Is it possible to get the evaluated value of f(x) from torch.autograd.functional.jacobian(f,x)

Since the function value f(x) is evaluated during the process of getting the jacobian matrix, is there a way to access it?

Thanks for the help.

Hmm, don’t think it is possible right now with the current API, but I think it is a valid request especially if you are setting vectorized=True. Feel to free to file an issue for this.