Something wrong after model.fix_symbolic()

[GuoMinghui07]

I encountered a RuntimeError while performing symbolic regression. The error message indicates that one of the arguments has an illegal value. Below is the detailed error message and my code.

Here is the code I used for symbolic regression:

from kan import *
f = lambda x: torch.cosh(3*x[:, [0]]**2 + 2*x[:, [0]] + x[:, [1]])
dataset = create_dataset(f, n_var=2, ranges=[-1,1], train_num=1000)
model = MultKAN(width=[2,1,1], base_fun='identity', auto_save=True,seed=4)
model.fit(dataset, steps=100)
model.get_act(dataset)
model.plot()

I found the result of fixing (1,0,0) with exp, r2=0.9992531538009644, c=2 was incorrect then I fixed it by using:

model.unfix_symbolic(1,0,0)
model.fix_symbolic(1,0,0,'cosh')
model.fit(dataset, steps=50)

and there was an error similar to #287