Fix LSTM training continuity for cloned nets #951
Conversation
This file contains hidden or 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
Fixes BrainJS#949 Update `src/recurrent.ts` to ensure cloned LSTM nets continue training from the point where the original stopped. * Add `fromJSON` method to properly restore the training state. * Modify `train` method to account for the state of the cloned net. * Update `trainPattern` method to consider the previous training state of the cloned net. * Adjust `initialize` method to handle state restoration for cloned nets. * Ensure `runInputs` method maintains continuity in training for cloned nets. Add a test case in `src/recurrent/lstm.test.ts` to verify that training a cloned LSTM net continues evolving from the point where the original stopped.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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Outdated suggestions cannot be applied.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.
Fixes #949
Update
src/recurrent.tsto ensure cloned LSTM nets continue training from the point where the original stopped.fromJSONmethod to properly restore the training state.trainmethod to account for the state of the cloned net.trainPatternmethod to consider the previous training state of the cloned net.initializemethod to handle state restoration for cloned nets.runInputsmethod maintains continuity in training for cloned nets.Add a test case in
src/recurrent/lstm.test.tsto verify that training a cloned LSTM net continues evolving from the point where the original stopped.