RLlib experiments

Hi everyone, I am a beginner to this field, in rllib experiments or in any RL experiment, when training agents on environments, whats the common practice, for instance, if i want to train a common gym environment, how should i go about it,whats the criteria of choosing algos? for example’s sake lets say i chose ppo, what should be configurations, should i search for people who have already trained them, if i find them,good enough, if not, what them? is it trial and error ?