ZooModel 's pretrained weights do not work

I have been trying to do some image learning using VGG16. It worked very well on Keras. I was trying to get it to run on DL4J but zoomodel’s initPretrained() method fails. It tries to download weights zip from blob.deeplearning4j. This URL is down and the download fails.

Is there an alternate location for this? If not, can I get reference to some code where I can take my Keras h5 model and transfer it into a Computational graph in DL4J?

Are you using an old version?

In beta6 the url is correct: https://github.com/eclipse/deeplearning4j/blob/release/1.0.0-beta6/deeplearning4j/deeplearning4j-common/src/main/java/org/deeplearning4j/common/resources/DL4JResources.java#L47

If you can’t, or don’t want to, upgrade for some reason, you can also set it as a system property on the command line by adding the following option:

-Dorg.deeplearning4j.resources.baseurl=https://dl4jdata.blob.core.windows.net/

Thanks a lot. This was helpful.

Is there a code sample for transfer learnt (Keras) VGG weights to a DL4J computational graph?

When you load your Keras model (as long as it is original keras, not tf.keras at the moment), you actually get a DL4J ComputationGraph, that also means that all transfer learning examples apply to it. You should be able to follow the documentation at https://deeplearning4j.konduit.ai/tuning-and-training/transfer-learning