Skip to content

titan_embedings

Since Docker image version 0.0.18

Using this Provider, you can use the Titan text embeddings model to output embeddings (float arrays) for text. The model configuration can be referenced in this document: Amazon Titan Embeddings Text

Parameter Configuration

Key Type Required Default value Description
modelId string Y Model id, amazon.titan-embed-text-v2:0 or amazon.titan-embed-text-v1
regions string[] or string N Same as the AWS_DEFAULT_REGION environment variable If you have applied for and specified multiple regions, one region will be randomly selected for the call. This feature can effectively alleviate performance bottlenecks.
dimensions number N 1024 Output dimension, only applicable to the amazon.titan-embed-text-v2:0 model.
normalize boolean N true Flag indicating whether or not to normalize the output embeddings, only applicable to the amazon.titan-embed-text-v2:0 model

An example configuration for titan_embeddings is as follows:

{
  "modelId": "amazon.titan-embed-text-v2:0",
  "regions": [
    "us-east-1",
    "us-west-2"
  ]
}