Component to finetune the model using the medical image data

Version: 0.0.1

View in Studio: https://ml.azure.com/registries/azureml/components/medimageinsight_adapter_finetune/version/0.0.1

Name	Description	Type	Default	Optional	Enum
train_data_path	Path to the training data file.	uri_file		False
validation_data_path	Path to the validation data file.	uri_file		False
validation_text_tsv	Path to the evaluation text TSV file.	uri_file		False
train_text_tsv	Path to the text TSV file.	uri_file		False
train_dataloader_batch_size	Batch size for the training dataloader.	integer	8	True
validation_dataloader_batch_size	Batch size for the validation dataloader.	integer	1	True
train_dataloader_workers	Number of workers for the training dataloader.	integer	2	True
validation_dataloader_workers	Number of workers for the validation dataloader.	integer	2	True
label_file	Path to the label file.	uri_file		False
hidden_dimensions	Number of hidden dimensions.	integer	512	True
input_channels	Number of input channels.	integer	1024	True
learning_rate	Learning rate for training.	number	0.0003	True
max_epochs	Maximum number of epochs for training.	integer	10	True
track_metric	Metric to track when calculating best model. acc or auc supported.	string	acc	True

Name	Description	Type
output_model_path	Path to save the output model.	uri_folder

azureml://registries/azureml/environments/acft-medimageinsight-adapter-finetune/versions/8