The function "validation_step" can not use the model parameters trained after the function "training_step".

[RunwenHu]

Recently, I used use the pytorch lightning to train the model, but I found that the function "validation_step" can not use the model parameters trained after the function "training_step". I then saved the model parameters and reloaded the parameters for new training, the "validation_step" function can ouput the right value. I want to know how can I solve the problem.