faysalbenahmed/AMF-v0.18-Qualified-Realization

AMF v0.18 — Qualified Hybrid Intelligence Realization

Status: QUALIFIED_INTELLIGENCE_REALIZATION under the AMF internal qualification protocol.
Selected realization: g02-reuse-g01-canonicalize-historical-be
Mission: explicit_evidence_sufficiency_v5_cumulative_scientist

This repository is a sanitized public release of the first AMF v0.18 realization that reached the terminal AMF verdict QUALIFIED_INTELLIGENCE_REALIZATION. It is not a claim of external certification.

The qualified object is not the LoRA alone. It is the composed realization:

Pinned Mistral-7B-Instruct-v0.3
+ exact historical LoRA adapter
+ CANONICALIZE_TO_RECORDS_V1
+ evidence-sufficiency prompt/output contract
+ inference runtime

AMF explored multiple intervention loci (continued LoRA adaptation, fresh adaptation, reuse, representation canonicalization, and a substrate switch). The final qualified realization retained the representation-canonicalized fabric rather than simply selecting the highest raw research score.

Public qualification results

Role	Parent accuracy	Qualified accuracy	Gain	Parent macro-F1	Qualified macro-F1	Schema valid
ROB	71.88%	90.63%	+18.75 pts	73.03%	90.63%	100%
OOD	50.00%	85.16%	+35.16 pts	35.92%	85.05%	100%
Fresh Final	47.66%	81.25%	+33.59 pts	39.50%	81.18%	100%

Each role above contains 128 evaluated samples. The exact aggregate measurement receipts are included under evidence/. Protected rows, Fresh Final rows, their seeds, and protected/final prediction files are deliberately not published.

Fresh Final is a bounded generalization test over supported representation families. It should not be interpreted as evidence of universal generalization to arbitrary unseen formats or domains.

Why this result matters

The strongest result is not the number 81.25% by itself. The campaign produced a coherent evidence chain across execution, integrity, scientific comparison, protected evaluation, candidate freeze and final qualification.

AMF also rejected a G2 continued-LoRA candidate that had a higher absolute SEARCH_CONFIRM score because its incremental improvement did not satisfy the configured confirmation rule. This is evidence that AMF does not simply choose the largest raw score; it selects among realizations according to the mission and qualification contracts.

Exact realization identity

• Base model: mistralai/Mistral-7B-Instruct-v0.3
• Pinned revision: c170c708c41dac9275d15a8fff4eca08d52bab71
• Adapter tree SHA-256 (AMF tree identity): 36c1f91df36aa00802ea65d721657b5a1af699f23b0beaddf9d95e3afd741145
• Adapter weights SHA-256: 88094b7f6a88586eb5201efa7051f10232602c618d3eb8e34046f18bb16dcb71
• Adapter size: 6,839,183 bytes (tree total recorded by AMF)
• Input transform: CANONICALIZE_TO_RECORDS_V1
• Canonicalizer source SHA-256: 5f32ab909812fa239233be764cfdc6286ea33197314108c2670c8b5ef5f03af3
• Qualified identity receipt: 5ed5db3b97bc011a19a66243f1accbc89db1c4afc1da4925c7c29317ff5c553e
• Qualification verdict receipt: 8949f3af288ee1ffa16a01e04f66467f7726414c26a9af5a3d84d9e569bb552e

qualified_adapter_exact/ contains the adapter files exactly as present in the qualified lineage. The original adapter_config.json intentionally retains its historical local cache path; the convenience runtime loads the pinned base model explicitly and then attaches the adapter.

Reproducibility / provenance

• Source mission package SHA-256: 3571f1f7b934e9e8cd5cafd2b021e96765a4adbe576c1072a5605f395c9aa030
• Canonical result archive SHA-256: de60ddb8b5dfc4061aa285c132d89b6fdbd168b508f81d419af1ce358c435493
• Ledger events: 92
• Ledger head SHA-256: 054311b729e4ab8b0d8d53cf74ae76a9ce5262dab80fdd0efc87898d2344d819
• Run state: SUCCEEDED
• Integrity state: VALID
• Qualification Authority verdict: QUALIFIED_INTELLIGENCE_REALIZATION

Machine-readable provenance is in AMF_EVIDENCE.json. Qualification contracts and exact public evidence receipts are included without disclosing protected rows.

The original RUN_RESULT.json contains publication_ready=false because the campaign's built-in gated Hugging Face exporter was not executed during the run. The qualification verdict itself is independent of that publication workflow and is preserved unchanged in evidence/QUALIFICATION_VERDICT.json.

Quick start

Install dependencies:

pip install -r runtime/requirements.txt

Run the convenience inference helper:

python runtime/inference.py examples/example_input.json

The base model is fetched from the pinned Hugging Face revision. A GPU suitable for Mistral 7B is expected for this convenience runner.

Important: runtime/inference.py is a post-campaign convenience helper. The qualification evidence is bound to the original AMF evaluation runtime recorded by the campaign; this helper is not itself the qualification authority.

Repository layout

qualified_adapter_exact/   exact qualified lineage adapter
canonicalizer/             exact CANONICALIZE_TO_RECORDS_V1 source
runtime/                   public inference helper + exact prompt/parser source
contracts/                 mission, qualification and operation contracts
evidence/                  aggregate receipts and qualification/freeze records
examples/                  synthetic public example
release/                   portable public bundle ZIP
AMF_EVIDENCE.json          machine-readable public provenance
MANIFEST.sha256            SHA-256 of every published file

What is intentionally not published

This repository does not contain protected dataset rows, Fresh Final rows, protected/Fresh Final seeds, protected/final prediction files, or the raw 170 MB result archive. Those remain private so the original evidence is not casually exposed or reused as if it were fresh protected evidence in future campaigns.

Scope and limitations

• QUALIFIED_INTELLIGENCE_REALIZATION is an AMF protocol verdict, not an external third-party certification.
• The release demonstrates bounded performance on this mission and its supported representation families; it does not imply universal reliability.
• The hashes establish integrity and lineage. They do not, by themselves, prove the correctness of every scientific design choice.
• The realization still uses Mistral 7B; this is not yet a demonstration of edge deployment or micro-model inference.
• No GDPR/RGPD compliance claim is made by this release.

License / third-party components

The Mistral base model weights are not included in this repository and remain subject to their upstream terms. This public AMF release is marked license: other because no broader AMF license is asserted here. Do not infer additional rights beyond the applicable upstream licenses and the rights granted by the repository owner.

Français — résumé

v0.18 démontre qu'AMF peut explorer plusieurs loci d'intervention, sélectionner une réalisation hybride admissible, la figer avant l'évaluation protégée, puis atteindre un verdict terminal QUALIFIED_INTELLIGENCE_REALIZATION. La publication reste volontairement assainie : les lignes protégées, le Fresh Final brut et leurs seeds ne sont pas exposés.