Skip to main content
Version: 1.0

aixplain.factories.finetune_factory

__author__​

Copyright 2022 The aiXplain SDK authors

Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at

http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software distributed under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License.

Author: Duraikrishna Selvaraju, Thiago Castro Ferreira, Shreyas Sharma and Lucas Pavanelli Date: June 14th 2023 Description: Finetune Factory Class

FinetuneFactory Objects​

class FinetuneFactory()

[view_source]

Factory class for creating and managing model fine-tuning operations.

This class provides static methods to create and manage fine-tuning jobs for machine learning models. It handles cost estimation, dataset preparation, and fine-tuning configuration.

Attributes:

  • backend_url str - Base URL for the aiXplain backend API.

create​

@classmethod
def create(cls,
           name: Text,
           dataset_list: List[Union[Dataset, Text]],
           model: Union[Model, Text],
           prompt_template: Optional[Text] = None,
           hyperparameters: Optional[Hyperparameters] = None,
           train_percentage: Optional[float] = 100,
           dev_percentage: Optional[float] = 0) -> Finetune

[view_source]

Create a new fine-tuning job with the specified configuration.

This method sets up a fine-tuning job by validating the configuration, estimating costs, and preparing the datasets and model. It supports both direct Dataset/Model objects and their IDs as inputs.

Arguments:

  • name Text - Name for the fine-tuning job.
  • dataset_list List[Union[Dataset, Text]] - List of Dataset objects or dataset IDs to use for fine-tuning.
  • model Union[Model, Text] - Model object or model ID to be fine-tuned.
  • prompt_template Text, optional - Template for formatting training examples. Use <<COLUMN_NAME>> to reference dataset columns. Defaults to None.
  • hyperparameters Hyperparameters, optional - Fine-tuning hyperparameters configuration. Defaults to None.
  • train_percentage float, optional - Percentage of data to use for training. Must be > 0. Defaults to 100.
  • dev_percentage float, optional - Percentage of data to use for validation. train_percentage + dev_percentage must be <= 100. Defaults to 0.

Returns:

  • Finetune - Configured fine-tuning job object, or None if creation failed.

Raises:

  • AssertionError - If train_percentage <= 0 or train_percentage + dev_percentage > 100.