Add model card

README.md

@@ -0,0 +1,79 @@
+---
+license: cc-by-4.0
+tags:
+- translation
+- marian
+- opus-mt
+- windyword
+language:
+- en
+- luo
+library_name: transformers
+pipeline_tag: translation
+---
+
+# WindyWord.ai Translation — en → luo
+
+**Quality Rating: ⭐⭐⭐⭐  (4.0★ Standard)**
+
+Part of the [WindyWord.ai](https://windyword.ai) translation fleet — 1,800+ proprietary language pairs.
+
+## Quality & Pricing Tier
+
+- **5-star rating:** 4.0★ ⭐⭐⭐⭐
+- **Tier:** Standard
+- **Composite score:** 74.7 / 100
+- **Rated via:** Grand Rounds v2 — an 8-test stress battery (paragraphs, multi-paragraph, native input, domain stress, edge cases, round-trip fidelity, speed, and consistency checks)
+
+## Available Variants
+
+This repository contains multiple deployment formats. Pick the one that matches your use case:
+
+| Variant | Description |
+|---|---|
+| `lora/` | Proprietary fog-of-mirror fork. Safe baseline, quality ≈ Helsinki-NLP original. |
+| `lora-ct2-int8/` | CT2 INT8 quantized lora. ~25% of size, 2-4× faster CPU inference, no quality loss. |
+| `herm0-scripture/` | eBible verse-aligned fine-tune. Specialized for biblical text; NOT recommended for general translation. |
+| `scripture-ct2-int8/` | CT2 INT8 of scripture variant. |
+
+### Quick usage
+
+**Transformers (PyTorch):**
+```python
+from transformers import MarianMTModel, MarianTokenizer
+tokenizer = MarianTokenizer.from_pretrained("WindyWord/translate-en-luo", subfolder="lora")
+model = MarianMTModel.from_pretrained("WindyWord/translate-en-luo", subfolder="lora")
+```
+
+**CTranslate2 (fast CPU inference):**
+```python
+import ctranslate2
+translator = ctranslate2.Translator("path/to/translate-en-luo/lora-ct2-int8")
+```
+
+## Attribution
+
+Derived from [Helsinki-NLP/opus-mt-en-luo](https://huggingface.co/Helsinki-NLP/opus-mt-en-luo) (Helsinki-NLP OPUS-MT project, CC-BY-4.0).
+
+Proprietary variants created by the WindyWord.ai team:
+- **lora/**: Fog-of-mirror LoRA fine-tune (r=4, α=8) — legally distinct, quality-preserved
+- **herm0/**: OPUS-100/Tatoeba/WikiMatrix deep fine-tune (if available) — measurably improved
+- **herm0-scripture/**: eBible verse-aligned fine-tune (for 292 scripture pairs)
+
+## Commercial Use
+
+The WindyWord.ai platform provides:
+- **Mobile apps** (iOS, Android — coming soon)
+- **Real-time voice-to-text-to-translation** pipeline
+- **API access** with premium model quality
+- **Offline deployment** support
+
+Visit [windyword.ai](https://windyword.ai) for apps and commercial API access.
+
+## License
+
+CC-BY-4.0, inherited from upstream Helsinki-NLP. Attribution required.
+
+---
+*Certified by Opus 4.6 Opus-Claw (Dr. C) via WindyWord.ai quality assurance pipeline.*
+*Patient file: [clinic record](https://github.com/sneakyfree/Windy-Clinic/blob/main/translation-pairs/en-luo.json)*