aixplain.modules.model.utility_model

Copyright 2024 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: Thiago Castro Ferreira, Shreyas Sharma and Lucas Pavanelli Date: November 25th 2024 Description: Utility Model Class

BaseUtilityModelParams Objects

class BaseUtilityModelParams(BaseModel)

[view_source]

Base model for utility model parameters.

This class defines the basic parameters required to create or update a utility model.

Attributes:

• name Text - The name of the utility model.
• code Union[Text, Callable] - The implementation code, either as a string or a callable function.
• description Optional[Text] - A description of what the utility model does. Defaults to None.

UtilityModelInput Objects

@dataclass
class UtilityModelInput()

[view_source]

A class representing an input parameter for a utility model.

This class defines the structure and validation rules for input parameters that can be used with utility models.

Attributes:

• name Text - The name of the input parameter.
• description Text - A description of what this input parameter represents.
• type DataType - The data type of the input parameter. Must be one of: TEXT, BOOLEAN, or NUMBER. Defaults to DataType.TEXT.

validate

def validate()

[view_source]

Validate that the input parameter has a supported data type.

Raises:

• ValueError - If the type is not one of: TEXT, BOOLEAN, or NUMBER.

to_dict

def to_dict()

[view_source]

Convert the input parameter to a dictionary representation.

Returns:

• dict - A dictionary containing the input parameter's name, description, and type (as a string value).

utility_tool

def utility_tool(name: Text,
                 description: Text,
                 inputs: List[UtilityModelInput] = None,
                 output_examples: Text = "",
                 status=AssetStatus.DRAFT)

[view_source]

Decorator for utility tool functions

Arguments:

• name - Name of the utility tool
• description - Description of what the utility tool does
• inputs - List of input parameters, must be UtilityModelInput objects
• output_examples - Examples of expected outputs
• status - Asset status

Raises:

• ValueError - If name or description is empty
• TypeError - If inputs contains non-UtilityModelInput objects

UtilityModel Objects

class UtilityModel(Model, DeployableMixin)

[view_source]

Ready-to-use Utility Model.

Note: Non-deployed utility models (status=DRAFT) will expire after 24 hours after creation. Use the .deploy() method to make the model permanent.

Attributes:

• id Text - ID of the Model
• name Text - Name of the Model
• code Union[Text, Callable] - code of the model.
• description Text - description of the model. Defaults to "".
• inputs List[UtilityModelInput] - inputs of the model. Defaults to [].
• output_examples Text - output examples. Defaults to "".
• api_key Text, optional - API key of the Model. Defaults to None.
• supplier Union[Dict, Text, Supplier, int], optional - supplier of the asset. Defaults to "aiXplain".
• version Text, optional - version of the model. Defaults to "1.0".
• function Function, optional - model AI function. Defaults to None.
• name0 bool, optional - Is the user subscribed. Defaults to False.
• name1 Dict, optional - model price. Defaults to None.
• name2 AssetStatus, optional - status of the model. Defaults to AssetStatus.DRAFT.
• name3 - Any additional Model info to be saved

__init__

def __init__(id: Text,
             name: Optional[Text] = None,
             code: Union[Text, Callable] = None,
             description: Optional[Text] = None,
             inputs: List[UtilityModelInput] = [],
             output_examples: Text = "",
             api_key: Optional[Text] = None,
             supplier: Union[Dict, Text, Supplier, int] = "aiXplain",
             version: Optional[Text] = None,
             function: Optional[Function] = None,
             is_subscribed: bool = False,
             cost: Optional[Dict] = None,
             status: AssetStatus = AssetStatus.DRAFT,
             function_type: Optional[FunctionType] = FunctionType.UTILITY,
             **additional_info) -> None

[view_source]

Initialize a new UtilityModel instance.

Arguments:

• id Text - ID of the utility model.
• name Optional[Text], optional - Name of the utility model. If not provided, will be extracted from the code if decorated. Defaults to None.
• code Union[Text, Callable], optional - Implementation code, either as a string or a callable function. Defaults to None.
• description Optional[Text], optional - Description of what the model does. If not provided, will be extracted from the code if decorated. Defaults to None.
• inputs List[UtilityModelInput], optional - List of input parameters the model accepts. If not provided, will be extracted from the code if decorated. Defaults to [].
• output_examples Text, optional - Examples of the model's expected outputs. Defaults to "".
• api_key Optional[Text], optional - API key for accessing the model. Defaults to None.
• supplier Union[Dict, Text, Supplier, int], optional - Supplier of the model. Defaults to "aiXplain".
• version Optional[Text], optional - Version of the model. Defaults to None.
• function Optional[Function], optional - Function type. Must be Function.UTILITIES. Defaults to None.
• name0 bool, optional - Whether the user is subscribed. Defaults to False.
• name1 Optional[Dict], optional - Cost information for the model. Defaults to None.
• name2 AssetStatus, optional - Current status of the model. Defaults to AssetStatus.DRAFT.
• name3 Optional[FunctionType], optional - Type of the function. Defaults to FunctionType.UTILITY.
• name4 - Any additional model info to be saved.

Raises:

• name5 - If function is not Function.UTILITIES.

Notes:

Non-deployed utility models (status=DRAFT) will expire after 24 hours. Use the .deploy() method to make the model permanent.

validate

def validate()

[view_source]

Validate the Utility Model.

This method checks if the utility model exists in the backend and if the code is a string with s3://. If not, it parses the code and updates the description and inputs and does the validation. If yes, it just does the validation on the description and inputs.

to_dict

def to_dict()

[view_source]

Convert the utility model to a dictionary representation.

This method creates a dictionary containing all the essential information about the utility model, suitable for API requests or serialization.

Returns:

• dict - A dictionary containing:
  • name (str): The model's name
  • description (str): The model's description
  • inputs (List[dict]): List of input parameters as dictionaries
  • code (Union[str, Callable]): The model's implementation code
  • function (str): The function type as a string value
  • outputDescription (str): Examples of expected outputs
  • status (str): Current status as a string value

update

def update()

[view_source]

Update the Utility Model.

This method validates the utility model and updates it in the backend.

Raises:

• Exception - If the update fails.

save

def save()

[view_source]

Save the Utility Model.

This method updates the utility model in the backend.

delete

def delete()

[view_source]

Delete the Utility Model.

This method deletes the utility model from the backend.

__repr__

def __repr__()

[view_source]

Return a string representation of the UtilityModel instance.

Returns:

• str - A string in the format "UtilityModel: <name> by <supplier> (id=<id>)". If supplier is a dictionary, uses supplier['name'], otherwise uses supplier directly.