Refresh model card: license chain + DISCLOSURE bandeau v2
Browse files
README.md
ADDED
|
@@ -0,0 +1,93 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
license: apache-2.0
|
| 3 |
+
base_model: mistralai/Devstral-Small-2-24B-Instruct-2512
|
| 4 |
+
library_name: peft
|
| 5 |
+
tags:
|
| 6 |
+
- mlx
|
| 7 |
+
- lora
|
| 8 |
+
- peft
|
| 9 |
+
- ailiance
|
| 10 |
+
- devstral
|
| 11 |
+
- typescript
|
| 12 |
+
language:
|
| 13 |
+
- en
|
| 14 |
+
- fr
|
| 15 |
+
pipeline_tag: text-generation
|
| 16 |
+
---
|
| 17 |
+
|
| 18 |
+
# Ailiance — Devstral-Small-2-24B-Instruct typescript LoRA
|
| 19 |
+
|
| 20 |
+
LoRA adapter fine-tuned on `mistralai/Devstral-Small-2-24B-Instruct-2512` for **typescript** tasks.
|
| 21 |
+
|
| 22 |
+
> Maintained by **Ailiance** — French AI org publishing EU AI Act aligned LoRA adapters and datasets.
|
| 23 |
+
|
| 24 |
+
## Quick start (MLX)
|
| 25 |
+
|
| 26 |
+
```python
|
| 27 |
+
from mlx_lm import load, generate
|
| 28 |
+
|
| 29 |
+
model, tokenizer = load(
|
| 30 |
+
"mistralai/Devstral-Small-2-24B-Instruct-2512",
|
| 31 |
+
adapter_path="Ailiance-fr/devstral-typescript-lora",
|
| 32 |
+
)
|
| 33 |
+
|
| 34 |
+
print(generate(model, tokenizer, prompt="..."))
|
| 35 |
+
```
|
| 36 |
+
|
| 37 |
+
## Training
|
| 38 |
+
|
| 39 |
+
| Hyperparameter | Value |
|
| 40 |
+
|------------------|------------------------|
|
| 41 |
+
| Base model | `mistralai/Devstral-Small-2-24B-Instruct-2512` |
|
| 42 |
+
| Method | LoRA via `mlx-lm` |
|
| 43 |
+
| Rank | 16 |
|
| 44 |
+
| Scale | 2.0 |
|
| 45 |
+
| Alpha | 32 |
|
| 46 |
+
| Max seq length | 2048 |
|
| 47 |
+
| Iterations | 500 |
|
| 48 |
+
| Optimizer | Adam, LR 1e-5 |
|
| 49 |
+
| Hardware | Apple M3 Ultra 512 GB |
|
| 50 |
+
|
| 51 |
+
## Training data lineage
|
| 52 |
+
|
| 53 |
+
Derived from the internal **eu-kiki / mascarade** curation. All upstream samples
|
| 54 |
+
are synthetic, permissively-licensed, or generated from Apache-2.0 base resources.
|
| 55 |
+
See the [Ailiance-fr catalog](https://huggingface.co/Ailiance-fr) for related cards.
|
| 56 |
+
|
| 57 |
+
## License chain
|
| 58 |
+
|
| 59 |
+
| Component | License |
|
| 60 |
+
|-----------------------------------|-------------------|
|
| 61 |
+
| Base model (`mistralai/Devstral-Small-2-24B-Instruct-2512`) | apache-2.0 |
|
| 62 |
+
| Training data (internal Ailiance curation (synthetic + permissive sources)) | apache-2.0 |
|
| 63 |
+
| **LoRA adapter (this repo)** | **apache-2.0**|
|
| 64 |
+
|
| 65 |
+
_All upstream components are Apache 2.0 / MIT — LoRA inherits permissive terms._
|
| 66 |
+
|
| 67 |
+
## EU AI Act compliance
|
| 68 |
+
|
| 69 |
+
- **Article 53(1)(c)**: training data licenses preserved (per-dataset cards declare upstream licenses).
|
| 70 |
+
- **Article 53(1)(d)**: training data summary — see upstream dataset cards on Ailiance-fr.
|
| 71 |
+
- **GPAI Code of Practice (July 2025)**: base `mistralai/Devstral-Small-2-24B-Instruct-2512` released under apache-2.0.
|
| 72 |
+
- **No web scraping by Ailiance**, **no licensed data**, **no PII**.
|
| 73 |
+
- Upstream Stack Exchange content (where applicable) is CC-BY-SA-4.0 and propagates to this adapter.
|
| 74 |
+
|
| 75 |
+
## License
|
| 76 |
+
|
| 77 |
+
LoRA weights: **apache-2.0** — see License chain table above for derivation rationale.
|
| 78 |
+
|
| 79 |
+
## Citation
|
| 80 |
+
|
| 81 |
+
```bibtex
|
| 82 |
+
@misc{ailiance_devstral_typescript_2026,
|
| 83 |
+
author = {Ailiance},
|
| 84 |
+
title = {Ailiance — Devstral-Small-2-24B-Instruct typescript LoRA},
|
| 85 |
+
year = {2026},
|
| 86 |
+
publisher = {Hugging Face},
|
| 87 |
+
url = {https://huggingface.co/Ailiance-fr/devstral-typescript-lora}
|
| 88 |
+
}
|
| 89 |
+
```
|
| 90 |
+
|
| 91 |
+
## Related
|
| 92 |
+
|
| 93 |
+
See the full [Ailiance-fr LoRA collection](https://huggingface.co/Ailiance-fr).
|