Repository navigation
feat(claims): check the voice's reply against the claims register - #8
Merged
Merged
Conversation
…OSIRIS_ON_QPU The register was checked against the user's message only. Now every mentor reply is checked too: a not-legit (or evidence-free) claim the voice brings up that the message did not is shown under the reply with its verdict and /legit link, and the exchange is not learned (ledger row: learnable false, flagged_claims). A reference is not an endorsement, as with match(): the flag keeps the reply out of training; it does not say the reply was wrong. New entry OSIRIS_ON_QPU (OVERCLAIM): OSIRIS's model running on, trained on or integrated with quantum hardware -- three sentences qwen2.5:1.5b wrote on 2026-11-09. The core is a NumPy transformer trained on the local CPU; hardware runs are separate recorded experiments. Full suite: 1792 passed, 13 skipped, 2 failed (the same two host-specific failures as main). Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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 claims register checked only the user's message. A small voice could still invent a claim on its own, and OSIRIS would show it and train on it. On 2026-11-09
qwen2.5:1.5bwrote "OSIRIS operates on a quantum computer with 156 qubits".claims.flagged_in_reply(): claims a reply brings up, minus those the message already raised, whose standing is not legit / not supported / not legit as worded, orNO_EVIDENCE. Pilots and untested proposals are not flagged.living.converse: showsclaims.reply_block()under the reply. The exchange is not learned. The ledger row recordslearnable: false, flagged_claims: [...].OSIRIS_ON_QPU(OVERCLAIM), with evidenceosiris_cli/train.pyand the Heron r2 campaign DOI.As with
match(), a reference is not an endorsement: a reply that says a claim was refuted is flagged too. The flag shows the verdict and keeps the reply out of training. It does not say the reply was wrong.Tests: 11 new (scripted mentor, no Ollama). Full suite locally: 1792 passed, 13 skipped, 2 failed. Both failures are host-specific and also fail on
main.🤖 Generated with Claude Code