Architectures for continuous vector labels prediction

Question

I have a small dataset of feature vectors of size 200 to each corresponds a larger vector of size 1000. I would like to "predict" a large vector for every small one. The task sounds like sort of regression or classification using "continuous category of labels" hence for that purpose I thought either using ResNet or convolutional autoencoder. I do wonder though which model to pick up. Which trade-offs should I consider and which model would you pick?