Skip to content

Model kinds

At the moment, we support the following kinds:

  • model: represents a generic ML model
  • mlflow: represents a MLflow model
  • sklearn: represents a scikit-learn model
  • huggingface: represents a HuggingFace model

For each different kind, the Model object has its own subclass with different spec and status attributes.

Model

The model kind indicates that the model is a generic ML model. It's usefull to represent a generic ML model as a Model object.

Model spec parameters

Parameter Type Description Default
path str Path of the model, can be a local path or a remote path, a single filepath or a directory/partition. required
framework str Model framework (e.g. 'pytorch'). None
algorithm str Model algorithm (e.g. 'resnet'). None
base_model str Base model. None
parameters dict Model parameters. None
metrics dict Model metrics. None

Model methods

The model kind has no additional methods.

Mlflow

The mlflow kind indicates that the model is an MLflow model. It's usefull to represent an MLflow model as a Model object.

Mlflow spec parameters

Parameter Type Description Default
path str Path of the model, can be a local path or a remote path, a single filepath or a directory/partition. required
framework str Model framework (e.g. 'pytorch'). None
algorithm str Model algorithm (e.g. 'resnet'). None
base_model str Base model. None
parameters dict Model parameters. None
metrics dict Model metrics. None
flavor str Mlflow model flavor. None
model_config dict Mlflow model config. None
input_datasets list[Dataset] Mlflow input datasets (see below). None
signature Signature Mlflow model signature (see below). None

Dataset

Parameter Type Description Default
name str Dataset name. None
digest str Dataset digest. None
profile str Dataset profile. None
schema str Dataset schema. None
source str Dataset source. None
source_type str Dataset source type. None

Signature

Parameter Type Description Default
inputs str Signature inputs. None
outputs str Signature outputs. None
parameters str Signature parameters. None

Mlflow methods

The mlflow kind has no additional methods.

Sklearn

The sklearn kind indicates that the model is an Sklearn model. It's usefull to represent an Sklearn model as a Model object.

Sklearn spec parameters

Parameter Type Description Default
path str Path of the model, can be a local path or a remote path, a single filepath or a directory/partition. required
framework str Model framework (e.g. 'pytorch'). None
algorithm str Model algorithm (e.g. 'resnet'). None
base_model str Base model. None
parameters dict Model parameters. None
metrics dict Model metrics. None

Sklearn methods

The sklearn kind has no additional methods.

Huggingface

The huggingface kind indicates that the model is an Huggingface model. It's usefull to represent an Huggingface model as a Model object.

Huggingface spec parameters

Parameter Type Description Default
path str Path of the model, can be a local path or a remote path, a single filepath or a directory/partition. required
framework str Model framework (e.g. 'pytorch'). None
algorithm str Model algorithm (e.g. 'resnet'). None
base_model str Base model. None
parameters dict Model parameters. None
metrics dict Model metrics. None
model_id str Huggingface model id. If not specified, the model is loaded from the model path None
model_revision str Huggingface model revision. None

Huggingface methods

The huggingface kind has no additional methods.