TensorFlow - Gobal Stop has been increased

I took the code example from TensorFlow of a BoostedTreeRegressor. (https://www.tensorflow.org/tutorials/estimator/boosted_trees)

Everything worked fine even if some warnings appeared and then I tried changing the dataset. From the titanic dataset I went with a dataset containing environmental concentrations to predict an index of quality.

The same warnings appear but this time it stopped after the gobal step warning without giving me any results.

I saw that this message appears when a NaN loss is encountered and that the batch is skipped and the loss scale is adjusted which end in the global step not being increased.

WARNING:tensorflow:It seems that global step (tf.train.get_global_step) has not been increased. 
Current value (could be stable): 0 vs previous value: 0. You could increase the global step by passing 
tf.train.get_global_step() to Optimizer.apply_gradients or Optimizer.minimize.

Process finished with exit code -1073740940 (0xC0000374)

How could I solve my issue? Usually the warning is only begnin so maybe the issue comes from somewhere else

Thank you for any help you could provide me