The lithology of the formation is known to affect the drilling operation. Litho-facies help in the quantification of the formation properties, which optimizes the drilling parameters. The proposed work uses the artificial neural network algorithm and an optimizer to develop a working model for predicting the lithology of any formation within the study area in real-time. The proposed model is trained using the formation data comprising 15-dependent variables from the Eagleford region of the United States of America. It builds a method for measuring or forecasting litho-facies in real-time when drilling through a formation. It uses general drilling parameters for better precision, including rate of penetration, rotation per minute, surface torque, differential pressure, gamma ray correlation, and a D-exponent correlation function. The proposed model compares and assesses various first-order optimization algorithm's efficiency, such as adaptive moment estimation, adaptive gradient, root-mean-square propagation, and stochastic gradient descent with traditional artificial neural network in quantitative litho-facies detection. The model can predict the complex lithology for vertical/inclined/horizontal wellbores in real-time, making it a novel algorithm in the industry. The developed algorithm illustrates an accuracy of 86% using ADAM optimizer when tested with the existing data and improves as the model is trained with more data.