# Generated by API Evangelist (build-phrasing.py). Our phrasing, not observed demand.
overlay: 1.0.0
info:
  title: API Evangelist conversational phrasing for Hugging Face Inference Natural Language Processing API
  version: 1.0.0
extends: openapi/hugging-face-natural-language-processing-api-openapi.yml
actions:
- target: $.info
  update:
    x-apievangelist-phrasing:
      method: generated
      generated: '2026-10-01'
      generator: build-phrasing.py
      label: Generated by API Evangelist
      operations: 11
- target: $.paths['/models/{model_id}'].post
  update:
    x-apievangelist-phrasing:
      intent: Run inference on any Hub model
      effect: write
      questions:
      - Can I call any model hosted on the Hub with a generic inference request?
      - Does the input and output format change depending on a model's pipeline task?
      instructions:
      - text: Run inference on model {model_id} with input {inputs}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Send {inputs} to {model_id} with task parameters {parameters}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
          parameters: requestBody.parameters
      method: generated
      generated: '2026-10-01'
- target: $.paths['/pipeline/{task}'].post
  update:
    x-apievangelist-phrasing:
      intent: Run inference by task with a recommended model
      effect: write
      questions:
      - Can I name just a task like summarization and let Hugging Face pick the best model?
      - Which pipeline tasks can I run without choosing a model myself?
      instructions:
      - text: Run the {task} pipeline on {inputs}.
        slots:
          task: path.task
          inputs: requestBody.inputs
      - text: Use the recommended model for {task} to process {inputs}.
        slots:
          task: path.task
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/text-generation'].post
  update:
    x-apievangelist-phrasing:
      intent: Continue a prompt with a language model
      effect: write
      questions:
      - Can I generate text from a prompt with a specific Hub model and control temperature and top_p?
      - Is there a way to cap max new tokens or penalize repetition in serverless text generation?
      instructions:
      - text: Continue the prompt {inputs} with model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Generate text from {inputs} on {model_id} using parameters {parameters}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
          parameters: requestBody.parameters
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/text-classification'].post
  update:
    x-apievangelist-phrasing:
      intent: Classify text, such as sentiment
      effect: write
      questions:
      - Can I tell whether a review is positive or negative with a sentiment model?
      - What labels and scores does text classification return?
      instructions:
      - text: Classify the sentiment of {inputs} with model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Label {inputs} using text classifier {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/question-answering'].post
  update:
    x-apievangelist-phrasing:
      intent: Extract an answer from a context passage
      effect: write
      questions:
      - Can a model pull the answer to a question out of a passage I provide?
      - Which task finds an answer span inside my own document text?
      instructions:
      - text: Answer the question in {inputs} from its context using model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Extract the answer span for {inputs} with QA model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/summarization'].post
  update:
    x-apievangelist-phrasing:
      intent: Summarize a long text
      effect: write
      questions:
      - Can I shorten a long article into a brief summary with a Hub model?
      - Is there a way to control summary length through task parameters?
      instructions:
      - text: Summarize {inputs} with model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Write a shorter version of {inputs} using {model_id} and parameters {parameters}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
          parameters: requestBody.parameters
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/translation'].post
  update:
    x-apievangelist-phrasing:
      intent: Translate text between languages
      effect: write
      questions:
      - Can I translate a sentence into another language with a translation model?
      - Which serverless task should I use to translate text with a model I pick?
      instructions:
      - text: Translate {inputs} with model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Run translation model {model_id} on {inputs}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/fill-mask'].post
  update:
    x-apievangelist-phrasing:
      intent: Fill a masked word in a sentence
      effect: write
      questions:
      - Can a model guess the missing word where I put a [MASK] token?
      - What candidates and scores come back for a masked token?
      instructions:
      - text: Fill the [MASK] in {inputs} using model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Predict the masked word in {inputs} with {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/feature-extraction'].post
  update:
    x-apievangelist-phrasing:
      intent: Extract embeddings with a chosen Hub model
      effect: write
      questions:
      - Can I get raw feature vectors from a specific Hub model by calling its feature-extraction task?
      - Which serverless model task returns dense vector representations for text?
      instructions:
      - text: Extract features for {inputs} using model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Get the feature-extraction vectors of {inputs} from {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/zero-shot-classification'].post
  update:
    x-apievangelist-phrasing:
      intent: Classify text into labels you supply
      effect: write
      questions:
      - Can I classify text into my own categories without training a model?
      - Where do I pass the candidate labels for zero-shot classification?
      instructions:
      - text: Classify {inputs} into the candidate labels in {parameters} with model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
          parameters: requestBody.parameters
      - text: Run zero-shot classification of {inputs} on {model_id} using labels {parameters}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
          parameters: requestBody.parameters
      method: generated
      generated: '2026-10-01'
- target: $.paths['/models/{model_id}/sentence-similarity'].post
  update:
    x-apievangelist-phrasing:
      intent: Score similarity between sentences
      effect: write
      questions:
      - Can I measure how similar one sentence is to a list of other sentences?
      - Which task returns similarity scores between a source sentence and targets?
      instructions:
      - text: Compare the source and target sentences in {inputs} with model {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      - text: Score sentence similarity for {inputs} using {model_id}.
        slots:
          model_id: path.model_id
          inputs: requestBody.inputs
      method: generated
      generated: '2026-10-01'