Skip to content
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

[onert-micro] Add GRU backward execution #13757

Merged
merged 1 commit into from
Aug 28, 2024

Conversation

BalyshevArtem
Copy link
Contributor

This pr adds GRU backward execution.

from #13737

ONE-DCO-1.0-Signed-off-by: Artem Balyshev <[email protected]

@chunseoklee

This comment was marked as resolved.

This pr adds GRU backward execution.

ONE-DCO-1.0-Signed-off-by: Artem Balyshev <[email protected]
@BalyshevArtem BalyshevArtem force-pushed the second_gru_training_pr branch from aa9e945 to 6571003 Compare August 27, 2024 09:33
@BalyshevArtem
Copy link
Contributor Author

@BalyshevArtem I have failed to train the model(generated by #13365 (comment) ) with the data, which were used for previous target model :

Need to change number of trainable layers. This pr support only weights gradients calculation, and input gradients calculation doesn't currently support. So, gru should be last layer for training.
For this pr use num_train_layers = 4 in TrainingDriver.cpp. I updated it in current TrainingDriver, please check.

Copy link
Contributor

@SlavikMIPT SlavikMIPT left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

LGTM

@chunseoklee chunseoklee merged commit 523d27b into Samsung:master Aug 28, 2024
4 checks passed
@@ -263,11 +264,11 @@ int entry(int argc, char **argv)
if (CLASSIFICATION_TASK)
{
// Evaluate cross_entropy and accuracy metrics
train_interpreter.evaluateMetric(onert_micro::CROSS_ENTROPY_METRICS,
train_interpreter.evaluateMetric(config, onert_micro::CROSS_ENTROPY_METRICS,
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We also need to introduce these changes to tests for target board

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants