Added error message for prompt-based loss and num_variants=1 #120
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
|
Ah it looks like torch.var by default computes the sample variance, which divides by n-1, rather than n. The previous code was dividing by n so returned the equivalent of CCS loss. But your solution seems fine too, since we don't really care about being able to run prompt_var with num_variants=1 |
|
@AlexTMallen yeah you can also do |
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Outdated suggestions cannot be applied.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.
The loss functions ccs_prompt_var and prompt_var_squared cause NaN errors when --num_variants is set to 1, because of the variance calculation.
I added an error message in train.py for this.
As ccs_prompt_var and num_variants=1 are set by default, this means an error is thrown by default, which is not ideal.