HyperNetWork is not being trained

[mleoelm]

With respect to the file "HyerIQASolver.py"

In line 20 you store the parameter of the HyperNetwork to be trained in the variable "self.hypernet_params".

self.hypernet_params = filter(lambda p: id(p) not in backbone_params, self.model_hyper.parameters())

From line 77 to 84 you update the optimizer with this variable self.hypernet_params

Since self.model_hyper.parameters() returns an iterator, self.hypernet_params is an iterator and after the optimizer is initialized, it becomes empty. Therefore, after the first epoch, the optimizer doesn't optimize the HyperNetwork anymore because self.hypernet_params is empty.

Setting self.hypernet_params as a list should fix the problem: self.hypernet_params = list(filter(lambda p: id(p) not in backbone_params, self.model_hyper.parameters()))

[SuperBruceJia]

@SSL92

[SuperBruceJia]

With respect to the file "HyerIQASolver.py"

In line 20 you store the parameter of the HyperNetwork to be trained in the variable "self.hypernet_params".

self.hypernet_params = filter(lambda p: id(p) not in backbone_params, self.model_hyper.parameters())

From line 77 to 84 you update the optimizer with this variable self.hypernet_params

Since self.model_hyper.parameters() returns an iterator, self.hypernet_params is an iterator and after the optimizer is initialized, it becomes empty. Therefore, after the first epoch, the optimizer doesn't optimize the HyperNetwork anymore because self.hypernet_params is empty.

Setting self.hypernet_params as a list should fix the problem: self.hypernet_params = list(filter(lambda p: id(p) not in backbone_params, self.model_hyper.parameters()))

After trained the model, I found the performances are almost the same with or without using the list.
Should we have to use list(...) for self. hypernet_params ?