Conversation
One image per corpus, generated by codecbench's profile_corpus.py and served from that repository: byte-value distribution, order-0 and order-1 entropy beside deflate reference rates, byte coverage by match distance and matched bytes by match length from a greedy 4-byte parse over sampled 32K-history windows, and per-file size, printable share, and most common byte share. Assisted-By: Claude Fable 5 <noreply@anthropic.com>
nmoinvaz
force-pushed
the
docs/corpus-profiles
branch
from
September 7, 2026 23:26
cf24b39 to
7ce0f94
Compare
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Adds a Profiles section to the README with one image per corpus, generated by codecbench's
profile_corpus.pyand served from that repository.Each profile shows the byte-value distribution, order-0 and order-1 entropy beside deflate reference rates, byte coverage by match distance and matched bytes by match length from a greedy 4-byte parse over sampled 32K-history windows, and per-file size, printable share, and most common byte share. The corpora's characters compare at a glance, silesia's x-ray is 73% literals with only short matches while nci is 3% literals, and the images track light and dark mode.