Trainer
E457347
Trainer is a TensorFlow Extended component responsible for training machine learning models within end-to-end ML pipelines.
Statements (48)
| Predicate | Object |
|---|---|
| instanceOf |
TFX component
ⓘ
TensorFlow Extended component ⓘ |
| canBeOrchestratedBy |
Apache Airflow
NERFINISHED
ⓘ
Kubeflow Pipelines NERFINISHED ⓘ Vertex AI Pipelines NERFINISHED ⓘ |
| canUse |
custom training loops
ⓘ
tf.estimator API ⓘ tf.keras API NERFINISHED ⓘ |
| configuredBy |
RunFn
ⓘ
Trainer component configuration ⓘ TrainerArgs ⓘ |
| documentationURL | https://www.tensorflow.org/tfx/guide/trainer ⓘ |
| executes |
metric logging
ⓘ
model checkpointing ⓘ training loop ⓘ |
| hasRole | model training component ⓘ |
| implementedIn | Python NERFINISHED ⓘ |
| input |
Example artifacts
ⓘ
Hyperparameters ⓘ Schema artifacts ⓘ TransformGraph artifacts ⓘ |
| integratesWith |
Evaluator
ⓘ
ExampleGen NERFINISHED ⓘ Pusher NERFINISHED ⓘ Transform ⓘ Tuner ⓘ |
| license | Apache License 2.0 ⓘ |
| maintainedBy | TensorFlow team NERFINISHED ⓘ |
| output |
Model artifacts
ⓘ
ModelEvaluation artifacts ⓘ SavedModel ⓘ |
| partOf |
TFX pipeline
NERFINISHED
ⓘ
TensorFlow Extended NERFINISHED ⓘ |
| requires |
data input function
ⓘ
model definition ⓘ user-defined training function ⓘ |
| supports |
GPU acceleration
ⓘ
TPU acceleration ⓘ distributed training ⓘ hyperparameter tuning ⓘ warm-starting models ⓘ |
| supportsFramework |
Keras
NERFINISHED
ⓘ
TensorFlow NERFINISHED ⓘ |
| usedFor |
deep learning model training
ⓘ
supervised learning model training ⓘ training machine learning models ⓘ |
| usedIn |
end-to-end ML pipelines
ⓘ
production ML pipelines ⓘ |
Referenced by (1)
Full triples — surface form annotated when it differs from this entity's canonical label.