Closed donupup closed 3 months ago
This error occurred when the model architecture was changed. I think you pruned or removed any edge or node to create a smaller network. If so, you need to initialize the new model the same way as your latest model architecture.
In the newest update, grid refinement can be done more easily model = model.refine(new_grid)
. Tutorial: https://github.com/KindXiaoming/pykan/blob/master/tutorials/Example_1_function_fitting.ipynb. Please let me know if your problem persists.
This error occurred when the model architecture was changed. I think you pruned or removed any edge or node to create a smaller network. If so, you need to initialize the new model the same way as your latest model architecture.
thanks for your reply, the problem was solved
In the newest update, grid refinement can be done more easily
model = model.refine(new_grid)
. Tutorial: https://github.com/KindXiaoming/pykan/blob/master/tutorials/Example_1_function_fitting.ipynb. Please let me know if your problem persists.
The problem was solved, and as @kaneyxx said, I did the pruning in subsequent code, and this leads to an error
做实验的过程中遇到下面这个问题,从小的grid初始化另一个较大的grid的KAN网络时,我并没有修改数据集的内容以及形状,但是initialize失败。我尝试将输入维度改为10,就可以成功的初始化,这是因为输入维度过高的原因吗?这个问题应该怎么解决? 代码如下:
报错如下: