While Adversarial Imitation Learning (AIL) algorithms have recently led to\nstate-of-the-art results on various imitation learning benchmarks, it is\nunclear as to what impact various design decisions have on performance. To this\nend, we present here an organizing, modular framework called\nReinforcement-learning-based Adversarial Imitation Learning (RAIL) that\nencompasses and generalizes a popular subclass of existing AIL approaches.\nUsing the view espoused by RAIL, we create two new IfO (Imitation from\nObservation) algorithms, which we term SAIfO: SAC-based Adversarial Imitation\nfrom Observation and SILEM (Skeletal Feature Compensation for Imitation\nLearning with Embodiment Mismatch). We go into greater depth about SILEM in a\nseparate technical report. In this paper, we focus on SAIfO, evaluating it on a\nsuite of locomotion tasks from OpenAI Gym, and showing that it outperforms\ncontemporaneous RAIL algorithms that perform IfO.\n