Reinforcement Learning Generation of 4-Qubits Entangled States