-
Notifications
You must be signed in to change notification settings - Fork 36
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
added DyRep #27
added DyRep #27
Conversation
neighbor_loader = LastNeighborLoader(data.num_nodes, size=10, device=device) | ||
|
||
|
||
|
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
should we move GraphAttentionEmbedding
to models
folder as well? Since it will be used in multiple example datasets
@@ -99,7 +90,7 @@ def __init__(self, in_channels): | |||
def forward(self, z_src, z_dst): | |||
h = self.lin_src(z_src) + self.lin_dst(z_dst) | |||
h = h.relu() | |||
return self.lin_final(h) | |||
return self.lin_final(h).sigmoid() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
better approach might be to specify the non-linearity in the arguments for the layer
ie. act="sigmoid"
and if act=="sigmoid", return self.lin_final(h).sigmoid()
@@ -0,0 +1,373 @@ | |||
""" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
after deciding, we should have one example per method for a given dataset. i.e. going with any vs. all or one vs. all
tgb/linkproppred/sample_negative.py
Outdated
@@ -0,0 +1,115 @@ | |||
""" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
should have
- function for generating all negatives (no negative sampling needed, use all)
- function for loading the existing sampled negatives from disc (to do)
Comments
|
The negative sampling with two different variations (RND & HIST-RND) is added (generate batches on the fly). |
Adding myket negative sampling
DyRep implementation is done. The following files are modified for implementing the DyRep model.