Hello, I would like to know if PINN can perform the following functions.
A model is first trained, and then the weights and biases of this model are saved such that the model can be recovered later and used to train a new model, which is obtained by modifying some of the parameters of the original model.
In short it is to use the already trained model to achieve faster training of the new model.
Thanks!
lululxvi
commented
1 year ago
Hello, I would like to know if PINN can perform the following functions. A model is first trained, and then the weights and biases of this model are saved such that the model can be recovered later and used to train a new model, which is obtained by modifying some of the parameters of the original model. In short it is to use the already trained model to achieve faster training of the new model. Thanks!