trainer

Documentation

Index

• type CommitInfo
• type ModelTrainer
  • func NewModelTrainer(trainAPIClient train_api_client.Client, connAPIClient conn_api_client.Client, ...) *ModelTrainer
  • func (mt *ModelTrainer) DownloadAlgorithm(k8sTraining *training.K8sTrainer) error
  • func (mt *ModelTrainer) SaveResult() error
  • func (mt *ModelTrainer) Setup() (err error)

Types

type CommitInfo

type CommitInfo struct {
	CommitID string
}

type ModelTrainer

type ModelTrainer struct {
	// contains filtered or unexported fields
}

func NewModelTrainer

func NewModelTrainer(
	trainAPIClient train_api_client.Client,
	connAPIClient conn_api_client.Client,
	trainerConfig config.TrainerConfig) *ModelTrainer

func (*ModelTrainer) DownloadAlgorithm

func (mt *ModelTrainer) DownloadAlgorithm(k8sTraining *training.K8sTrainer) error

func (*ModelTrainer) SaveResult

func (mt *ModelTrainer) SaveResult() error

This function saves a training result. To do this, it performs the following steps:

1. It extracts the training entity from repository storage, for example, from the API server.
2. It creates the training zip archive.
2. It uploads the training zip archive to object storage.
4. Finally, it saves the training entity results.

func (*ModelTrainer) Setup

func (mt *ModelTrainer) Setup() (err error)

This function prepares a training environment. To do this, it performs the following steps:

1. It extracts the training entity from repository storage, for example, from the API server.
2. It downloads the source code of model training.
3. The setup function downloads all training data.
4. Finally, it saves the training entity to allow an ML toolchain to use it.