Refresh model card: license chain + DISCLOSURE bandeau v2

README.md

@@ -0,0 +1,107 @@
+---
+license: cc-by-sa-4.0
+base_model: mistralai/Devstral-Small-2-24B-Instruct-2512
+library_name: peft
+tags:
+- mlx
+- lora
+- peft
+- ailiance
+- devstral
+- kicad-pcb
+language:
+- en
+- fr
+pipeline_tag: text-generation
+---
+
+# Ailiance — Devstral-Small-2-24B-BF16 kicad-pcb (fullseq) LoRA
+
+LoRA adapter fine-tuned on `mistralai/Devstral-Small-2-24B-Instruct-2512` for **kicad-pcb** tasks.
+
+> **Variant**: trained with full-sequence loss for stronger schema adherence.
+
+> Maintained by **Ailiance** — French AI org publishing EU AI Act aligned LoRA adapters and datasets.
+
+> ## ATTRIBUTION AUDIT COMPLETED
+>
+> The training dataset
+> [`Ailiance-fr/mascarade-kicad-dataset`](https://huggingface.co/datasets/Ailiance-fr/mascarade-kicad-dataset)
+> went through a full Stack Exchange attribution audit (2026-05-11):
+> 61 samples (~2.3%) carry per-sample URL+author+post_id attribution;
+> 169 samples flagged `not_found_on_se` (likely synthetic);
+> 2 413 samples (~91%) are LLM-synthetic.
+> Audit report: `docs/audit_mascarade_se_attribution.md` in `electron-bench`.
+
+## Quick start (MLX)
+
+```python
+from mlx_lm import load, generate
+
+model, tokenizer = load(
+    "mistralai/Devstral-Small-2-24B-Instruct-2512",
+    adapter_path="Ailiance-fr/devstral-kicad-pcb-fullseq-lora",
+)
+
+print(generate(model, tokenizer, prompt="..."))
+```
+
+## Training
+
+| Hyperparameter   | Value                  |
+|------------------|------------------------|
+| Base model       | `mistralai/Devstral-Small-2-24B-Instruct-2512`     |
+| Method           | LoRA via `mlx-lm`      |
+| Rank             | 16            |
+| Scale            | 2.0           |
+| Alpha            | 32           |
+| Max seq length   | 16384  |
+| Iterations       | 1000           |
+| Optimizer        | Adam, LR 1e-5          |
+| Hardware         | Apple M3 Ultra 512 GB  |
+
+## Training data lineage
+
+| Role            | Dataset                                                                                          | License        |
+|-----------------|--------------------------------------------------------------------------------------------------|----------------|
+| Primary corpus  | [`Ailiance-fr/mascarade-kicad-dataset`](https://huggingface.co/datasets/Ailiance-fr/mascarade-kicad-dataset)                          | cc-by-sa-4.0 |
+
+For per-sample provenance and attribution status, consult the dataset card.
+
+## License chain
+
+| Component                         | License           |
+|-----------------------------------|-------------------|
+| Base model (`mistralai/Devstral-Small-2-24B-Instruct-2512`)        | apache-2.0    |
+| Training data ([`Ailiance-fr/mascarade-kicad-dataset`](https://huggingface.co/datasets/Ailiance-fr/mascarade-kicad-dataset))         | cc-by-sa-4.0      |
+| **LoRA adapter (this repo)**      | **cc-by-sa-4.0**|
+
+_Most restrictive license in the chain (CC-BY-SA-4.0 share-alike) propagates to derivatives._
+
+## EU AI Act compliance
+
+- **Article 53(1)(c)**: training data licenses preserved (per-dataset cards declare upstream licenses).
+- **Article 53(1)(d)**: training data summary — see upstream dataset cards on Ailiance-fr.
+- **GPAI Code of Practice (July 2025)**: base `mistralai/Devstral-Small-2-24B-Instruct-2512` released under apache-2.0.
+- **No web scraping by Ailiance**, **no licensed data**, **no PII**.
+- Upstream Stack Exchange content (where applicable) is CC-BY-SA-4.0 and propagates to this adapter.
+
+## License
+
+LoRA weights: **cc-by-sa-4.0** — see License chain table above for derivation rationale.
+
+## Citation
+
+```bibtex
+@misc{ailiance_devstral_kicad_pcb_fullseq_2026,
+  author    = {Ailiance},
+  title     = {Ailiance — Devstral-Small-2-24B-BF16 kicad-pcb (fullseq) LoRA},
+  year      = {2026},
+  publisher = {Hugging Face},
+  url       = {https://huggingface.co/Ailiance-fr/devstral-kicad-pcb-fullseq-lora}
+}
+```
+
+## Related
+
+See the full [Ailiance-fr LoRA collection](https://huggingface.co/Ailiance-fr).