Downloads · 30 days
201
46% of all-time downloads
SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-v3
szl-receiptagent-qwen35-0.8b-v3 is a text generation model from SZLHOLDINGS. Use it when you need the model to write or continue text. It is set up for peft. The card lists the license as apache-2.0.
🜂 SZL Holdings estate artifact — the estate that receipts its own thinking.
Downloads · 30 days
201
46% of all-time downloads
All-time downloads
437
Public
Parameters
752M
6.1 GB on disk
Likes
1
Public
Click a slice to open those files.
.safetensors3 GB · 99%
From the Hugging Face model README
🜂 SZL Holdings estate artifact — the estate that receipts its own thinking.
<!-- szl:artifact-identity-reconciled --><!-- SZL-ESTATE-CARD:v2:START --> <p align="center"><a href="https://a-11-oy.com/"><img src="https://huggingface.co/spaces/SZLHOLDINGS/README/resolve/main/assets/estate-banner-v2.svg" alt="SZL Holdings — governed, receipted, verifiable" width="100%"></a></p>Artifact identity (noted 2026-11-12). Besides the LoRA adapter, this repository's root carries a merged full-precision checkpoint (
model.safetensors+config.json, loadable withtransformers) produced by the receipted CPU merge of this adapter into its declared base (merge_receipt.json). The adapter remains the artifact of record for every figure on this card; the merged bytes carry no separate held-out receipt and add no claim. Metadata-only note.
Control before action. Evidence after.
Product · Proof · GitHub · HF org
<!-- SZL-ESTATE-CARD:v2:END -->Canonical family model — proposal-only receipt-drafting LoRA (bf16, r=16, α=32) on Qwen/Qwen3.5-0.8B + derived merge. Non-release placeholder (weights present; promotion false).
| Field | Value |
|---|---|
| Status | NON-RELEASE · publication_eligible=false |
| Train loss | 0.0537 MEASURED (train metric ≠ eval) |
| License | Apache-2.0 |
| Doctrine | v11 LOCKED · Λ = Conjecture 1 (advisory) |
Does not overwrite v2 or SZL-Forge-1.5B-ReceiptAgent.
Drafts / recovers / refuses receipt-shaped outputs under a proposal-only contract. Stamp that never mints authority.
*-candidate / *-laptop-lora / *-authenticated confusion idsload_in_4bit=false)| Dataset / source | Role |
|---|---|
Committed v3 train.jsonl (180 rows; SHA in training receipt) | Primary train rows (receipt-stated) |
SZLHOLDINGS/governed-receipts-bench | Bench adjacency |
SZLHOLDINGS/szl-training-receipts | Training receipt corpus |
SZLHOLDINGS/receipted-unsloth | Unsloth receipt adjacency |
SZLHOLDINGS/governed-agent-bench | Agent-bench adjacency |
Dataset SHA-256 (receipt-stated): ad30ce9b478eff50d78064f387143b750be3f746b459f98f7961784b9ce1081f.
| Suite | Result | Class |
|---|---|---|
| Historical DEV n=12 (2026-08-29) | 11/12 (4/4 draft · 3/4 recovery · 4/4 refuse) | MEASURED (DEV; test.jsonl never opened) |
| Additive adapter-bound DEV n=12 (2026-09-29) | 12/12 | MEASURED · base UNRECORDED · promotion_effect=NONE |
| Adapter SHA Hub binding | — | UNVERIFIED vs receipt-stated local SHA |
| Release / autonomy | false | MEASURED policy |
| Field | Status |
|---|---|
| Train energy | UNAVAILABLE (laptop Blackwell run recorded; no attested kWh meter on card) |
| BOM | bom/model-bom.cdx.json present — packaging |
| DSSE | Training receipt unsigned — honest UNSIGNED, not fabricated PASS |
from peft import PeftModel
from transformers import AutoModelForCausalLM, AutoTokenizer
base_id = "Qwen/Qwen3.5-0.8B"
base_rev = "<BASE_PIN>"
adapter = "SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-v3"
adapter_rev = "<PINNED_SHA>"
tok = AutoTokenizer.from_pretrained(base_id, revision=base_rev)
base = AutoModelForCausalLM.from_pretrained(base_id, revision=base_rev, device_map="auto")
model = PeftModel.from_pretrained(base, adapter, revision=adapter_rev)
SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-v3training_receipt.laptop-blackwell.json, eval_dev_named_n.json, merge_receipt.jsonStephen Lutar / SZL Holdings
Thank you NVIDIA NeMo/NIM, Unsloth, Hugging Face Hub, vLLM / PagedAttention, and FlashAttention authors. SZL builds clean-room silhouettes with DSSE receipts and MEASURED-or-null energy (never fabricates joules, signatures, or green evals).
Apache-2.0. See LICENSE in-repo when present.
NVIDIA-style card upgrade · wave2 2026-11-04T18:00Z · clean-room silhouette (DSSE receipts + MEASURED-or-null energy).
Does not authorize train, serve, or autonomy. Does not overwrite sibling SKUs.
| Surface | Link |
|---|---|
| Company | szlholdings.com |
| Sponsors | Become a sponsor |
| GitHub | szl-holdings |
| Receipts chain | frontier-54-runs |
| Live dashboard | SI command center |
The listed weight files in this repository are byte-identical (same LFS SHA-256; checkpoint-* excluded) to files in another repository: SZLHOLDINGS/szl-receiptagent-qwen35-0.8b-v3-laptop-lora@7c0c0cacd3dc (shared: adapter_model.safetensors) (checked 2026-11-12 at 73fbb0832128). Identical bytes are one artifact published under more than one name; this repository is not independent evidence of a separate training run, and any results stated for it apply to that same artifact. Existing result lines above are left unchanged.