Repository navigation
bench: report llama-perplexity failures instead of a silent row of "-… - #146
Merged
Merged
Conversation
…-", never keep a partial KL base FrogNano, 2026-10-05: the box's disk filled while writing the KL base, so it stopped at chunk 109. Every rung then scored 109 chunks, failed reading chunk 109, and llama-perplexity -- which prints its summary only after the last chunk -- printed none. pollard-bench recorded null PPL, KLD and top-1 for three rungs and exited 0 with no reason. _ppl_kl now prints llama-perplexity's own error lines (exit code + the failing line) whenever it exits non-zero or yields no metrics. Building the KL base, a non-zero exit or a disk left under 1 GB free (a full disk can still exit 0) deletes the partial base instead of keeping it for every later run to trip over.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
…-", never keep a partial KL base
FrogNano, 2026-10-05: the box's disk filled while writing the KL base, so it stopped at chunk 109. Every rung then scored 109 chunks, failed reading chunk 109, and llama-perplexity -- which prints its summary only after the last chunk -- printed none. pollard-bench recorded null PPL, KLD and top-1 for three rungs and exited 0 with no reason.
_ppl_kl now prints llama-perplexity's own error lines (exit code + the failing line) whenever it exits non-zero or yields no metrics. Building the KL base, a non-zero exit or a disk left under 1 GB free (a full disk can still exit 0) deletes the partial base instead of keeping it for every later run to trip over.