sneakyfree commited on
Commit
576def7
·
verified ·
1 Parent(s): 3dae6ab

Add model card

Browse files
Files changed (1) hide show
  1. README.md +75 -0
README.md ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: cc-by-4.0
3
+ tags:
4
+ - translation
5
+ - marian
6
+ - windyword
7
+ - english
8
+ language:
9
+ - wls
10
+ - en
11
+ library_name: transformers
12
+ pipeline_tag: translation
13
+ ---
14
+
15
+ # WindyWord.ai Translation — wls → English
16
+
17
+ **Translates wls → English.**
18
+
19
+ **Quality Rating: ⭐⭐½ (2.5★ Basic)**
20
+
21
+ Part of the [WindyWord.ai](https://windyword.ai) translation fleet — 1,800+ proprietary language pairs.
22
+
23
+ ## Quality & Pricing Tier
24
+
25
+ - **5-star rating:** 2.5★ ⭐⭐½
26
+ - **Tier:** Basic
27
+ - **Composite score:** 52.5 / 100
28
+ - **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)
29
+
30
+ ## Available Variants
31
+
32
+ This repository contains multiple deployment formats. Pick the one that matches your use case:
33
+
34
+ | Variant | Description |
35
+ |---|---|
36
+ | `lora/` | **WindyStandard** — our proprietary production baseline. Stable, reliable, optimized for GPU inference. |
37
+ | `lora-ct2-int8/` | **WindyStandard · CPU INT8** — CTranslate2 quantized version of WindyStandard. ~25% of the size, 2–4× faster on CPU, no measurable quality loss. |
38
+ | `herm0/` | **WindyEnhanced** — deep fine-tuned on OPUS-100, Tatoeba, and WikiMatrix parallel corpora. Measurably higher translation quality on supported pairs. |
39
+ | `herm0-ct2-int8/` | **WindyEnhanced · CPU INT8** — CTranslate2 quantized WindyEnhanced. Premium quality, CPU-efficient. |
40
+
41
+ ### Quick usage
42
+
43
+ **Transformers (PyTorch):**
44
+ ```python
45
+ from transformers import MarianMTModel, MarianTokenizer
46
+ tokenizer = MarianTokenizer.from_pretrained("WindyWord/translate-wls-en", subfolder="lora")
47
+ model = MarianMTModel.from_pretrained("WindyWord/translate-wls-en", subfolder="lora")
48
+ ```
49
+
50
+ **CTranslate2 (fast CPU inference):**
51
+ ```python
52
+ import ctranslate2
53
+ translator = ctranslate2.Translator("path/to/translate-wls-en/lora-ct2-int8")
54
+ ```
55
+
56
+ ## Commercial Use
57
+
58
+ The WindyWord.ai platform provides:
59
+ - **Mobile apps** (iOS, Android — coming soon)
60
+ - **Real-time voice-to-text-to-translation** pipeline
61
+ - **API access** with premium model quality
62
+ - **Offline deployment** support
63
+
64
+ Visit [windyword.ai](https://windyword.ai) for apps and commercial API access.
65
+
66
+ ---
67
+
68
+ ## Provenance & License
69
+
70
+ Weights derived from the OPUS-MT project ([Helsinki-NLP/opus-mt-wls-en](https://huggingface.co/Helsinki-NLP/opus-mt-wls-en)) under CC-BY-4.0. WindyStandard, WindyEnhanced, and WindyScripture variants are proprietary to WindyWord.ai, independently trained and quality-certified via our Grand Rounds v2 test battery.
71
+
72
+ Licensed CC-BY-4.0 — attribution preserved as required.
73
+
74
+ *Certified by Opus 4.6 Opus-Claw (Dr. C) on Veron-1 (RTX 5090).*
75
+ *Patient file: [clinic record](https://github.com/sneakyfree/Windy-Clinic/blob/main/translation-pairs/wls-en.json)*