Many IC50 values are outside valid bounds. We can consider update the fit_curve.py script to include a 'valid_ic50' flag that determines if the corresponding value is within a range. This will help filtering down the line for those that are predicting against IC50.