FEA: Add SGL in General models #1004
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
2017pxy
approved these changes
Oct 20, 2021
ChangxinTian
suggested changes
Oct 25, 2021
Comment on lines
54
to
58
| - ``layers (int)`` : The number of layers in SGL. Defaults to ``3``. | ||
| - ``lamb (float)`` : The temperature in softmax. Defaults to ``0.5``. | ||
| - ``embedding_dim (int)`` : the embedding size of users and items. Defaults to ``64``. | ||
| - ``ratio (float)`` The dropout ratio. Defaults to ``0.1``. | ||
| - ``reg (float)`` : The L2 regularization weight. Defaults to ``1e-05``. |
There was a problem hiding this comment.
Rename parameters to be consistent with other general models.
| self.node_num = self.user_num + self.item_num | ||
| self.user_embedding = torch.nn.Embedding(self.n_users, self.embed_dim, device=self.device) | ||
| self.item_embedding = torch.nn.Embedding(self.n_items, self.embed_dim, device=self.device) | ||
| self.dataset = dataset |
There was a problem hiding this comment.
Defined but not used.
| self.apply(xavier_uniform_initialization) | ||
| self.update = True | ||
|
|
||
| def comp(self): |
There was a problem hiding this comment.
Let function name make sense, such as graph_augment() ?
| #calc user side | ||
| u_emd1 = F.normalize(user_sub1[user_list], dim=1) | ||
| u_emd2 = F.normalize(user_sub2[user_list], dim=1) | ||
| #all_emd2 = F.normalize(self.user_sub2,dim=1) |
There was a problem hiding this comment.
Remove useless code.
| #calc item side | ||
| i_emd1 = F.normalize(item_sub1[pos_item_list], dim=1) | ||
| i_emd2 = F.normalize(item_sub2[pos_item_list], dim=1) | ||
| #all_item = F.normalize(self.item_sub2,dim=1) |
There was a problem hiding this comment.
Remove useless code.
Comment on lines
246
to
250
| v1 = torch.sum(u_emd1 * u_emd2, dim=1) | ||
| v2 = u_emd1.matmul(u_emd2.T) | ||
| v1 = torch.exp(v1 / self.ssl_lamb) | ||
| v2 = torch.sum(torch.exp(v2 / self.ssl_lamb), dim=1) | ||
| ssl_user = -torch.sum(torch.log(v1 / v2)) |
There was a problem hiding this comment.
According to the original paper, SGl treats the views of any different nodes as the negative pairs, rather than other nodes in the batch.
Comment on lines
256
to
260
| v3 = torch.sum(i_emd1 * i_emd2, dim=1) | ||
| v4 = i_emd1.matmul(i_emd2.T) | ||
| v3 = torch.exp(v3 / self.ssl_lamb) | ||
| v4 = torch.sum(torch.exp(v4 / self.ssl_lamb), dim=1) | ||
| ssl_item = -torch.sum(torch.log(v3 / v4)) |
| self.reg_loss = EmbLoss() | ||
| self.train_graph = self.csr2tensor(self.create_adjust_matrix(is_sub=False)) | ||
| self.apply(xavier_uniform_initialization) | ||
| self.update = True |
There was a problem hiding this comment.
Let para name make sense.
ChangxinTian
approved these changes
Oct 26, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.