Fine-Tune Command R+ Model on Your Own Data Locally

 This video is a simple hands-on tutorial to show how to do fine-tuning of Command R+ model from cohere locally by using your own data.

Code:

#pip install cohere

import os

import cohere

from cohere.finetuning import (

    BaseModel,

    BaseType,

    FinetunedModel,

    Settings,

)

api_key = os.getenv('COHERE_API_KEY')

co = cohere.Client(api_key)

# create a dataset

single_label_dataset = co.datasets.create(name="single-label-dataset",

                                         data=open("train.csv", "rb"),

                                         type="single-label-classification-finetune-input")

#print(co.wait(single_label_dataset))

datasetid=single_label_dataset.id

finetuned_model = co.finetuning.create_finetuned_model(

    request=FinetunedModel(

        name="test-finetuned-model",

        settings=Settings(

            base_model=BaseModel(

                base_type="BASE_TYPE_CLASSIFICATION",

            ),

            dataset_id=datasetid,

        ),

    )

)

print(finetuned_model)