![](/img/trans.png)
[英]TFX component CsvExampleGen always yields Examples with empty outputs (and inputs)
[英]TFX - How to inspect records from CsvExampleGen
如何檢查加載到 TFX CsvExampleGen 中的數據?
california_housing_train.csv的前 3 行如下所示。
經度 | 緯度 | 住房中位數年齡 | total_rooms | total_bedrooms | 人口 | 家庭 | 收入中位數 | 中值房屋價值 |
---|---|---|---|---|---|---|---|---|
-122.05 | 37.37 | 27 | 3885 | 661 | 1537 | 606 | 6.6085 | 344700 |
-118.3 | 34.26 | 43 | 1510 | 310 | 809 | 277 | 3.599 | 176500 |
-117.81 | 33.78 | 27 | 3589 | 507 | 1484 | 495 | 5.7934 | 270500 |
CSV 被加載到 CsvExampleGen 中。 據我了解,XXXExampleGen 是生成 tf.Record 實例,因此我想知道是否有一種方法可以遍歷 CsvExampleGen 中的記錄。
from tfx.components import (
CsvExampleGen
)
housing = CsvExampleGen("sample_data/california_housing_train.csv")
housing
----------
CsvExampleGen(
spec: <tfx.types.standard_component_specs.FileBasedExampleGenSpec object at 0x7fcd90435450>,
executor_spec: <tfx.dsl.components.base.executor_spec.BeamExecutorSpec object at 0x7fcd90435850>,
driver_class: <class 'tfx.components.example_gen.driver.FileBasedDriver'>,
component_id: CsvExampleGen,
inputs: {},
outputs: {
'examples': OutputChannel(artifact_type=Examples,
producer_component_id=CsvExampleGen,
output_key=examples,
additional_properties={},
additional_custom_properties={})
}
)
for record in housing.outputs['examples']:
print(record)
----> 1 中的 TypeError Traceback (last last call last) for record in Housing.outputs['examples']: 2 print(record)
TypeError:“OutputChannel”對象不可迭代
您是否有機會查看教程中的這一部分,其中解釋了如何顯示ExampleGen
組件的工件? 您可以修改下面的代碼(來源: TFX 教程)以實現相同的目的。
# Get the URI of the output artifact representing the training examples, which is a directory
train_uri = os.path.join(example_gen.outputs['examples'].get()[0].uri, 'Split-train')
# Get the list of files in this directory (all compressed TFRecord files)
tfrecord_filenames = [os.path.join(train_uri, name)
for name in os.listdir(train_uri)]
# Create a `TFRecordDataset` to read these files
dataset = tf.data.TFRecordDataset(tfrecord_filenames, compression_type="GZIP")
# Iterate over the first 3 records and decode them.
for tfrecord in dataset.take(3):
serialized_example = tfrecord.numpy()
example = tf.train.Example()
example.ParseFromString(serialized_example)
pp.pprint(example)
聲明:本站的技術帖子網頁,遵循CC BY-SA 4.0協議,如果您需要轉載,請注明本站網址或者原文地址。任何問題請咨詢:yoyou2525@163.com.