Skip to main content

cohere.rerank

Conduit processor for Cohere's rerank model.

Description​

Conduit processor for Cohere's rerank model.

Configuration parameters​

version: 2.2
pipelines:
  - id: example
    status: running
    connectors:
      # define source and destination ...
    processors:
      - id: example
        plugin: "cohere.rerank"
        settings:
          # APIKey is the API key for Cohere api calls.
          # Type: string
          apiKey: ""
          # Maximum number of retries for an individual record when backing off
          # following an error.
          # Type: float
          backoffRetry.count: "0"
          # The multiplying factor for each increment step.
          # Type: float
          backoffRetry.factor: "2"
          # The maximum waiting time before retrying.
          # Type: duration
          backoffRetry.max: "5s"
          # The minimum waiting time before retrying.
          # Type: duration
          backoffRetry.min: "100ms"
          # Model is one of the name of a compatible rerank model version.
          # Type: string
          model: "rerank-v3.5"
          # Query is the search query.
          # Type: string
          query: ""
          # RequestBodyRef specifies the api request field.
          # Type: string
          request.body: ".Payload.After"
          # ResponseBodyRef specifies in which field should the response body be
          # saved.
          # Type: string
          response.body: ".Payload.After"
          # Whether to decode the record key using its corresponding schema from
          # the schema registry.
          # Type: bool
          sdk.schema.decode.key.enabled: "true"
          # Whether to decode the record payload using its corresponding schema
          # from the schema registry.
          # Type: bool
          sdk.schema.decode.payload.enabled: "true"
          # Whether to encode the record key using its corresponding schema from
          # the schema registry.
          # Type: bool
          sdk.schema.encode.key.enabled: "true"
          # Whether to encode the record payload using its corresponding schema
          # from the schema registry.
          # Type: bool
          sdk.schema.encode.payload.enabled: "true"

Examples​

Generate responses using Cohere's rerank model​

This example demonstrates how to use the Cohere rerank processor.This takes in a query and a list of texts and produces an ordered array with each text assigned a relevance score. The processor extracts text from the specified input field (default: ".Payload.After"), sends it to the Cohere API, and stores the response in the record's ".Payload.After" field.

In this example, the processor is configured with a mock client and an API key. The input record's metadata is updated to include the rerank model used ("rerank-v3.5").

Configuration parameters​

version: 2.2
pipelines:
  - id: example
    status: running
    connectors:
      # define source and destination ...
    processors:
      - id: example
        plugin: "cohere.rerank"
        settings:
          apiKey: "fakeapiKey"
          backoffRetry.count: "0"
          backoffRetry.factor: "2"
          backoffRetry.max: "5s"
          backoffRetry.min: "100ms"
          model: "rerank-v3.5"
          query: "What is the capital of the United States?"
          request.body: ".Payload.After"
          response.body: ".Payload.After"

Record difference​

Before
After
1
{
1
{
2
  "position": "cG9zLTE=",
2
  "position": "cG9zLTE=",
3
  "operation": "update",
3
  "operation": "update",
4
  "metadata": null,
4
  "metadata": null,
5
  "key": null,
5
  "key": null,
6
  "payload": {
6
  "payload": {
7
    "before": null,
7
    "before": null,
8
-
    "after": "Carson City is the capital city of the American state of Nevada."
8
+
    "after": "{\"document\":{\"text\":\"Carson City is the capital city of the American state of Nevada.\"},\"index\":0,\"relevance_score\":0.9}"
9
  }
9
  }
10
}
10
}


scarf pixel conduit-site-docs-using-processors