🜂 SZL Holdings estate artifact — the estate that receipts its own thinking.
A 0.8B fine-tune on acoustic physics + governance math — pairing the Aero-FF acoustoelastic operating values with the estate's Λ-gate governance semantics, built for the collaboration line with Molecular Audio Canada (Chad Barnard) on falsifiable acoustic-physics predictions.
Artifact: merged 16-bit checkpoint (not adapter-only) · Stage: research · Wave: acoustic-w27 · Status: BASELINE_MEASURED_DELTA_PENDING
Chart data and sources: assets/chart-data.json. Every bar is a count already stated in this card; standard benchmarks are drawn as "not yet evaluated".
Sizes are Hub file sizes on 2026-11-19. They say nothing about quality.
from transformers import AutoTokenizer, AutoModelForCausalLM
tok = AutoTokenizer.from_pretrained("SZLHOLDINGS/SZL-AcousticGovernance-0.8B-v1")
model = AutoModelForCausalLM.from_pretrained("SZLHOLDINGS/SZL-AcousticGovernance-0.8B-v1", torch_dtype="auto")
msgs = [{"role": "user", "content": "<controller-supplied context + request>"}]
inputs = tok.apply_chat_template(msgs, add_generation_prompt=True, return_tensors="pt", return_dict=True)
out = model.generate(**inputs, max_new_tokens=256, do_sample=False)
print(tok.decode(out[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))
In transformers 5.x, apply_chat_template returns a BatchEncoding: pass return_dict=True and **inputs as shown (the older ids.shape[1] form raises AttributeError).
Run check (2026-11-19, transformers 5.23.0, torch 2.15.0 (CPU)): the code path of this snippet (model load, apply_chat_template, generate, decode) ran end to end for the qwen3_5 architecture, using a tokenizer with a chat template and a randomly initialised, shrunken model of that architecture. No weights were downloaded, so it was not run with this repository's weights and output quality is not evaluated. Private or gated repositories need hf auth login first.
| Key | Value |
|---|---|
| Base model | Qwen 0.8B class (szl-receiptagent-qwen35-0.8b-sft-v3) |
| Parameters | 0.8B, merged 16-bit |
| Training corpus | SZLHOLDINGS/acoustic-governance-sft-v1 — 26 examples |
| Method | QLoRA SFT (Unsloth + TRL), merged-16bit push |
| Hardware | NVIDIA RTX A6000 48GB (Massed Compute) |
| Wave log | SZLHOLDINGS/wave-logs (acoustic-w27) |
26 examples across 6 topics: Navier-Stokes regularization pathway (4 falsifiable checks), Schur-concavity prediction (Lean-proved Robin Hood lemma → NS regularization hypothesis), acoustic OAM (Liu et al. 2025), Aero-FF operating values, and the Λ-gate governance semantics that bind physical claims to proof status.
Status: citation probes NOT_RUN on this model. The estate's honest convention: no eval, no claim. The physics labels in the corpus are the corpus's own honesty contract, not evidence the model learned them.
| Eval | Status | Result |
|---|---|---|
| Citation probes | NOT_RUN | No claim until the receipt exists |
Model output is an untrusted proposal. Consequential actions pass a separate deterministic controller, policy gate (Λ-spine, fail-closed, zero-veto), and human approval boundary. The estate's measured confabulation floor at 0.8B (~3/6 citation probes across three training ablations) is exactly why the gate exists.
vllm serve SZLHOLDINGS/SZL-AcousticGovernance-0.8B-v1 --dtype bfloat16 --max-model-len 4096
No production endpoint serves this model. No x402 monetization route is bound to it.
Apache-2.0 (per base model). Aero-FF material used under conditional verification permission from Chad Barnard (Molecular Audio Canada); Liu et al. 2025 cited per its license.
@misc{{szl-acousticgovernance-08b-v1,
title = {{SZL-AcousticGovernance-0.8B-v1}} — SFT on acoustic physics + governance math,
author = {{Lutar, Stephen P. Jr.}},
year = {{2026}},
url = {{https://huggingface.co/SZLHOLDINGS/SZL-AcousticGovernance-0.8B-v1}}
}}
Related DOI-minted work: Schur-Concavity as a Falsifiable Physics Prediction (10.5281/zenodo.24005485) · Information-Thermodynamic Certification (10.5281/zenodo.24005483)
The evaluation receipts behind the numbers on this card are in
SZLHOLDINGS/model-eval-receipts, which is private as of 2026-11-12, so outside readers cannot open them yet. Until the owner publishes those receipts (or copies them to the public chainSZLHOLDINGS/frontier-54-runs), treat the figures below as not publicly verifiable. The MEASURED labels above are unchanged.
| Receipt (private) | What it records for this model |
|---|---|
fleet-confabulation-sweep-20261108.json |
6-probe confabulation sweep (2026-11-08, greedy): abstained 1, answered 5, verdict FAIL. This is the estate confabulation sweep, not the citation-probe battery marked NOT_RUN above |
Training log: acoustic-governance-w27-train.log in SZLHOLDINGS/wave-logs (private).
The estate published measurement receipts naming this repository in the dataset 'SZLHOLDINGS/model-eval-receipts' (dataset commit '44fd9dac4145a5e9e55f28df50390925d984631f'). Each is mirrored byte-identically under `receipts/eval/` with its sha256 in the table below. Estate doctrine: results publish verbatim, win or lose.
| Receipt | schema | date | Measured headline | sha256 |
|---|---|---|---|---|
| `receipts/eval/fleet-confabulation-sweep-20261108.json` | szl.fleet-confabulation-sweep/v1 | 2026-11-08 | see receipt | `9a16903f6943917a…` |
Verbatim honest notes (excerpted):
6-probe honesty battery (3 answerable + 3 unanswerable), measured on NVIDIA GeForce RTX 4090 (SaladCloud run szl-estate-battery-v2). Measured result: 0 of 6 probes abstained-or-refused. Per-probe verdicts and raw samples in the public receipt receipts/estate-battery-szl_acousticgovernance_0_8b_v1.json in dataset SZLHOLDINGS/frontier-54-runs (run manifest sha256 2de61060e55d55d5...). Honest scores only - no curated claims.
b405c0d05177d48494866a46cadf0f433e25e29f. The prior card remains in revision history.UNAVAILABLE.receipts/eval/fleet-confabulation-sweep-20261108.json.apache-2.0. This edit changed no license and asserts none beyond the observed tag.UNMAPPED). No source-side binding is claimed.07d4d3d80261822da7d38365524d6c45240b45d9. The prior card remains in revision history.UNAVAILABLE.receipts/eval/fleet-confabulation-sweep-20261108.json.apache-2.0. This edit changed no license and asserts none beyond the observed tag.SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-sft-v3