databricks / spark-deep-learning

Deep Learning Pipelines for Apache Spark
https://databricks.github.io/spark-deep-learning
Apache License 2.0
1.99k stars 494 forks source link

No module named keras.backend #210

Closed christophelebrun closed 4 years ago

christophelebrun commented 4 years ago

Hello

Trying to use sparkdl on Amazon EMR, I import Keras with a booststrap script on my nodes. In the log of my bootstrap step, I got : Successfully installed h5py-2.10.0 keras-2.3.1 keras-applications-1.0.8 keras-preprocessing-1.1.0 numpy-1.16.5 scipy-1.2.2 six-1.13.0.

Nevertheless, when I run my SparkApp, I got an import error. Here is the traceback :

    from sparkdl import DeepImageFeaturizer
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1576070154153_0001/container_1576070154153_0001_02_000001/databricks_spark-deep-learning-1.5.0-spark2.4-s_2.11.jar/sparkdl/__init__.py", line 16, in <module>

  File "/mnt1/yarn/usercache/hadoop/appcache/application_1576070154153_0001/container_1576070154153_0001_02_000001/databricks_spark-deep-learning-1.5.0-spark2.4-s_2.11.jar/sparkdl/transformers/keras_image.py", line 17, in <module>
  File "/mnt1/yarn/usercache/hadoop/appcache/application_1576070154153_0001/container_1576070154153_0001_02_000001/databricks_spark-deep-learning-1.5.0-spark2.4-s_2.11.jar/sparkdl/param/__init__.py", line 16, in <module>

  File "/mnt1/yarn/usercache/hadoop/appcache/application_1576070154153_0001/container_1576070154153_0001_02_000001/databricks_spark-deep-learning-1.5.0-spark2.4-s_2.11.jar/sparkdl/param/shared_params.py", line 22, in <module>
ImportError: No module named keras.backend

Any idea of how to handle this ?

Thank you very much ! Christophe