Repeating eval-F1 scores with seed + data randomization

I wonder if the model output has reached the ideal value…