Version 1.11 fitting ability worse than 1.4

Hi developers of pytorch. I am switching my model from version 1.4 to version 1.11 and I am pretty sure that the neural network models in 1.4 and 1.11 are the same. The model trained in 1.4 fits well with the data and the model trained in 1.11 does not fit as well as its counterpart does in 1.4. So to me, it seems that some regularization technique is added to version 1.11?! How can I solve this problem? Thank you for any guidance.