Skip to main content

Cohere

Introduction​

Cohere Inc. is a Canadian multinational technology firm that concentrates on enterprise artificial intelligence, specializing particularly in large-scale language models.

Deploying your model​

Cohere offers the capability to deploy various Chat Models(LLM) under its umbrella.

Chat Model​

Refer to their documentation to understand the latest offerings, with feature and cost comparisons.

Usage with MAAP​

To use the Cohere chat model with the MAAP framework, you would need to feed the below values.

  • Config File :​

    Provided below are the values required to be added in the config.yaml file in the LLM section.

    llms:
        class_name: Cohere
        model_name: <check_references_below>
        temperature: <integer_value>

  • Environment Variable :​

    The below value(s) are to be added in the .env file, present at builder/partnerproduct/.

    COHERE_API_KEY : <check_references_below>

Embedding Model​

Refer to their documentation to understand the latest offerings, with feature and cost comparisons.

Usage with MAAP​

To use the Cohere embedding model with the MAAP framework, you would need to feed the below values.

  • Config File :​

    Provided below are the values required to be added in the config.yaml file located in the embedding section.

    embedding:
        class_name: Cohere
        model_name: <check_references_below>

  • Environment Variable :​

    The below value(s) are to be added in the .env file, present at builder/partnerproduct/.

    COHERE_API_KEY = <check_references_below>

Deploying your model using the LlamaIndex framework​

MAAP now provides the option to choose if you want to use LlamaIndex as your main framework to deploy your LLM models and embeddings.

This can be done by adding the 'framework' configuration to the config.yaml file

Chat Model​

  • Config File​

    llms:
      class_name: Cohere
      model_name: <check_references_below>
      framework: 'LlamaIndex'

Embedding Model​

  • Config File​

    embedding:
      class_name: Cohere
      model_name: <check_references_below>
      framework: 'LlamaIndex'

For optimal quality the recommended length of each chunk should be under 512 tokens.

References​

Provided below are the instructions on how to procure the right values for building your MAAP framework.

  • Model Name​

    You can pick any model from the updated list given in Cohere documentation.

  • API Key​

    You will need to sign up and retrieve a Cohere API Key.

  • API Documentation​

For more information about how it was implemented for LlamaIndex click here.