Send a Text Classification Request

Use the /classify endpoint to classify one or more text inputs.

curl -X POST \
  --header 'Content-Type: application/json' \
  --header 'Accept: application/json' \
  --header "Authorization: Bearer $API_KEY" \
  --cacert "$SECRETS_DIR/cert.pem" \
  --data '{
    "model": "twitter-roberta-base-sentiment-latest",
    "input": [
      "I loved the service.",
      "The experience was frustrating."
    ]
  }' \
  https://localhost:9091/classify

The request body includes the following parameters:

• model: The name of the text classification model to use.
• input: A string or an array of strings to classify.

The response includes a classification result for each input string. Each result includes the input index, predicted label, class probabilities, and number of classes.

{
  "id": "classify-814098a1f6521d7e",
  "model": "twitter-roberta-base-sentiment-latest",
  "data": [
    {
      "index": 0,
      "label": "positive",
      "probs": [
        0.005621818359941244,
        0.02003072202205658,
        0.9743474125862122
      ],
      "num_classes": 3
    },
    {
      "index": 1,
      "label": "negative",
      "probs": [
        0.9102542400360107,
        0.08134238421916962,
        0.008403426967561245
      ],
      "num_classes": 3
    }
  ],
  "usage": {
    "prompt_tokens": 14,
    "total_tokens": 14,
    "completion_tokens": 0
  }
}

The response includes the following fields:

• id: A unique identifier for the classification request.
• model: The name of the model used for classification.
• data: An array of classification results.
• index: The position of the input text in the request.
• label: The predicted class label.
• probs: The probability for the model's classes.
• num_classes: The number of classes returned by the model.
• usage: Token usage information for the request.