Accurate prediction of crop production is of utmost importance in the agricultural sector, as it allows farmers and decision-makers to plan and optimize their resources effectively. Numerous factors, including weather conditions, soil moisture, and temperature, have a direct impact on crop yield. To address this challenge, this paper proposes a hybrid model that combines 1D CNN and LSTMwith an attention layer, aiming to provide precise crop production forecasting. The study targets on the prediction of two frequently cultivated crops in India, namely wheat and rice. These crops were chosen due to their significant contribution to the agricultural landscape of the country. Different modification were donein CNN-LSTM to achieve the higher accuracy and at last CNN-LSTM multi-head attention with multiplication skip connection was modified. The suggested hybrid model was compared to conventional machine learning methods in order to assess its efficiency, including Support Vector Regressor, Decision Tree Regressor, and Random Forest Regressor. The experiments’ findings illustrate how the modified hybrid model outperforms other models and previous studies, achieving impressive performance metrics. The RMSE value of 0.017 indicates the small margin of error in the predicted crop production. The MAE value of 0.09 further emphasizes the accuracy of the model’s predictions. Additionally, the achieved R2 value of 0.967 signifies a high level of correlation between the predicted and actual crop production