PADAM: Parallel averaged Adam reduces the error for stochastic optimization in scientific machine learning | Read Paper on Bytez