Skip to main content

Replicate

Replicate is an AI model platform designed for developers and organizations to create, share, and deploy machine learning models effortlessly. With a user-friendly interface, it streamlines the entire model development process from experimentation to deployment.

Requisites​

To use models provided by Replicate, you need to have a Replicate API Token. You can get one by signing up at Llama API.

Required credentials:

  • REPLICATE_API_KEY: Your Replicate API Token.

Supported Models:​

All models provided by Llama API has the following properties:

  • Function call: Not Supported
  • Streaming: Not supported

For extra configs when generating texts, different models may accept different parameters, please check your target model's documentation. All the following parameters will be accepted by TaskingAI, but may not take effect if the target model rejects:

  • temperature
  • top_p
  • max_tokens
  • stop
  • top_k

Wildcard​

  • Model schema id: replicate/wildcard

Replicate hosted hundreds of models on its platform from various organizations and developers. TaskingAI created a wildcard model that can integrate all eligible models on Replicate.

Replicate currently runs on two types of APIs. The first one requires the model's name, for example 'meta/llama-2-7b-chat'. Another one requires a version id.

No matter what API schema your target model requires, pass the version id or the model's name to provider_schema_id when creating the model.

replicate-name replicate-id