(Re)quantize existing QKeras model with model_quantize #19
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
|
Hello @thesps, thank you for the PR! Could you also add file to demonstrate your usage and tests? I want more information to review this PR. |
1 similar comment
|
Hello @thesps, thank you for the PR! Could you also add file to demonstrate your usage and tests? I want more information to review this PR. |
|
Hi, there are conflicts in utils.py. Please solve if you want to proceed. thanks! |
# for free
to join this conversation on GitHub.
Already have an account?
# 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Outdated suggestions cannot be applied.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.
Add to
model_quantizethe capability to accept an existingqkerasmodel and return a model with a differentquantizer_config.This could be used, for example, to train at higher precision then initialize a lower precision model with
transfer_weights=Trueto speedup convergence of the lower precision model.