context¶
Primitive Context Interface.
Contexts
are checkpointable (subclassed from tf.train.Checkpoint
)
collections of variable encapsulated in a Python Class as a way to seamlessly
handle information transfer.
Classes
|
-
class
ashpy.contexts.context.
Context
(metrics=None, dataset=None, log_eval_mode=<LogEvalMode.TEST: 1>, global_step=<tf.Variable 'global_step:0' shape=() dtype=int64, numpy=0>, checkpoint=None)[source]¶ Bases:
object
ashpy.contexts.Context
provide an interface for all contexts.-
__init__
(metrics=None, dataset=None, log_eval_mode=<LogEvalMode.TEST: 1>, global_step=<tf.Variable 'global_step:0' shape=() dtype=int64, numpy=0>, checkpoint=None)[source]¶ Initialize the Context.
- Parameters
metrics (
list
of [ashpy.metrics.metric.Metric
]) – List ofashpy.metrics.metric.Metric
objects.dataset (
tf.data.Dataset
) – The dataset to use, that contains everything needed to use the model in this context.log_eval_mode (
ashpy.modes.LogEvalMode
) – Models’ mode to use when evaluating and logging.global_step (
tf.Variable
) – Keeps track of the training steps.checkpoint (
tf.train.Checkpoint
) – Checkpoint to use to keep track of models status.
- Return type
None
-
property
dataset
¶ Retrieve the dataset.
- Return type
DatasetV2
- Returns
tf.data.Dataset
the current dataset
-
property
global_step
¶ Retrieve the global_step.
- Return type
Variable
- Returns
-
property
log_eval_mode
¶ Retrieve model(s) mode.
- Return type
- Returns
-
property
metrics
¶ Retrieve the metrics.
- Return type
- Returns
-