    "# The tokenizer will automatically add any model specific separators (i.e. <CLS> and <SEP>) and tokens to\n",
    "# the sequence, as well as compute the attention masks.\n",
    "paraphrase = tokenizer(sequence_0, sequence_2, return_tensors=\"tf\")\n",
    "not_paraphrase = tokenizer(sequence_0, sequence_1, return_tensors=\"tf\")\n",
    "\n",
    "paraphrase_classification_logits = model(paraphrase).logits\n",
    "not_paraphrase_classification_logits = model(not_paraphrase).logits\n",
    "\n",
    "paraphrase_results = tf.nn.softmax(paraphrase_classification_logits, axis=1).numpy()[0]\n",
    "not_paraphrase_results = tf.nn.softmax(not_paraphrase_classification_logits, axis=1).numpy()[0]\n",
    "\n",
    "# Should be paraphrase\n",
    "for i in range(len(classes)):\n",
    "    print(f\"{classes[i]}: {int(round(paraphrase_results[i] * 100))}%\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "not paraphrase: 94%\n",
       "is paraphrase: 6%"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Should not be paraphrase\n",
    "for i in range(len(classes)):\n",
    "    print(f\"{classes[i]}: {int(round(not_paraphrase_results[i] * 100))}%\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Extractive Question Answering"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Extractive Question Answering is the task of extracting an answer from a text given a question. An example of a\n",
    "question answering dataset is the SQuAD dataset, which is entirely based on that task. If you would like to fine-tune a\n",
    "model on a SQuAD task, you may leverage the [run_qa.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/question-answering/run_qa.py) and\n",
    "[run_tf_squad.py](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/question-answering/run_tf_squad.py)\n",
    "scripts.\n",
    "\n",
    "\n",
    "Here is an example of using pipelines to do question answering: extracting an answer from a text given a question. It\n",
    "leverages a fine-tuned model on SQuAD."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "question_answerer = pipeline(\"question-answering\")\n",
    "\n",
    "context = r\"\"\"\n",
    "Extractive Question Answering is the task of extracting an answer from a text given a question. An example of a\n",
    "question answering dataset is the SQuAD dataset, which is entirely based on that task. If you would like to fine-tune\n",
    "a model on a SQuAD task, you may leverage the examples/pytorch/question-answering/run_squad.py script.\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This returns an answer extracted from the text, a confidence score, alongside \"start\" and \"end\" values, which are the\n",
    "positions of the extracted answer in the text."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Answer: 'the task of extracting an answer from a text given a question', score: 0.6177, start: 34, end: 95"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = question_answerer(question=\"What is extractive question answering?\", context=context)\n",
    "print(\n",
    "    f\"Answer: '{result['answer']}', score: {round(result['score'], 4)}, start: {result['start']}, end: {result['end']}\"\n",
    ")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Answer: 'SQuAD dataset', score: 0.5152, start: 147, end: 160"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = question_answerer(question=\"What is a good example of a question answering dataset?\", context=context)\n",
    "print(\n",
    "    f\"Answer: '{result['answer']}', score: {round(result['score'], 4)}, start: {result['start']}, end: {result['end']}\"\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is an example of question answering using a model and a tokenizer. The process is the following:\n",
    "\n",
    "1. Instantiate a tokenizer and a model from the checkpoint name. The model is identified as a BERT model and loads it\n",
    "   with the weights stored in the checkpoint.\n",
    "2. Define a text and a few questions.\n",
    "3. Iterate over the questions and build a sequence from the text and the current question, with the correct\n",
    "   model-specific separators, token type ids and attention masks.\n",
    "4. Pass this sequence through the model. This outputs a range of scores across the entire sequence tokens (question and\n",
    "   text), for both the start and end positions.\n",
    "5. Compute the softmax of the result to get probabilities over the tokens.\n",
    "6. Fetch the tokens from the identified start and stop values, convert those tokens to a string.\n",
    "7. Print the results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Question: How many pretrained models are available in 🤗 Transformers?\n",
       "Answer: over 32 +\n",
       "Question: What does 🤗 Transformers provide?\n",
       "Answer: general - purpose architectures\n",
       "Question: 🤗 Transformers provides interoperability between which frameworks?\n",
       "Answer: tensorflow 2. 0 and pytorch"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoTokenizer, AutoModelForQuestionAnswering\n",
    "import torch\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"bert-large-uncased-whole-word-masking-finetuned-squad\")\n",
    "model = AutoModelForQuestionAnswering.from_pretrained(\"bert-large-uncased-whole-word-masking-finetuned-squad\")\n",
    "\n",
    "text = r\"\"\"\n",
    "🤗 Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides general-purpose\n",
    "architectures (BERT, GPT-2, RoBERTa, XLM, DistilBert, XLNet…) for Natural Language Understanding (NLU) and Natural\n",
    "Language Generation (NLG) with over 32+ pretrained models in 100+ languages and deep interoperability between\n",
    "TensorFlow 2.0 and PyTorch.\n",
    "\"\"\"\n",
    "\n",
    "questions = [\n",
    "    \"How many pretrained models are available in 🤗 Transformers?\",\n",
    "    \"What does 🤗 Transformers provide?\",\n",
    "    \"🤗 Transformers provides interoperability between which frameworks?\",\n",
    "]\n",
    "\n",
    "for question in questions:\n",
    "    inputs = tokenizer(question, text, add_special_tokens=True, return_tensors=\"pt\")\n",
    "    input_ids = inputs[\"input_ids\"].tolist()[0]\n",
    "\n",
    "    outputs = model(**inputs)\n",
    "    answer_start_scores = outputs.start_logits\n",
    "    answer_end_scores = outputs.end_logits\n",
    "\n",
    "    # Get the most likely beginning of answer with the argmax of the score\n",
    "    answer_start = torch.argmax(answer_start_scores)\n",
    "    # Get the most likely end of answer with the argmax of the score\n",
    "    answer_end = torch.argmax(answer_end_scores) + 1\n",
    "\n",
    "    answer = tokenizer.convert_tokens_to_string(\n",
    "        tokenizer.convert_ids_to_tokens(input_ids[answer_start:answer_end])\n",
    "    )\n",
    "\n",
    "    print(f\"Question: {question}\")\n",
    "    print(f\"Answer: {answer}\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Question: How many pretrained models are available in 🤗 Transformers?\n",
       "Answer: over 32 +\n",
       "Question: What does 🤗 Transformers provide?\n",
       "Answer: general - purpose architectures\n",
       "Question: 🤗 Transformers provides interoperability between which frameworks?\n",
       "Answer: tensorflow 2. 0 and pytorch"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoTokenizer, TFAutoModelForQuestionAnswering\n",
    "import tensorflow as tf\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"bert-large-uncased-whole-word-masking-finetuned-squad\")\n",
    "model = TFAutoModelForQuestionAnswering.from_pretrained(\"bert-large-uncased-whole-word-masking-finetuned-squad\")\n",
    "\n",
    "text = r\"\"\"\n",
    "🤗 Transformers (formerly known as pytorch-transformers and pytorch-pretrained-bert) provides general-purpose\n",
    "architectures (BERT, GPT-2, RoBERTa, XLM, DistilBert, XLNet…) for Natural Language Understanding (NLU) and Natural\n",
    "Language Generation (NLG) with over 32+ pretrained models in 100+ languages and deep interoperability between\n",
    "TensorFlow 2.0 and PyTorch.\n",
    "\"\"\"\n",
    "\n",
    "questions = [\n",
    "    \"How many pretrained models are available in 🤗 Transformers?\",\n",
    "    \"What does 🤗 Transformers provide?\",\n",
    "    \"🤗 Transformers provides interoperability between which frameworks?\",\n",
    "]\n",
    "\n",
    "for question in questions:\n",
    "    inputs = tokenizer(question, text, add_special_tokens=True, return_tensors=\"tf\")\n",
    "    input_ids = inputs[\"input_ids\"].numpy()[0]\n",
    "\n",
    "    outputs = model(inputs)\n",
    "    answer_start_scores = outputs.start_logits\n",
    "    answer_end_scores = outputs.end_logits\n",
    "\n",
    "    # Get the most likely beginning of answer with the argmax of the score\n",
    "    answer_start = tf.argmax(answer_start_scores, axis=1).numpy()[0]\n",
    "    # Get the most likely end of answer with the argmax of the score\n",
    "    answer_end = tf.argmax(answer_end_scores, axis=1).numpy()[0] + 1\n",
    "\n",
    "    answer = tokenizer.convert_tokens_to_string(\n",
    "        tokenizer.convert_ids_to_tokens(input_ids[answer_start:answer_end])\n",
    "    )\n",
    "\n",
    "    print(f\"Question: {question}\")\n",
    "    print(f\"Answer: {answer}\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Language Modeling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Language modeling is the task of fitting a model to a corpus, which can be domain specific. All popular\n",
    "transformer-based models are trained using a variant of language modeling, e.g. BERT with masked language modeling,\n",
    "GPT-2 with causal language modeling.\n",
    "\n",
    "Language modeling can be useful outside of pretraining as well, for example to shift the model distribution to be\n",
    "domain-specific: using a language model trained over a very large corpus, and then fine-tuning it to a news dataset or\n",
    "on scientific papers e.g. [LysandreJik/arxiv-nlp](https://huggingface.co/lysandre/arxiv-nlp)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Masked Language Modeling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Masked language modeling is the task of masking tokens in a sequence with a masking token, and prompting the model to\n",
    "fill that mask with an appropriate token. This allows the model to attend to both the right context (tokens on the\n",
    "right of the mask) and the left context (tokens on the left of the mask). Such a training creates a strong basis for\n",
    "downstream tasks requiring bi-directional context, such as SQuAD (question answering, see [Lewis, Lui, Goyal et al.](https://arxiv.org/abs/1910.13461), part 4.2). If you would like to fine-tune a model on a masked language modeling\n",
    "task, you may leverage the [run_mlm.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/language-modeling/run_mlm.py) script.\n",
    "\n",
    "Here is an example of using pipelines to replace a mask from a sequence:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "unmasker = pipeline(\"fill-mask\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This outputs the sequences with the mask filled, the confidence score, and the token id in the tokenizer vocabulary:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'score': 0.1793,\n",
       "  'sequence': 'HuggingFace is creating a tool that the community uses to solve '\n",
       "              'NLP tasks.',\n",
       "  'token': 3944,\n",
       "  'token_str': ' tool'},\n",
       " {'score': 0.1135,\n",
       "  'sequence': 'HuggingFace is creating a framework that the community uses to '\n",
       "              'solve NLP tasks.',\n",
       "  'token': 7208,\n",
       "  'token_str': ' framework'},\n",
       " {'score': 0.0524,\n",
       "  'sequence': 'HuggingFace is creating a library that the community uses to '\n",
       "              'solve NLP tasks.',\n",
       "  'token': 5560,\n",
       "  'token_str': ' library'},\n",
       " {'score': 0.0349,\n",
       "  'sequence': 'HuggingFace is creating a database that the community uses to '\n",
       "              'solve NLP tasks.',\n",
       "  'token': 8503,\n",
       "  'token_str': ' database'},\n",
       " {'score': 0.0286,\n",
       "  'sequence': 'HuggingFace is creating a prototype that the community uses to '\n",
       "              'solve NLP tasks.',\n",
       "  'token': 17715,\n",
       "  'token_str': ' prototype'}]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from pprint import pprint\n",
    "\n",
    "pprint(\n",
    "    unmasker(\n",
    "        f\"HuggingFace is creating a {unmasker.tokenizer.mask_token} that the community uses to solve NLP tasks.\"\n",
    "    )\n",
    ")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is an example of doing masked language modeling using a model and a tokenizer. The process is the following:\n",
    "\n",
    "1. Instantiate a tokenizer and a model from the checkpoint name. The model is identified as a DistilBERT model and\n",
    "   loads it with the weights stored in the checkpoint.\n",
    "2. Define a sequence with a masked token, placing the `tokenizer.mask_token` instead of a word.\n",
    "3. Encode that sequence into a list of IDs and find the position of the masked token in that list.\n",
    "4. Retrieve the predictions at the index of the mask token: this tensor has the same size as the vocabulary, and the\n",
    "   values are the scores attributed to each token. The model gives higher score to tokens it deems probable in that\n",
    "   context.\n",
    "5. Retrieve the top 5 tokens using the PyTorch `topk` or TensorFlow `top_k` methods.\n",
    "6. Replace the mask token by the tokens and print the results"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help reduce our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help increase our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help decrease our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help offset our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help improve our carbon footprint."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoModelForMaskedLM, AutoTokenizer\n",
    "import torch\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"distilbert-base-cased\")\n",
    "model = AutoModelForMaskedLM.from_pretrained(\"distilbert-base-cased\")\n",
    "\n",
    "sequence = (\n",
    "    \"Distilled models are smaller than the models they mimic. Using them instead of the large \"\n",
    "    f\"versions would help {tokenizer.mask_token} our carbon footprint.\"\n",
    ")\n",
    "\n",
    "inputs = tokenizer(sequence, return_tensors=\"pt\")\n",
    "mask_token_index = torch.where(inputs[\"input_ids\"] == tokenizer.mask_token_id)[1]\n",
    "\n",
    "token_logits = model(**inputs).logits\n",
    "mask_token_logits = token_logits[0, mask_token_index, :]\n",
    "\n",
    "top_5_tokens = torch.topk(mask_token_logits, 5, dim=1).indices[0].tolist()\n",
    "\n",
    "for token in top_5_tokens:\n",
    "    print(sequence.replace(tokenizer.mask_token, tokenizer.decode([token])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help reduce our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help increase our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help decrease our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help offset our carbon footprint.\n",
       "Distilled models are smaller than the models they mimic. Using them instead of the large versions would help improve our carbon footprint."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import TFAutoModelForMaskedLM, AutoTokenizer\n",
    "import tensorflow as tf\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"distilbert-base-cased\")\n",
    "model = TFAutoModelForMaskedLM.from_pretrained(\"distilbert-base-cased\")\n",
    "\n",
    "sequence = (\n",
    "    \"Distilled models are smaller than the models they mimic. Using them instead of the large \"\n",
    "    f\"versions would help {tokenizer.mask_token} our carbon footprint.\"\n",
    ")\n",
    "\n",
    "inputs = tokenizer(sequence, return_tensors=\"tf\")\n",
    "mask_token_index = tf.where(inputs[\"input_ids\"] == tokenizer.mask_token_id)[0, 1]\n",
    "\n",
    "token_logits = model(**inputs).logits\n",
    "mask_token_logits = token_logits[0, mask_token_index, :]\n",
    "\n",
    "top_5_tokens = tf.math.top_k(mask_token_logits, 5).indices.numpy()\n",
    "\n",
    "for token in top_5_tokens:\n",
    "    print(sequence.replace(tokenizer.mask_token, tokenizer.decode([token])))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This prints five sequences, with the top 5 tokens predicted by the model."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Causal Language Modeling"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Causal language modeling is the task of predicting the token following a sequence of tokens. In this situation, the\n",
    "model only attends to the left context (tokens on the left of the mask). Such a training is particularly interesting\n",
    "for generation tasks. If you would like to fine-tune a model on a causal language modeling task, you may leverage the\n",
    "[run_clm.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/language-modeling/run_clm.py) script.\n",
    "\n",
    "Usually, the next token is predicted by sampling from the logits of the last hidden state the model produces from the\n",
    "input sequence.\n",
    "\n",
    "Here is an example of using the tokenizer and model and leveraging the\n",
    "[top_k_top_p_filtering()](https://huggingface.co/docs/transformers/main/en/internal/generation_utils#transformers.top_k_top_p_filtering) method to sample the next token following an input sequence\n",
    "of tokens."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Hugging Face is based in DUMBO, New York City, and ..."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoModelForCausalLM, AutoTokenizer, top_k_top_p_filtering\n",
    "import torch\n",
    "from torch import nn\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"gpt2\")\n",
    "model = AutoModelForCausalLM.from_pretrained(\"gpt2\")\n",
    "\n",
    "sequence = f\"Hugging Face is based in DUMBO, New York City, and\"\n",
    "\n",
    "inputs = tokenizer(sequence, return_tensors=\"pt\")\n",
    "input_ids = inputs[\"input_ids\"]\n",
    "\n",
    "# get logits of last hidden state\n",
    "next_token_logits = model(**inputs).logits[:, -1, :]\n",
    "\n",
    "# filter\n",
    "filtered_next_token_logits = top_k_top_p_filtering(next_token_logits, top_k=50, top_p=1.0)\n",
    "\n",
    "# sample\n",
    "probs = nn.functional.softmax(filtered_next_token_logits, dim=-1)\n",
    "next_token = torch.multinomial(probs, num_samples=1)\n",
    "\n",
    "generated = torch.cat([input_ids, next_token], dim=-1)\n",
    "\n",
    "resulting_string = tokenizer.decode(generated.tolist()[0])\n",
    "print(resulting_string)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is an example of using the tokenizer and model and leveraging the\n",
    "[tf_top_k_top_p_filtering()](https://huggingface.co/docs/transformers/main/en/internal/generation_utils#transformers.tf_top_k_top_p_filtering) method to sample the next token following an input sequence\n",
    "of tokens."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Hugging Face is based in DUMBO, New York City, and ..."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import TFAutoModelForCausalLM, AutoTokenizer, tf_top_k_top_p_filtering\n",
    "import tensorflow as tf\n",
    "\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"gpt2\")\n",
    "model = TFAutoModelForCausalLM.from_pretrained(\"gpt2\")\n",
    "\n",
    "sequence = f\"Hugging Face is based in DUMBO, New York City, and\"\n",
    "\n",
    "inputs = tokenizer(sequence, return_tensors=\"tf\")\n",
    "input_ids = inputs[\"input_ids\"]\n",
    "\n",
    "# get logits of last hidden state\n",
    "next_token_logits = model(**inputs).logits[:, -1, :]\n",
    "\n",
    "# filter\n",
    "filtered_next_token_logits = tf_top_k_top_p_filtering(next_token_logits, top_k=50, top_p=1.0)\n",
    "\n",
    "# sample\n",
    "next_token = tf.random.categorical(filtered_next_token_logits, dtype=tf.int32, num_samples=1)\n",
    "\n",
    "generated = tf.concat([input_ids, next_token], axis=1)\n",
    "\n",
    "resulting_string = tokenizer.decode(generated.numpy().tolist()[0])\n",
    "print(resulting_string)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This outputs a (hopefully) coherent next token following the original sequence, which in our case is the word *is* or\n",
    "*features*.\n",
    "\n",
    "In the next section, we show how [generation_utils.GenerationMixin.generate()](https://huggingface.co/docs/transformers/main/en/main_classes/text_generation#transformers.generation_utils.GenerationMixin.generate) can be used to\n",
    "generate multiple tokens up to a specified length instead of one token at a time."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Text Generation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "In text generation (*a.k.a* *open-ended text generation*) the goal is to create a coherent portion of text that is a\n",
    "continuation from the given context. The following example shows how *GPT-2* can be used in pipelines to generate text.\n",
    "As a default all models apply *Top-K* sampling when used in pipelines, as configured in their respective configurations\n",
    "(see [gpt-2 config](https://huggingface.co/gpt2/blob/main/config.json) for example)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'generated_text': 'As far as I am concerned, I will be the first to admit that I am not a fan of the idea of a\n",
       "\"free market.\" I think that the idea of a free market is a bit of a stretch. I think that the idea'}]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "text_generator = pipeline(\"text-generation\")\n",
    "print(text_generator(\"As far as I am concerned, I will\", max_length=50, do_sample=False))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here, the model generates a random text with a total maximal length of *50* tokens from context *\"As far as I am\n",
    "concerned, I will\"*. Behind the scenes, the pipeline object calls the method\n",
    "[PreTrainedModel.generate()](https://huggingface.co/docs/transformers/main/en/main_classes/text_generation#transformers.generation_utils.GenerationMixin.generate) to generate text. The default arguments for this method can be\n",
    "overridden in the pipeline, as is shown above for the arguments `max_length` and `do_sample`.\n",
    "\n",
    "Below is an example of text generation using `XLNet` and its tokenizer, which includes calling `generate()` directly:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Today the weather is really nice and I am planning ..."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoModelForCausalLM, AutoTokenizer\n",
    "\n",
    "model = AutoModelForCausalLM.from_pretrained(\"xlnet-base-cased\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"xlnet-base-cased\")\n",
    "\n",
    "# Padding text helps XLNet with short prompts - proposed by Aman Rusia in https://github.com/rusiaaman/XLNet-gen#methodology\n",
    "PADDING_TEXT = \"\"\"In 1991, the remains of Russian Tsar Nicholas II and his family\n",
    "(except for Alexei and Maria) are discovered.\n",
    "The voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the\n",
    "remainder of the story. 1883 Western Siberia,\n",
    "a young Grigori Rasputin is asked by his father and a group of men to perform magic.\n",
    "Rasputin has a vision and denounces one of the men as a horse thief. Although his\n",
    "father initially slaps him for making such an accusation, Rasputin watches as the\n",
    "man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n",
    "the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous,\n",
    "with people, even a bishop, begging for his blessing. <eod> </s> <eos>\"\"\"\n",
    "\n",
    "prompt = \"Today the weather is really nice and I am planning on \"\n",
    "inputs = tokenizer(PADDING_TEXT + prompt, add_special_tokens=False, return_tensors=\"pt\")[\"input_ids\"]\n",
    "\n",
    "prompt_length = len(tokenizer.decode(inputs[0]))\n",
    "outputs = model.generate(inputs, max_length=250, do_sample=True, top_p=0.95, top_k=60)\n",
    "generated = prompt + tokenizer.decode(outputs[0])[prompt_length + 1 :]\n",
    "\n",
    "print(generated)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Today the weather is really nice and I am planning ..."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import TFAutoModelForCausalLM, AutoTokenizer\n",
    "\n",
    "model = TFAutoModelForCausalLM.from_pretrained(\"xlnet-base-cased\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"xlnet-base-cased\")\n",
    "\n",
    "# Padding text helps XLNet with short prompts - proposed by Aman Rusia in https://github.com/rusiaaman/XLNet-gen#methodology\n",
    "PADDING_TEXT = \"\"\"In 1991, the remains of Russian Tsar Nicholas II and his family\n",
    "(except for Alexei and Maria) are discovered.\n",
    "The voice of Nicholas's young son, Tsarevich Alexei Nikolaevich, narrates the\n",
    "remainder of the story. 1883 Western Siberia,\n",
    "a young Grigori Rasputin is asked by his father and a group of men to perform magic.\n",
    "Rasputin has a vision and denounces one of the men as a horse thief. Although his\n",
    "father initially slaps him for making such an accusation, Rasputin watches as the\n",
    "man is chased outside and beaten. Twenty years later, Rasputin sees a vision of\n",
    "the Virgin Mary, prompting him to become a priest. Rasputin quickly becomes famous,\n",
    "with people, even a bishop, begging for his blessing. <eod> </s> <eos>\"\"\"\n",
    "\n",
    "prompt = \"Today the weather is really nice and I am planning on \"\n",
    "inputs = tokenizer(PADDING_TEXT + prompt, add_special_tokens=False, return_tensors=\"tf\")[\"input_ids\"]\n",
    "\n",
    "prompt_length = len(tokenizer.decode(inputs[0]))\n",
    "outputs = model.generate(inputs, max_length=250, do_sample=True, top_p=0.95, top_k=60)\n",
    "generated = prompt + tokenizer.decode(outputs[0])[prompt_length + 1 :]\n",
    "\n",
    "print(generated)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Text generation is currently possible with *GPT-2*, *OpenAi-GPT*, *CTRL*, *XLNet*, *Transfo-XL* and *Reformer* in\n",
    "PyTorch and for most models in Tensorflow as well. As can be seen in the example above *XLNet* and *Transfo-XL* often\n",
    "need to be padded to work well. GPT-2 is usually a good choice for *open-ended text generation* because it was trained\n",
    "on millions of webpages with a causal language modeling objective.\n",
    "\n",
    "For more information on how to apply different decoding strategies for text generation, please also refer to our text\n",
    "generation blog post [here](https://huggingface.co/blog/how-to-generate)."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Named Entity Recognition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Named Entity Recognition (NER) is the task of classifying tokens according to a class, for example, identifying a token\n",
    "as a person, an organisation or a location. An example of a named entity recognition dataset is the CoNLL-2003 dataset,\n",
    "which is entirely based on that task. If you would like to fine-tune a model on an NER task, you may leverage the\n",
    "[run_ner.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/token-classification/run_ner.py) script.\n",
    "\n",
    "Here is an example of using pipelines to do named entity recognition, specifically, trying to identify tokens as\n",
    "belonging to one of 9 classes:\n",
    "\n",
    "- O, Outside of a named entity\n",
    "- B-MIS, Beginning of a miscellaneous entity right after another miscellaneous entity\n",
    "- I-MIS, Miscellaneous entity\n",
    "- B-PER, Beginning of a person's name right after another person's name\n",
    "- I-PER, Person's name\n",
    "- B-ORG, Beginning of an organisation right after another organisation\n",
    "- I-ORG, Organisation\n",
    "- B-LOC, Beginning of a location right after another location\n",
    "- I-LOC, Location\n",
    "\n",
    "It leverages a fine-tuned model on CoNLL-2003, fine-tuned by [@stefan-it](https://github.com/stefan-it) from [dbmdz](https://github.com/dbmdz)."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "ner_pipe = pipeline(\"ner\")\n",
    "\n",
    "sequence = \"\"\"Hugging Face Inc. is a company based in New York City. Its headquarters are in DUMBO,\n",
    "therefore very close to the Manhattan Bridge which is visible from the window.\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This outputs a list of all words that have been identified as one of the entities from the 9 classes defined above.\n",
    "Here are the expected results:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'entity': 'I-ORG', 'score': 0.9996, 'index': 1, 'word': 'Hu', 'start': 0, 'end': 2}\n",
       "{'entity': 'I-ORG', 'score': 0.9910, 'index': 2, 'word': '##gging', 'start': 2, 'end': 7}\n",
       "{'entity': 'I-ORG', 'score': 0.9982, 'index': 3, 'word': 'Face', 'start': 8, 'end': 12}\n",
       "{'entity': 'I-ORG', 'score': 0.9995, 'index': 4, 'word': 'Inc', 'start': 13, 'end': 16}\n",
       "{'entity': 'I-LOC', 'score': 0.9994, 'index': 11, 'word': 'New', 'start': 40, 'end': 43}\n",
       "{'entity': 'I-LOC', 'score': 0.9993, 'index': 12, 'word': 'York', 'start': 44, 'end': 48}\n",
       "{'entity': 'I-LOC', 'score': 0.9994, 'index': 13, 'word': 'City', 'start': 49, 'end': 53}\n",
       "{'entity': 'I-LOC', 'score': 0.9863, 'index': 19, 'word': 'D', 'start': 79, 'end': 80}\n",
       "{'entity': 'I-LOC', 'score': 0.9514, 'index': 20, 'word': '##UM', 'start': 80, 'end': 82}\n",
       "{'entity': 'I-LOC', 'score': 0.9337, 'index': 21, 'word': '##BO', 'start': 82, 'end': 84}\n",
       "{'entity': 'I-LOC', 'score': 0.9762, 'index': 28, 'word': 'Manhattan', 'start': 114, 'end': 123}\n",
       "{'entity': 'I-LOC', 'score': 0.9915, 'index': 29, 'word': 'Bridge', 'start': 124, 'end': 130}"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for entity in ner_pipe(sequence):\n",
    "    print(entity)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Note how the tokens of the sequence \"Hugging Face\" have been identified as an organisation, and \"New York City\",\n",
    "\"DUMBO\" and \"Manhattan Bridge\" have been identified as locations.\n",
    "\n",
    "Here is an example of doing named entity recognition, using a model and a tokenizer. The process is the following:\n",
    "\n",
    "1. Instantiate a tokenizer and a model from the checkpoint name. The model is identified as a BERT model and loads it\n",
    "   with the weights stored in the checkpoint.\n",
    "2. Define a sequence with known entities, such as \"Hugging Face\" as an organisation and \"New York City\" as a location.\n",
    "3. Split words into tokens so that they can be mapped to predictions. We use a small hack by, first, completely\n",
    "   encoding and decoding the sequence, so that we're left with a string that contains the special tokens.\n",
    "4. Encode that sequence into IDs (special tokens are added automatically).\n",
    "5. Retrieve the predictions by passing the input to the model and getting the first output. This results in a\n",
    "   distribution over the 9 possible classes for each token. We take the argmax to retrieve the most likely class for\n",
    "   each token.\n",
    "6. Zip together each token with its prediction and print it."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import AutoModelForTokenClassification, AutoTokenizer\n",
    "import torch\n",
    "\n",
    "model = AutoModelForTokenClassification.from_pretrained(\"dbmdz/bert-large-cased-finetuned-conll03-english\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"bert-base-cased\")\n",
    "\n",
    "sequence = (\n",
    "    \"Hugging Face Inc. is a company based in New York City. Its headquarters are in DUMBO, \"\n",
    "    \"therefore very close to the Manhattan Bridge.\"\n",
    ")\n",
    "\n",
    "inputs = tokenizer(sequence, return_tensors=\"pt\")\n",
    "tokens = inputs.tokens()\n",
    "\n",
    "outputs = model(**inputs).logits\n",
    "predictions = torch.argmax(outputs, dim=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import TFAutoModelForTokenClassification, AutoTokenizer\n",
    "import tensorflow as tf\n",
    "\n",
    "model = TFAutoModelForTokenClassification.from_pretrained(\"dbmdz/bert-large-cased-finetuned-conll03-english\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"bert-base-cased\")\n",
    "\n",
    "sequence = (\n",
    "    \"Hugging Face Inc. is a company based in New York City. Its headquarters are in DUMBO, \"\n",
    "    \"therefore very close to the Manhattan Bridge.\"\n",
    ")\n",
    "\n",
    "inputs = tokenizer(sequence, return_tensors=\"tf\")\n",
    "tokens = inputs.tokens()\n",
    "\n",
    "outputs = model(**inputs)[0]\n",
    "predictions = tf.argmax(outputs, axis=2)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "This outputs a list of each token mapped to its corresponding prediction. Differently from the pipeline, here every\n",
    "token has a prediction as we didn't remove the \"0\"th class, which means that no particular entity was found on that\n",
    "token.\n",
    "\n",
    "In the above example, `predictions` is an integer that corresponds to the predicted class. We can use the\n",
    "`model.config.id2label` property in order to recover the class name corresponding to the class number, which is\n",
    "illustrated below:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "('[CLS]', 'O')\n",
       "('Hu', 'I-ORG')\n",
       "('##gging', 'I-ORG')\n",
       "('Face', 'I-ORG')\n",
       "('Inc', 'I-ORG')\n",
       "('.', 'O')\n",
       "('is', 'O')\n",
       "('a', 'O')\n",
       "('company', 'O')\n",
       "('based', 'O')\n",
       "('in', 'O')\n",
       "('New', 'I-LOC')\n",
       "('York', 'I-LOC')\n",
       "('City', 'I-LOC')\n",
       "('.', 'O')\n",
       "('Its', 'O')\n",
       "('headquarters', 'O')\n",
       "('are', 'O')\n",
       "('in', 'O')\n",
       "('D', 'I-LOC')\n",
       "('##UM', 'I-LOC')\n",
       "('##BO', 'I-LOC')\n",
       "(',', 'O')\n",
       "('therefore', 'O')\n",
       "('very', 'O')\n",
       "('close', 'O')\n",
       "('to', 'O')\n",
       "('the', 'O')\n",
       "('Manhattan', 'I-LOC')\n",
       "('Bridge', 'I-LOC')\n",
       "('.', 'O')\n",
       "('[SEP]', 'O')"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "for token, prediction in zip(tokens, predictions[0].numpy()):\n",
    "    print((token, model.config.id2label[prediction]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Summarization"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Summarization is the task of summarizing a document or an article into a shorter text. If you would like to fine-tune a\n",
    "model on a summarization task, you may leverage the [run_summarization.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/summarization/run_summarization.py)\n",
    "script.\n",
    "\n",
    "An example of a summarization dataset is the CNN / Daily Mail dataset, which consists of long news articles and was\n",
    "created for the task of summarization. If you would like to fine-tune a model on a summarization task, various\n",
    "approaches are described in this [document](https://github.com/huggingface/transformers/tree/main/examples/pytorch/summarization/README.md).\n",
    "\n",
    "Here is an example of using the pipelines to do summarization. It leverages a Bart model that was fine-tuned on the CNN\n",
    "/ Daily Mail data set."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "summarizer = pipeline(\"summarization\")\n",
    "\n",
    "ARTICLE = \"\"\" New York (CNN)When Liana Barrientos was 23 years old, she got married in Westchester County, New York.\n",
    "A year later, she got married again in Westchester County, but to a different man and without divorcing her first husband.\n",
    "Only 18 days after that marriage, she got hitched yet again. Then, Barrientos declared \"I do\" five more times, sometimes only within two weeks of each other.\n",
    "In 2010, she married once more, this time in the Bronx. In an application for a marriage license, she stated it was her \"first and only\" marriage.\n",
    "Barrientos, now 39, is facing two criminal counts of \"offering a false instrument for filing in the first degree,\" referring to her false statements on the\n",
    "2010 marriage license application, according to court documents.\n",
    "Prosecutors said the marriages were part of an immigration scam.\n",
    "On Friday, she pleaded not guilty at State Supreme Court in the Bronx, according to her attorney, Christopher Wright, who declined to comment further.\n",
    "After leaving court, Barrientos was arrested and charged with theft of service and criminal trespass for allegedly sneaking into the New York subway through an emergency exit, said Detective\n",
    "Annette Markowski, a police spokeswoman. In total, Barrientos has been married 10 times, with nine of her marriages occurring between 1999 and 2002.\n",
    "All occurred either in Westchester County, Long Island, New Jersey or the Bronx. She is believed to still be married to four men, and at one time, she was married to eight men at once, prosecutors say.\n",
    "Prosecutors said the immigration scam involved some of her husbands, who filed for permanent residence status shortly after the marriages.\n",
    "Any divorces happened only after such filings were approved. It was unclear whether any of the men will be prosecuted.\n",
    "The case was referred to the Bronx District Attorney\\'s Office by Immigration and Customs Enforcement and the Department of Homeland Security\\'s\n",
    "Investigation Division. Seven of the men are from so-called \"red-flagged\" countries, including Egypt, Turkey, Georgia, Pakistan and Mali.\n",
    "Her eighth husband, Rashid Rajput, was deported in 2006 to his native Pakistan after an investigation by the Joint Terrorism Task Force.\n",
    "If convicted, Barrientos faces up to four years in prison.  Her next court appearance is scheduled for May 18.\n",
    "\"\"\""
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because the summarization pipeline depends on the `PreTrainedModel.generate()` method, we can override the default\n",
    "arguments of `PreTrainedModel.generate()` directly in the pipeline for `max_length` and `min_length` as shown\n",
    "below. This outputs the following summary:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'summary_text': ' Liana Barrientos, 39, is charged with two counts of \"offering a false instrument for filing in\n",
       "the first degree\" In total, she has been married 10 times, with nine of her marriages occurring between 1999 and\n",
       "2002 . At one time, she was married to eight men at once, prosecutors say .'}]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(summarizer(ARTICLE, max_length=130, min_length=30, do_sample=False))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Here is an example of doing summarization using a model and a tokenizer. The process is the following:\n",
    "\n",
    "1. Instantiate a tokenizer and a model from the checkpoint name. Summarization is usually done using an encoder-decoder\n",
    "   model, such as `Bart` or `T5`.\n",
    "2. Define the article that should be summarized.\n",
    "3. Add the T5 specific prefix \"summarize: \".\n",
    "4. Use the `PreTrainedModel.generate()` method to generate the summary.\n",
    "\n",
    "In this example we use Google's T5 model. Even though it was pre-trained only on a multi-task mixed dataset (including\n",
    "CNN / Daily Mail), it yields very good results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "prosecutors say the marriages were part of an immigration scam. if convicted, barrientos faces two criminal\n",
       "counts of \"offering a false instrument for filing in the first degree\" she has been married 10 times, nine of them\n",
       "between 1999 and 2002."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoModelForSeq2SeqLM, AutoTokenizer\n",
    "\n",
    "model = AutoModelForSeq2SeqLM.from_pretrained(\"t5-base\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"t5-base\")\n",
    "\n",
    "# T5 uses a max_length of 512 so we cut the article to 512 tokens.\n",
    "inputs = tokenizer(\"summarize: \" + ARTICLE, return_tensors=\"pt\", max_length=512, truncation=True)\n",
    "outputs = model.generate(\n",
    "    inputs[\"input_ids\"], max_length=150, min_length=40, length_penalty=2.0, num_beams=4, early_stopping=True\n",
    ")\n",
    "\n",
    "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "prosecutors say the marriages were part of an immigration scam. if convicted, barrientos faces two criminal\n",
       "counts of \"offering a false instrument for filing in the first degree\" she has been married 10 times, nine of them\n",
       "between 1999 and 2002."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import TFAutoModelForSeq2SeqLM, AutoTokenizer\n",
    "\n",
    "model = TFAutoModelForSeq2SeqLM.from_pretrained(\"t5-base\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"t5-base\")\n",
    "\n",
    "# T5 uses a max_length of 512 so we cut the article to 512 tokens.\n",
    "inputs = tokenizer(\"summarize: \" + ARTICLE, return_tensors=\"tf\", max_length=512)\n",
    "outputs = model.generate(\n",
    "    inputs[\"input_ids\"], max_length=150, min_length=40, length_penalty=2.0, num_beams=4, early_stopping=True\n",
    ")\n",
    "\n",
    "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Translation"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Translation is the task of translating a text from one language to another. If you would like to fine-tune a model on a\n",
    "translation task, you may leverage the [run_translation.py](https://github.com/huggingface/transformers/tree/main/examples/pytorch/translation/run_translation.py) script.\n",
    "\n",
    "An example of a translation dataset is the WMT English to German dataset, which has sentences in English as the input\n",
    "data and the corresponding sentences in German as the target data. If you would like to fine-tune a model on a\n",
    "translation task, various approaches are described in this [document](https://github.com/huggingface/transformers/tree/main/examples/pytorch/translation/README.md).\n",
    "\n",
    "Here is an example of using the pipelines to do translation. It leverages a T5 model that was only pre-trained on a\n",
    "multi-task mixture dataset (including WMT), yet, yielding impressive translation results."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'translation_text': 'Hugging Face ist ein Technologieunternehmen mit Sitz in New York und Paris.'}]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "translator = pipeline(\"translation_en_to_de\")\n",
    "print(translator(\"Hugging Face is a technology company based in New York and Paris\", max_length=40))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Because the translation pipeline depends on the `PreTrainedModel.generate()` method, we can override the default\n",
    "arguments of `PreTrainedModel.generate()` directly in the pipeline as is shown for `max_length` above.\n",
    "\n",
    "Here is an example of doing translation using a model and a tokenizer. The process is the following:\n",
    "\n",
    "1. Instantiate a tokenizer and a model from the checkpoint name. Summarization is usually done using an encoder-decoder\n",
    "   model, such as `Bart` or `T5`.\n",
    "2. Define the article that should be summarized.\n",
    "3. Add the T5 specific prefix \"translate English to German: \"\n",
    "4. Use the `PreTrainedModel.generate()` method to perform the translation."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Hugging Face ist ein Technologieunternehmen mit Sitz in New York und Paris."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoModelForSeq2SeqLM, AutoTokenizer\n",
    "\n",
    "model = AutoModelForSeq2SeqLM.from_pretrained(\"t5-base\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"t5-base\")\n",
    "\n",
    "inputs = tokenizer(\n",
    "    \"translate English to German: Hugging Face is a technology company based in New York and Paris\",\n",
    "    return_tensors=\"pt\",\n",
    ")\n",
    "outputs = model.generate(inputs[\"input_ids\"], max_length=40, num_beams=4, early_stopping=True)\n",
    "\n",
    "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Hugging Face ist ein Technologieunternehmen mit Sitz in New York und Paris."
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import TFAutoModelForSeq2SeqLM, AutoTokenizer\n",
    "\n",
    "model = TFAutoModelForSeq2SeqLM.from_pretrained(\"t5-base\")\n",
    "tokenizer = AutoTokenizer.from_pretrained(\"t5-base\")\n",
    "\n",
    "inputs = tokenizer(\n",
    "    \"translate English to German: Hugging Face is a technology company based in New York and Paris\",\n",
    "    return_tensors=\"tf\",\n",
    ")\n",
    "outputs = model.generate(inputs[\"input_ids\"], max_length=40, num_beams=4, early_stopping=True)\n",
    "\n",
    "print(tokenizer.decode(outputs[0], skip_special_tokens=True))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We get the same translation as with the pipeline example."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Audio classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Audio classification assigns a class to an audio signal. The Keyword Spotting dataset from the [SUPERB](https://huggingface.co/datasets/superb) benchmark is an example dataset that can be used for audio classification fine-tuning. This dataset contains ten classes of keywords for classification. If you'd like to fine-tune a model for audio classification, take a look at the [run_audio_classification.py](https://github.com/huggingface/transformers/blob/main/examples/pytorch/audio-classification/run_audio_classification.py) script or this [how-to guide](https://huggingface.co/docs/transformers/main/en/./tasks/audio_classification).\n",
    "\n",
    "The following examples demonstrate how to use a [pipeline()](https://huggingface.co/docs/transformers/main/en/main_classes/pipelines#transformers.pipeline) and a model and tokenizer for audio classification inference:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[{'score': 0.1315, 'label': 'calm'}, {'score': 0.1307, 'label': 'neutral'}, {'score': 0.1274, 'label': 'sad'}, {'score': 0.1261, 'label': 'fearful'}, {'score': 0.1242, 'label': 'happy'}]"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "from datasets import load_dataset\n",
    "import torch\n",
    "\n",
    "torch.manual_seed(42)\n",
    "dataset = load_dataset(\"hf-internal-testing/librispeech_asr_demo\", \"clean\", split=\"validation\")\n",
    "dataset = dataset.sort(\"id\")\n",
    "audio_file = dataset[0][\"audio\"][\"path\"]\n",
    "\n",
    "audio_classifier = pipeline(\n",
    "    task=\"audio-classification\", model=\"ehcalabres/wav2vec2-lg-xlsr-en-speech-emotion-recognition\"\n",
    ")\n",
    "predictions = audio_classifier(audio_file)\n",
    "predictions = [{\"score\": round(pred[\"score\"], 4), \"label\": pred[\"label\"]} for pred in predictions]\n",
    "predictions"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The general process for using a model and feature extractor for audio classification is:\n",
    "\n",
    "1. Instantiate a feature extractor and a model from the checkpoint name.\n",
    "2. Process the audio signal to be classified with a feature extractor.\n",
    "3. Pass the input through the model and take the `argmax` to retrieve the most likely class.\n",
    "4. Convert the class id to a class name with `id2label` to return an interpretable result."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'_unknown_'"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoFeatureExtractor, AutoModelForAudioClassification\n",
    "from datasets import load_dataset\n",
    "import torch\n",
    "\n",
    "dataset = load_dataset(\"hf-internal-testing/librispeech_asr_demo\", \"clean\", split=\"validation\")\n",
    "dataset = dataset.sort(\"id\")\n",
    "sampling_rate = dataset.features[\"audio\"].sampling_rate\n",
    "\n",
    "feature_extractor = AutoFeatureExtractor.from_pretrained(\"superb/wav2vec2-base-superb-ks\")\n",
    "model = AutoModelForAudioClassification.from_pretrained(\"superb/wav2vec2-base-superb-ks\")\n",
    "\n",
    "inputs = feature_extractor(dataset[0][\"audio\"][\"array\"], sampling_rate=sampling_rate, return_tensors=\"pt\")\n",
    "\n",
    "with torch.no_grad():\n",
    "    logits = model(**inputs).logits\n",
    "\n",
    "predicted_class_ids = torch.argmax(logits, dim=-1).item()\n",
    "predicted_label = model.config.id2label[predicted_class_ids]\n",
    "predicted_label"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Automatic speech recognition"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Automatic speech recognition transcribes an audio signal to text. The [Common Voice](https://huggingface.co/datasets/common_voice) dataset is an example dataset that can be used for automatic speech recognition fine-tuning. It contains an audio file of a speaker and the corresponding sentence. If you'd like to fine-tune a model for automatic speech recognition, take a look at the [run_speech_recognition_ctc.py](https://github.com/huggingface/transformers/blob/main/examples/pytorch/speech-recognition/run_speech_recognition_ctc.py) or [run_speech_recognition_seq2seq.py](https://github.com/huggingface/transformers/blob/main/examples/pytorch/speech-recognition/run_speech_recognition_seq2seq.py) scripts or this [how-to guide](https://huggingface.co/docs/transformers/main/en/./tasks/asr).\n",
    "\n",
    "The following examples demonstrate how to use a [pipeline()](https://huggingface.co/docs/transformers/main/en/main_classes/pipelines#transformers.pipeline) and a model and tokenizer for automatic speech recognition inference:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'text': 'MISTER QUILTER IS THE APOSTLE OF THE MIDDLE CLASSES AND WE ARE GLAD TO WELCOME HIS GOSPEL'}"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "from datasets import load_dataset\n",
    "\n",
    "dataset = load_dataset(\"hf-internal-testing/librispeech_asr_demo\", \"clean\", split=\"validation\")\n",
    "dataset = dataset.sort(\"id\")\n",
    "audio_file = dataset[0][\"audio\"][\"path\"]\n",
    "\n",
    "speech_recognizer = pipeline(task=\"automatic-speech-recognition\", model=\"facebook/wav2vec2-base-960h\")\n",
    "speech_recognizer(audio_file)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The general process for using a model and processor for automatic speech recognition is:\n",
    "\n",
    "1. Instantiate a processor (which regroups a feature extractor for input processing and a tokenizer for decoding) and a model from the checkpoint name.\n",
    "2. Process the audio signal and text with a processor.\n",
    "3. Pass the input through the model and take the `argmax` to retrieve the predicted text.\n",
    "4. Decode the text with a tokenizer to obtain the transcription."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'MISTER QUILTER IS THE APOSTLE OF THE MIDDLE CLASSES AND WE ARE GLAD TO WELCOME HIS GOSPEL'"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoProcessor, AutoModelForCTC\n",
    "from datasets import load_dataset\n",
    "import torch\n",
    "\n",
    "dataset = load_dataset(\"hf-internal-testing/librispeech_asr_demo\", \"clean\", split=\"validation\")\n",
    "dataset = dataset.sort(\"id\")\n",
    "sampling_rate = dataset.features[\"audio\"].sampling_rate\n",
    "\n",
    "processor = AutoProcessor.from_pretrained(\"facebook/wav2vec2-base-960h\")\n",
    "model = AutoModelForCTC.from_pretrained(\"facebook/wav2vec2-base-960h\")\n",
    "\n",
    "inputs = processor(dataset[0][\"audio\"][\"array\"], sampling_rate=sampling_rate, return_tensors=\"pt\")\n",
    "with torch.no_grad():\n",
    "    logits = model(**inputs).logits\n",
    "predicted_ids = torch.argmax(logits, dim=-1)\n",
    "\n",
    "transcription = processor.batch_decode(predicted_ids)\n",
    "transcription[0]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Image classification"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Like text and audio classification, image classification assigns a class to an image. The [CIFAR-100](https://huggingface.co/datasets/cifar100) dataset is an example dataset that can be used for image classification fine-tuning. It contains an image and the corresponding class. If you'd like to fine-tune a model for image classification, take a look at the [run_image_classification.py](https://github.com/huggingface/transformers/blob/main/examples/pytorch/image-classification/run_image_classification.py) script or this [how-to guide](https://huggingface.co/docs/transformers/main/en/./tasks/image_classification).\n",
    "\n",
    "The following examples demonstrate how to use a [pipeline()](https://huggingface.co/docs/transformers/main/en/main_classes/pipelines#transformers.pipeline) and a model and tokenizer for image classification inference:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Class lynx, catamount with score 0.4335                                                    \n",
       "Class cougar, puma, catamount, mountain lion, painter, panther, Felis concolor with score 0.0348\n",
       "Class snow leopard, ounce, Panthera uncia with score 0.0324          \n",
       "Class Egyptian cat with score 0.0239                                                       \n",
       "Class tiger cat with score 0.0229"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import pipeline\n",
    "\n",
    "vision_classifier = pipeline(task=\"image-classification\")\n",
    "result = vision_classifier(\n",
    "    images=\"https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/pipeline-cat-chonk.jpeg\"\n",
    ")\n",
    "print(\"\\n\".join([f\"Class {d['label']} with score {round(d['score'], 4)}\" for d in result]))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "The general process for using a model and feature extractor for image classification is:\n",
    "\n",
    "1. Instantiate a feature extractor and a model from the checkpoint name.\n",
    "2. Process the image to be classified with a feature extractor.\n",
    "3. Pass the input through the model and take the `argmax` to retrieve the predicted class.\n",
    "4. Convert the class id to a class name with `id2label` to return an interpretable result."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Egyptian cat"
      ]
     },
     "execution_count": null,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from transformers import AutoFeatureExtractor, AutoModelForImageClassification\n",
    "import torch\n",
    "from datasets import load_dataset\n",
    "\n",
    "dataset = load_dataset(\"huggingface/cats-image\")\n",
    "image = dataset[\"test\"][\"image\"][0]\n",
    "\n",
    "feature_extractor = AutoFeatureExtractor.from_pretrained(\"google/vit-base-patch16-224\")\n",
    "model = AutoModelForImageClassification.from_pretrained(\"google/vit-base-patch16-224\")\n",
    "\n",
    "inputs = feature_extractor(image, return_tensors=\"pt\")\n",
    "\n",
    "with torch.no_grad():\n",
    "    logits = model(**inputs).logits\n",
    "\n",
    "predicted_label = logits.argmax(-1).item()\n",
    "print(model.config.id2label[predicted_label])"
   ]
  }
 ],
 "metadata": {},
 "nbformat": 4,
 "nbformat_minor": 4
}