Skip to content

eval_results

Endpoint Documentation: /v1/lamini/train/jobs/{job_id}/eval

Note

You can see these results by going to the train tab at https://app.lamini.ai/train

Get the training evaluation results for a completed training job.

Request

HTTP Method: GET

Path: https://api.lamini.ai/v1/lamini/train/jobs/{job_id}/eval

Headers:

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

Parameters:

  • {job_id} - The unique identifier of the completed training job.

Response

The response will be the evaluation results for a completed training job.

Body (JSON):

  • job_id - ID of the training job
  • eval_results - A list of input strings and output objects. Eval results contain model outputs from both the newly finetuned model and the base model.

Example

This example cancels the training job with the ID 418. The request is authenticated using the LAMINI_API_KEY bearer token.

Request

curl --location --request GET 'https://api.lamini.ai/v1/lamini/train/jobs/418/eval' \
--header 'Authorization: Bearer <LAMINI_API_KEY>' \
--header 'Content-Type: application/json'

Response

{
    "job_id":2514,
    "eval_results": [
        {
            "input":"What is Lamini?",
            "outputs":[
                {
                    "model_name":"4321dvwgeb9c483750b213afc78b49fe875d43db27d508e821c2e92e2701e018",
                    "output":"Lamini is the world's most powerful LLM Engine."
                },
                {
                    "model_name":"meta-llama/Llama-2-7b-hf",
                    "output":"finetune me first!"
                }
            ]
        },
        ...
    ]
}