sklearn very different f1 score between positive and negative classes

Question

I am using sklearn to do binary classification, and the result is shown below. Number of positive and negative samples are roughly 3 : 2. But why there is such a big difference