tests(qwen36): provenance for the tensor-kind lists -- index hashes, the regenerating one-liner, an index-backed check - #1613
Merged
Conversation
…the regenerating one-liner, an index-backed check Asked for in the JustVugg#1326 review: the name lists were transcribed by hand from the two real model.safetensors.index.json files and the contract rests on them. The header now carries the sha256 of both indexes and the one line that regenerates a collapsed list from an index; IndexBackedTest runs the lists against the real files when QWEN36_INDEX_DIR points at them (skipped in CI, where the indexes are not in the tree). For the 2.4T index the collapsed output is QWEN38_2P4T exactly; for the 35B index it is the same 46 names under model.language_model. plus 21 model.visual.* kinds. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01X2HPETYXNwT4RmQkdMMyPS
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
2 participants
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Follow-up to #1326, the auditability request from the review: the name lists in
tests/test_qwen36_tensor_kinds.pywere transcribed by hand from the two realmodel.safetensors.index.jsonfiles, and the contract rests on them.e36c40d4…28ee, Qwen/Qwen3.6-35B-A3B41b93561…be83) and the one line that regenerates a collapsed list from an index (re.sub(r"\.(\d+)\.", ".N.", name)overweight_map, sorted, unique).IndexBackedTestruns the lists against the real files whenQWEN36_INDEX_DIRpoints at them; skipped in CI, where the indexes are not in the tree. For the 2.4T index the collapsed output isQWEN38_2P4T(46 names). For the 35B index it is the same 46 undermodel.language_model.plus 21model.visual.*kinds (the 333 vision tensors collapse to 21; the three listed are among them, the prefix is what the contract keys on).Verified here with both files: 9 tests, 0 skipped, clean under
-W error::ResourceWarning; without the files: 7 run, 2 skipped, as CI will see it.Thanks to Claude Code for the help along the way.
https://claude.ai/code/session_01X2HPETYXNwT4RmQkdMMyPS