Deep learning draws a lot of attention as a new way of generating unseen structures for drug discovery. We propose a Transformer-based deep model for de novo target-specific molecular design. The proposed method is capable of generating both drug-like compounds and target-specific compounds. The latter are generated by enforcing different keys and values of the multi-head attention for each target. We allow the generation of SMILES strings to be conditional on the specified target. The sampled compounds largely occupy the real target-specific data's chemical space and also cover a significant fraction of novel compounds.