utils.types.ModelTypes

View source on Github

Arize enum to specify your model type represented in the platform and validate applicable performance metrics.

Model Types

from arize.utils.types import ModelTypes

Specify a model_type when logging a prediction for the first time for a new model.

Method

list_types()

View Source

Returns a list of all model types.

ModelTypes.list_types()

Use Case	SDK ModelType	Description
Regression	ModelTypes.REGRESSION	Regression models predict continuous values
Binary Classification	ModelType.BINARY_CLASSIFICATION	Binary classification models predict only two categorical values, typically represented as 0 or 1
Ranking	ModelType.RANKING	Ranking models predict the relative ordering of a set of items based on their features
Multi Class	ModelType.MULTI_CLASS	Multiclass models predict multiple categorical values
Natural Language Processing (NLP)	ModelType.SCORE_CATEGORICAL	NLP models are categorical models specifically designed to work with text data and perform various tasks (i.e. sentiment analysis and language translation)
Computer Vision (CV)	ModelType.SCORE_CATEGORICAL	CV models are categorical models specifically designed to work with visual data and perform various tasks (i.e. object detection and image classification)
Large Language Models (LLMs)	ModelTypes.GENERATIVE_LLM	Models that use vast amounts of data to generate human-like language and perform a wide range of natural language processing tasks
Object Detection	ModelTypes.OBJECT_DETECTION	Object detection models identify and locate objects within images or videos by assigning them specific bounding boxes

Code Example

response = arize_client.log(
    model_id='sample-binary-classification-model', 
    ...
    model_type=ModelTypes.BINARY_CLASSIFICATION
)

response = arize_client.log(
    model_id='sample-regression-model', 
    ...
    model_type=ModelTypes.REGRESSION
)

response = arize_client.log(
    model_id='sample-ranking-model', 
    ...
    model_type=ModelTypes.RANKING
)