I tried fixing it but it ended up in an error on line 1288 in sklearn/preprocessing/data.py in normalize() because the function inplace_csr_row_normalize_l2(X) didn't seem to be able to handle float32