Update in inference.py -- omitting the optimizer's state #165
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
During the inference phase when loading a trained model, there's a focus on exclusively loading the model's weights while disregarding or not loading the optimizer's state. This adjustment can be particularly beneficial when utilizing a pre-trained model for tasks such as inference or transfer learning, where the optimizer's state information might not be necessary.
|
@microsoft-github-policy-service agree |
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Outdated suggestions cannot be applied.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.
During the inference phase when loading a trained model, there's a focus on exclusively loading the model's weights while disregarding or not loading the optimizer's state. This adjustment can be particularly beneficial when utilizing a pre-trained model for tasks such as inference or transfer learning, where the optimizer's state information might not be necessary.