temporian.from_tensorflow_record #
from_tensorflow_record(
path: Union[str, List[str]],
schema: Schema,
timestamps: str = "timestamp",
format: TFRecordEventSetFormatChoices = TFRecordEventSetFormat.GROUPED_BY_INDEX,
num_parallel_reads: Optional[int] = None,
buffer_size: Optional[int] = None,
) -> EventSet
Imports an EventSet from a TF.Records of TF.Examples.
TF.Records of TF.Examples is one of the standard solution to store data for TensorFlow models. https://www.tensorflow.org/tutorials/load_data/tfrecord
The GZIP compression is used.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
path |
Union[str, List[str]]
|
Path to TF.Record file or list of path to TF.Record files. |
required |
timestamps |
str
|
Name of the output column containing timestamps. |
'timestamp'
|
format |
TFRecordEventSetFormatChoices
|
Format of the events inside the received record. At the moment only TFRecordEventSetFormat.GROUPED_BY_INDEX is supported. See TFRecordEventSetFormat for more. |
GROUPED_BY_INDEX
|
num_parallel_reads |
Optional[int]
|
Number of files to read in parallel. Only used if
|
None
|
buffer_size |
Optional[int]
|
Number of bytes in the buffer. If not set, use a sensible default value. |
None
|
Returns:
Type | Description |
---|---|
EventSet
|
Imported EventSet. |