To accelerate the exploration of novel materials, the deep-learning-based inverse design for the intelligent discovery of organic molecules was introduced by experts in computational materials. The novel molecules with desired properties can be generated from the trained continuous latent space, which describes the high-level feature for chemical structure. In addition to continuous representation which relates to actual chemical structures, we reschedule each loss in our model and search optimal molecules with precise chemical structures efficiently. This approach helps us not only to generate the valid chemical structures by our deep generative model precisely but also to correspond molecules to the perspective of physical significance in each chemical property. We implement organic molecules on our model with electrical properties.