Fixing bug with random weights between wmax and wmin #61
Conversation
bindsnet/models/__init__.py
Outdated
| wmax=wmax, | ||
| norm=norm, | ||
| decay=0.8), | ||
| decay=0), |
There was a problem hiding this comment.
Shouldn’t decay be None by default?
There was a problem hiding this comment.
It should but not on Diehl and Cook implementation
I'm still searching for the good one...
bindsnet/network/topology.py
Outdated
|
|
||
| self.w = kwargs.get('w', torch.rand(*source.shape, *target.shape)) | ||
| self.w = torch.clamp(self.w, self.wmin, self.wmax) | ||
| #self.w = torch.clamp(self.w, self.wmin, self.wmax) |
There was a problem hiding this comment.
Remove this commented out line.
bindsnet/network/topology.py
Outdated
|
|
||
| self.w = kwargs.get('w', torch.rand(*source.shape, *target.shape)) | ||
| self.w = torch.clamp(self.w, self.wmin, self.wmax) | ||
| self.w = self.wmin + self.w*(self.wmax-self.wmin) |
There was a problem hiding this comment.
Actually, this only makes sense if the weights are randomly initialized. This should only happen if the weight aren’t passed in; otherwise, the clamping should still take place.
There was a problem hiding this comment.
If the user pass its weight and it need to be clamping, that a problem. It will bias the weight matrix and the user need to be notify...
bindsnet/network/topology.py
Outdated
| print("Warning: provided weights matrix contain values smaller then :", self.wmin) | ||
| bo = True | ||
| if bo: | ||
| print("Warning: the weights matrix as been clamp between: ", self.wmin," to ", self.wmin," \n The matrix values can be bais to max and min values!!!") |
There was a problem hiding this comment.
If you're going to use warnings, using warnings.
| torch.rand(*source.shape, *target.shape) | ||
| self.w = self.wmin + self.w*(self.wmax-self.wmin) | ||
| else: | ||
| bo = False |
There was a problem hiding this comment.
This is a mess; I'm okay with the logic, but I'm going to refactor this once it's merged.
|
@Hananel-Hazan there's a merge conflict in |
Noted by @Devdher
Clamping make the random values to be more populated my wmax and wmin values.