Repository navigation
fix(engine-vector): re-pin the TinyLlama anchor to vitni-regime-2 - #2
Merged
Merged
Conversation
The engine moved from vitni-regime-1 to vitni-regime-2 (this run's v2 model_digest went ffebe862... -> 7a2e28c9...) without the conformance vector following. The reproducer requires the engine's reported regime to match the vector (engine.py), so against the current vitni-receipt binary the vector was rejected and silently reported SKIPPED rather than reproduced -- the engine tier had no passing vector. Re-pin vectors/engine.json to vitni-regime-2 / 7a2e28c9...; the frozen tier-1 v1 digest 9c0754... is unchanged across the regime move. Update the README anchor and add CHANGELOG 0.3.1. Re-verified end-to-end against the vitni-receipt binary: the TinyLlama vector reproduces bit-for-bit (1 reproduced); receipt tier still 16/16.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 engine moved from `vitni-regime-1` to `vitni-regime-2` — this run's v2 `model_digest` went `ffebe862…` → `7a2e28c9…` — without the conformance vector following.
Why this matters: the engine reproducer (`engine.py`) requires the engine's reported `regime` to equal the vector's. So against the current `vitni-receipt` binary the regime-1 vector was rejected and silently reported SKIPPED — the engine tier had no passing vector at all, while the README/CHANGELOG still claimed "1 reproduced".
Change
Verification — re-run end-to-end against the `vitni-receipt` binary (TinyLlama-1.1B-Chat Q4_K_M, "Once upon a time,", n=20): the vector reproduces bit-for-bit —
1 reproduced, 0 skipped; receipt tier still 16/16. The regime-2 digest was also reproduced byte-identically on arm64 and x86_64.Follow-up (separate): the drift happened because no check ties the committed anchors to what the engine currently emits — worth an opt-in CI guard.