sneakyfree commited on
Commit
f957545
·
verified ·
1 Parent(s): 87653c2

Add model card

Browse files
Files changed (1) hide show
  1. README.md +78 -0
README.md ADDED
@@ -0,0 +1,78 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ tags:
4
+ - translation
5
+ - marian
6
+ - opus-mt
7
+ - windyword
8
+ language:
9
+ - multilingual
10
+ library_name: transformers
11
+ pipeline_tag: translation
12
+ ---
13
+
14
+ # WindyWord.ai Translation — es → NORWAY
15
+
16
+ **Quality Rating: ⭐⭐½ (2.5★ Basic)**
17
+
18
+ Part of the [WindyWord.ai](https://windyword.ai) translation fleet — 1,800+ proprietary language pairs.
19
+
20
+ ## Quality & Pricing Tier
21
+
22
+ - **5-star rating:** 2.5★ ⭐⭐½
23
+ - **Tier:** Basic
24
+ - **Composite score:** 52.5 / 100
25
+ - **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)
26
+
27
+ ## Available Variants
28
+
29
+ This repository contains multiple deployment formats. Pick the one that matches your use case:
30
+
31
+ | Variant | Description |
32
+ |---|---|
33
+ | `lora/` | Proprietary fog-of-mirror fork. Safe baseline, quality ≈ Helsinki-NLP original. |
34
+ | `lora-ct2-int8/` | CT2 INT8 quantized lora. ~25% of size, 2-4× faster CPU inference, no quality loss. |
35
+ | `herm0/` | Deep OPUS-100 fine-tuned improvement. Highest quality when available. |
36
+ | `herm0-ct2-int8/` | CT2 INT8 of herm0. Best quality + efficient inference. |
37
+
38
+ ### Quick usage
39
+
40
+ **Transformers (PyTorch):**
41
+ ```python
42
+ from transformers import MarianMTModel, MarianTokenizer
43
+ tokenizer = MarianTokenizer.from_pretrained("WindyWord/translate-es-NORWAY", subfolder="lora")
44
+ model = MarianMTModel.from_pretrained("WindyWord/translate-es-NORWAY", subfolder="lora")
45
+ ```
46
+
47
+ **CTranslate2 (fast CPU inference):**
48
+ ```python
49
+ import ctranslate2
50
+ translator = ctranslate2.Translator("path/to/translate-es-NORWAY/lora-ct2-int8")
51
+ ```
52
+
53
+ ## Attribution
54
+
55
+ Derived from [Helsinki-NLP/opus-mt-es-NORWAY](https://huggingface.co/Helsinki-NLP/opus-mt-es-NORWAY) (Helsinki-NLP OPUS-MT project, CC-BY-4.0).
56
+
57
+ Proprietary variants created by the WindyWord.ai team:
58
+ - **lora/**: Fog-of-mirror LoRA fine-tune (r=4, α=8) — legally distinct, quality-preserved
59
+ - **herm0/**: OPUS-100/Tatoeba/WikiMatrix deep fine-tune (if available) — measurably improved
60
+ - **herm0-scripture/**: eBible verse-aligned fine-tune (for 292 scripture pairs)
61
+
62
+ ## Commercial Use
63
+
64
+ The WindyWord.ai platform provides:
65
+ - **Mobile apps** (iOS, Android — coming soon)
66
+ - **Real-time voice-to-text-to-translation** pipeline
67
+ - **API access** with premium model quality
68
+ - **Offline deployment** support
69
+
70
+ Visit [windyword.ai](https://windyword.ai) for apps and commercial API access.
71
+
72
+ ## License
73
+
74
+ CC-BY-4.0, inherited from upstream Helsinki-NLP. Attribution required.
75
+
76
+ ---
77
+ *Certified by Opus 4.6 Opus-Claw (Dr. C) via WindyWord.ai quality assurance pipeline.*
78
+ *Patient file: [clinic record](https://github.com/sneakyfree/Windy-Clinic/blob/main/translation-pairs/es-NORWAY.json)*