Additional KL-divergence statistics #5081
Merged
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
JohannesGaessler
approved these changes
Jan 22, 2024
|
|
||
| std::sort(kld_values.begin(), kld_values.end()); | ||
|
|
||
| printf("===== KL-divergence statistics\n"); |
There was a problem hiding this comment.
Suggested change
| printf("===== KL-divergence statistics\n"); | |
| printf("===== KL-divergence statistics =====\n"); |
Was this intentional? I usually see prints like this.
examples/perplexity/perplexity.cpp
Outdated
Comment on lines
1763
to
1768
| const int n_1percent = nearest_int(0.01f*kld_values.size()); | ||
| printf("KLD_01 : %10.6f\n", kld_values[n_1percent]); | ||
| printf("KLD_99 : %10.6f\n", kld_values[kld_values.size()-1-n_1percent]); | ||
| const int n_5percent = nearest_int(0.05f*kld_values.size()); | ||
| printf("KLD_05 : %10.6f\n", kld_values[n_5percent]); | ||
| printf("KLD_95 : %10.6f\n", kld_values[kld_values.size()-1-n_5percent]); |
There was a problem hiding this comment.
This is not the correct way to calculate percentiles (due to integer rounding errors) but if the number of tokes is >> 100 it should be fine.
|
I am unable to get it to print anything. It simply exits without evaluating.
Regenerating it with or without GPU layers did not resolve the issue. Removing --kl-divergence causes:
This issue may be Windows specific. EDIT: On WSL I get a segfault. |
jordankanter
pushed a commit
to jordankanter/llama.cpp
that referenced
this pull request
Feb 3, 2024
* perplexity: add top-token probability * perplexity: add additional KL-divergence statistics * perplexity: a better organized KL-divergence statistics output --------- Co-authored-by: Iwan Kawrakow <[email protected]>
hodlen
pushed a commit
to hodlen/llama.cpp
that referenced
this pull request
Apr 1, 2024
* perplexity: add top-token probability * perplexity: add additional KL-divergence statistics * perplexity: a better organized KL-divergence statistics output --------- Co-authored-by: Iwan Kawrakow <[email protected]>
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.
I have added:
I'll keep it as a draft PR for a bit.
Please express your wishes for additional things one may want to know.
Here is an example of what you currently get: