At text classification example #945
Open
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
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
l2 1e-6 > 1e-3 Signed-off-by: atuzhykov <[email protected]>
l2 1e-3 > 1e-6 Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
) Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: atuzhykov <[email protected]>
…r1e-4 Signed-off-by: atuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
Signed-off-by: Andrii Tuzhykov <[email protected]>
AlexDBlack
suggested changes
Mar 11, 2020
There was a problem hiding this comment.
Looking good, just a few minor improvements to make.
Can we also make a backup of the branch, then flatten + sign on this branch as described here: https://deeplearning4j.org/eclipse-contributors
Otherwise I'm happy with this 👍
| * As far model is predisposed to overfitting we also add l2 regularization and dropout for certain layers. | ||
| * To prepare reviews we use BertIterator, which is MultiDataSetIterator for training BERT (Transformer) models. | ||
| * We congigure BertIterator for supervised sequence classification: | ||
| * 0. As tokenizer we use BertWordPieceTokenizerFactory with provided BERT BASE UNCASED vocabulary. |
There was a problem hiding this comment.
Maybe let's improve this slightly, add another line under 0.:
BertIterator and BertWordPieceTokenizer implement the Word Piece sub-word tokenization algorithm, with a vocabulary size of 30522 tokens.
| int listenerFrequency = 20; | ||
| net.setListeners(new StatsListener(statsStorage, listenerFrequency), new ScoreIterationListener(50)); | ||
| //Attach the StatsStorage instance to the UI: this allows the contents of the StatsStorage to be visualized | ||
| uiServer.attach(statsStorage); |
There was a problem hiding this comment.
Maybe let's comment out the UI by default, as it adds some overhead (slows down training a bit). Users can uncomment it if they want to run it with UI. That would look like this:
/*
//Uncomment this section to run the example with the user interface
UIServer uiServer = UIServer.getInstance();
//Configure where the network information (gradients, activations, score vs. time etc) is to be stored
//Then add the StatsListener to collect this information from the network, as it trains
StatsStorage statsStorage = new FileStatsStorage(new File(System.getProperty("java.io.tmpdir"), "ui-stats-" + System.currentTimeMillis() + ".dl4j"));
int listenerFrequency = 20;
net.setListeners(new StatsListener(statsStorage, listenerFrequency), new ScoreIterationListener(50));
//Attach the StatsStorage instance to the UI: this allows the contents of the StatsStorage to be visualized
uiServer.attach(statsStorage);
*/
net.setListeners(new ScoreIterationListener(50));
| net.fit(train); | ||
|
|
||
| // Get and print accuracy, precision, recall & F1 and confusion matrix | ||
| Evaluation eval = net.doEvaluation(test, new Evaluation[]{new Evaluation()})[0]; |
There was a problem hiding this comment.
For MultiLayerNetwork, we can use net.evaluate(test)
pom.xml
Outdated
| @@ -28,7 +28,7 @@ | |||
| <properties> | |||
| <!-- Change the nd4j.backend property to nd4j-cuda-9.2-platform,nd4j-cuda-10.0-platform or nd4j-cuda-10.1-platform to use CUDA GPUs --> | |||
| <nd4j.backend>nd4j-native-platform</nd4j.backend> | |||
| <!-- <nd4j.backend>nd4j-cuda-10.2-platform</nd4j.backend>--> | |||
| <!-- <nd4j.backend>nd4j-cuda-10.0-platform</nd4j.backend>--> | |||
There was a problem hiding this comment.
Leave this commented out with 10.2
Signed-off-by: Andrii Tuzhykov <[email protected]>
…ation class Signed-off-by: Andrii Tuzhykov <[email protected]>
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.
What changes were proposed in this pull request?
(Please fill in changes proposed in this fix)
How was this patch tested?
(Please explain how this patch was tested. E.g. unit tests, integration tests, manual tests)
Please review
https://github.com/eclipse/deeplearning4j/blob/master/CONTRIBUTING.md before opening a pull request.