Training of MobileNetV1 and MobileNetV2 on Mnist and Cifar10

[warthecatalyst]

trafficstars

I think the training all get good results and the accuracy seems pretty good as well. It's just that After the training of Cifar10 with MBNv2 will cause an exception in the end...

Exception in thread "main" ai.djl.MalformedModelException: Unsupported encoding version: 3 at ai.djl.nn.SequentialBlock.loadMetadata(SequentialBlock.java:209) at ai.djl.nn.AbstractBaseBlock.loadParameters(AbstractBaseBlock.java:288) at ai.djl.nn.AbstractBaseBlock.loadParameters(AbstractBaseBlock.java:293) at ai.djl.nn.AbstractBaseBlock.loadParameters(AbstractBaseBlock.java:293) at ai.djl.BaseModel.readParameters(BaseModel.java:367) at ai.djl.mxnet.engine.MxModel.loadParameters(MxModel.java:187) at ai.djl.mxnet.engine.MxModel.load(MxModel.java:129) at ai.djl.repository.zoo.BaseModelLoader.loadModel(BaseModelLoader.java:160) at ai.djl.repository.zoo.Criteria.loadModel(Criteria.java:168) at ai.djl.examples.training.transferlearning.TrainMobileNetV2WithCifar10.testSaveParameters(TrainMobileNetV2WithCifar10.java:199) at ai.djl.examples.training.transferlearning.TrainMobileNetV2WithCifar10.runExample(TrainMobileNetV2WithCifar10.java:110) at ai.djl.examples.training.transferlearning.TrainMobileNetV2WithCifar10.main(TrainMobileNetV2WithCifar10.java:69)