add different GNN models #60
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
trsvchn
reviewed
Mar 4, 2024
flexynesis/modules.py
Outdated
Comment on lines
340
to
343
| self.fc1 = nn.Linear(input_dim, hidden_dim).to(self.device) | ||
| self.bn_ff1 = nn.BatchNorm1d(hidden_dim).to(self.device) | ||
|
|
||
| self.fc2 = nn.Linear(hidden_dim, output_dim).to(self.device) |
There was a problem hiding this comment.
What is the motivation of using fully connected layers here?
flexynesis/modules.py
Outdated
Comment on lines
391
to
394
| x = self.activation(self.bn_ff1(self.fc1(x))) | ||
| x = self.dropout(x) | ||
|
|
||
| out = self.fc2(x) |
| @@ -261,6 +432,7 @@ def forward(self, x, edge_index, batch): | |||
| Returns: | |||
| Tensor: The output tensor after processing through the GCNN, with shape [num_nodes, output_dim]. | |||
| """ | |||
| #print('batch:', batch.x) | |||
There was a problem hiding this comment.
nit
Suggested change
| #print('batch:', batch.x) |
- reorder, cleanup code - remove device management from the module - set number_layers, dropout, activation function as hyperparameters - make convolution type an option exposed to DirectPredGCNN main class - add GraphConv as an additional convolution option
…ution from CLI; make sanity checks
trsvchn
reviewed
Mar 4, 2024
Comment on lines
+357
to
+359
| x = self.dropout(self.activation(batch_norm(conv(x, edge_index)))) | ||
| else: | ||
| x = self.dropout(self.activation(batch_norm(conv(x, edge_index)))) |
There was a problem hiding this comment.
Dropout will damage the embeddings during training step no?
trsvchn
reviewed
Mar 4, 2024
| x = self.dropout(self.activation(batch_norm(conv(x, edge_index)))) | ||
| else: | ||
| x = self.dropout(self.activation(batch_norm(conv(x, edge_index)))) | ||
| return x |
There was a problem hiding this comment.
why don't you aggregate here? like in forward
There was a problem hiding this comment.
again DO will damage the embeddings during training
@borauyar
potential source of problem with new GNN training
There was a problem hiding this comment.
We don't actually use the extract_embeddings function; the embeddings are extracted from the forward function, which has the same lines. You think we need to disable dropout in the forward?
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.
Hey @trsvchn,
I added new GNN architectures in this branch.