TextEmbeddingPredictionInstance

Prediction input format for Text Embedding. An embedding is a numerical representation (a vector of floating-point numbers) of text that captures its semantic meaning, enabling tasks like semantic search, classification, and clustering. LINT.IfChange

Fields
content string

The text passage to generate the embedding for. This can be a phrase, sentence, or document.

title string

The title for the text passage in the content field. If you are embedding documents for the RETRIEVAL_DOCUMENT task type, providing a title can improve the quality of the embedding.

taskType enum (TaskType)

The task that the generated embeddings will be used for. The model uses this hint to generate embeddings that are optimized for your use case. If not set, the default task type is used.

JSON representation 
{
  "content": string,
  "title": string,
  "taskType": enum (TaskType)
}