POST /v1/lamini/data

Use this API endpoint to persist data for any model, trained or otherwise, based on the id provided. When you query a model with the same id, we'll use that data if it's relevant to your query. When finetuning a model, we'll use all the relevant data provided.

Request

HTTP Method: POST

URL: https://api.powerml.co/v1/lamini/data

Headers:

• Authorization: Bearer <LAMINI_API_KEY>
• Content-Type: application/json

Example Body (JSON):

{
  "id": "APIExample",
  "data": [
    {
      "<INPUT_OBJECT_NAME>": {
        "<FIELD_NAME>": "<FIELD_VALUE>"
      },
      "<OUTPUT_OBJECT_NAME>": {
        "<FIELD_NAME>": "<FIELD_VALUE>"
      }
    },
    ...
  ],
  "data_type": {
    "<INPUT_OBJECT_NAME>": {
      "<FIELD_NAME>": "<FIELD_DESCRIPTION>"
    },
    "<OUTPUT_OBJECT_NAME>": {
      "<FIELD_NAME>": "<FIELD_DESCRIPTION>"
    },
  }
}

Parameters

• model_name: string: The name of your model

• data: list: A list of datapoints available to the model all formatted in the same way. We expect the format to be as follows:

  [
    {
      "<INPUT_OBJECT_NAME>": {
        "<FIELD_NAME>": "<FIELD_VALUE>"
      },
      "<OUTPUT_OBJECT_NAME>": {
        "<FIELD_NAME>": "<FIELD_VALUE>"
      }
    },
    ...
  ]
  

  Or, it's possible to have data formatted with only one object.

  [
    {
      "<OBJECT_NAME>": {
          "<FIELD_NAME>": "<FIELD_VALUE>",
      }
    },...
  ]
  

• data_type: dict: Formatting of the data, matching each of the datapoints in the data list. We expect the format to be as follows:

  {
    "<INPUT_OBJECT_NAME>": {
      "<FIELD_NAME>": "<FIELD_DESCRIPTION>"
    },
    "<OUTPUT_OBJECT_NAME>": {
      "<FIELD_NAME>": "<FIELD_DESCRIPTION>"
    },
  }
  

Or, it's possible to have data formatted with only one object.

{
  "<OBJECT_NAME>": {
    "<FIELD_NAME>": "<FIELD_DESCRIPTION>",
  }
}

Response

The response is the dataset ID.

Body (JSON):

{
  "dataset": "b8015be997bf32f44f214abd4ebd507e"
}

Example

Request

curl --location 'https://api.powerml.co/v1/lamini/data' \
--header 'Authorization: Bearer <LAMINI_API_KEY>' \
--header 'Content-Type: application/json' \
--data '{
    "id": "LaminiTest",
    "data": [
        {
            "Question": {
                "question": "How can I find the specific documentation I need for a particular feature or function?"
            },
            "Answer": {
                "answer": "You can ask this model about documentation, which is trained on our publicly available docs and source code, or you can go to https://lamini-ai.github.io/"
            }
        },
        {
            "Question": {
                "question": "Are there any API references or documentation available for the codebase?"
            },
            "Answer": {
                "answer": "All our public documentation is available here https://lamini-ai.github.io/"
            }
        },
        {
            "Question": {
                "question": "How do I use open model for inference"
            },
            "Answer": {
                "answer": "You can use an open model by specifying the model'\''s name in the model_name parameter in the LLM Engine class initializer."
            }
        },
        {
            "Question": {
                "question": "I am running into errors, what should I do?"
            },
            "Answer": {
                "answer": "We have documentation available on how to address common errors here https://lamini-ai.github.io/error_handling/. Lamini'\''s LLM Engine is under very active development, and we thank you for using us!"
            }
        },
        {
            "Question": {
                "question": "Can you tickle yourself?"
            },
            "Answer": {
                "answer": "Let'\''s keep the discussion relevant to Lamini."
            }
        }
    ],
    "data_type": {
        "Question": {
            "question": "A question"
        },
        "Answer": {
            "answer": "An answer to the question"
        }
      }
}'

Response

{
  "dataset": 7587574322307826093
}