refactor: check in pre-split AILuminate benchmark CSVs with reserve-set columns - #9474
Merged
Merged
Conversation
…et columns The gemma enclave notebooks downloaded the MLCommons AILuminate demo prompt set at runtime, then split it into mock and private halves. Instead, check in the pre-split CSVs under notebooks/enclave/gemma/data and document their provenance. Columns now match the real AILuminate reserve prompt set: release_prompt_id is renamed to prompt_uid, persona and prompt_hash are dropped, and the order is prompt_uid, hazard, locale, prompt_text. The job code reads row["prompt_uid"] accordingly.
|
Check out this pull request on See visual diffs & provide feedback on Jupyter Notebooks. Powered by ReviewNB |
The notebooks can be run from any directory, so ignore the downloaded copies by filename pattern rather than by path, and re-include the checked-in canonical copies under notebooks/enclave/gemma/data.
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
The gemma enclave notebooks downloaded the MLCommons AILuminate demo prompt set at runtime, then split it into mock and private halves. This checks the pre-split CSVs into the repo instead.
Changes
notebooks/enclave/gemma/data/withsafety_prompts_mock.csv(5 rows) andsafety_prompts.csv(5 rows), plus aREADME.mdrecording provenance. Both are a deterministic 10-row sample (first prompt per hazard) of the MLCommons AILuminate demo prompt set — a hazard spread rather than the previousrows[:5]/rows[5:10], which were allcse.release_prompt_id→prompt_uid,personaandprompt_hashdropped, orderprompt_uid, hazard, locale, prompt_text. Both notebooks document this divergence from the MLCommons file.colab/2. DO-benchmark-owner-gemma-restrict.ipynb: replaces the download-and-split code with two!curlcalls againstraw.githubusercontent.com/.../gemma/data.dev/1. enclave_gemma_inmem_restrict_ailumniate.ipynb: reads the CSVs from the checkout via../data(no network), staging copies into temp dirs so the repo files stay untouched.row["prompt_uid"]instead ofrow["release_prompt_id"].Test plan
prompt_uid/prompt_textnon-empty.!curlpath through IPython (against a local URL, since the CSVs are not ondevyet) and the dev../datapath in place — both load 5 mock + 5 private rows and pass the column assertion.pre-commitclean on the changed files.Note: the colab notebook's
DATA_URLpoints atrefs/heads/dev, so its!curlcells only work once this is merged.