Currently the hyperparamter for the learning rate for all sections is chosen through evaluating the adverse reactions section. It may be that the learning-rate is suboptimal for the boxed warnings section. Add an experiment to evaluate different learning rates for this section, all other parameters could be the same as those used in Experiment 10F.
Currently the hyperparamter for the learning rate for all sections is chosen through evaluating the adverse reactions section. It may be that the learning-rate is suboptimal for the boxed warnings section. Add an experiment to evaluate different learning rates for this section, all other parameters could be the same as those used in Experiment 10F.