seqgra.model.model.modeldefinition module¶

class ModelDefinition(model_id: str = '', name: str = '', description: str = '', task: str = 'multi-class classification', sequence_space: str = 'DNA', library: str = 'PyTorch', implementation: Optional[str] = None, input_encoding: Optional[str] = '1D', labels: Optional[List[str]] = None, seed: int = 0, architecture: Optional[seqgra.model.model.architecture.Architecture] = None, loss_hyperparameters: Optional[Dict[str, str]] = None, optimizer_hyperparameters: Optional[Dict[str, str]] = None, training_process_hyperparameters: Optional[Dict[str, str]] = None)[source]¶

Bases: object

TODO

id¶

learner ID, used for output folder name

Type: str

name¶

learner name

Type: str

description¶

concise description of the model architecture

Type: str

task¶

one of the following: multi-class classification, multi-label classification, multiple regression, multivariate regression

Type: str

sequence_space¶

one of the following: DNA, protein

Type: str

library¶

one of the following: TensorFlow, PyTorch

Type: str

implementation¶

class name of the learner implementation, e.g., KerasDNAMultiLabelClassificationLearner

Type: Optional[str]

labels¶

class labels expected from output layer

Type: List[str]

seed¶

seed for Python, NumPy, and machine learning library

Type: int

architecture¶

model architecture

Type: Architecture

loss_hyperparameters¶

hyperparmeters for loss function, e.g., type of loss function

Type: Dict[str, str]

optimizer_hyperparameters¶

hyperparmeters for optimizer, e.g., optimizer type

Type: Dict[str, str]

training_process_hyperparameters¶

hyperparmeters regarding the training process, e.g., batch size

Type: Dict[str, str]