I wrote a small (and dirty) code using DNN Regressor in tensorflow. What I tried : - Use a list of numbers as a training set (training set : x = [1,2,3], y =[2, 4, 6] )- Use DNN Regressor to build a model - Make predictions with the model using a python list as a test set (test set : x = [2, 3, 5]) The results are different in every run, anyway I could get the following {'loss': 2.74980