Open ClaudiaShu opened 9 months ago
Hi, I am trying to implement neuralSDE using torchdyn, my code is:
# drift_func f = nn.Sequential(...) # diffusion function g = nn.Sequential(...) self.func = NeuralSDE(f, g, solver=args.solver, rtol=args.rtol, atol=args.atol)
But I got this error:
TypeError: __init__() got an unexpected keyword argument 'func'
I checked the source code and it seems that the initialisation of the model class is different from what input it takes.
class NeuralSDE(SDEProblem, pl.LightningModule): def __init__(self, drift_func, diffusion_func, noise_type ='diagonal', sde_type = 'ito', order=1, sensitivity='autograd', s_span=torch.linspace(0, 1, 2), solver='srk', atol=1e-4, rtol=1e-4, ds = 1e-3, intloss=None): super().__init__(func=SDEFunc(f=drift_func, g=diffusion_func, order=order), order=order, sensitivity=sensitivity, s_span=s_span, solver=solver, atol=atol, rtol=rtol)
What should be given to the model as the input or is there a bug that remains to fix?
Thanks in advance!
Got the same! Has it been resolved? Curious about the solution to this!
Hi, I am trying to implement neuralSDE using torchdyn, my code is:
But I got this error:
I checked the source code and it seems that the initialisation of the model class is different from what input it takes.
What should be given to the model as the input or is there a bug that remains to fix?
Thanks in advance!