| --- |
| license: cc-by-4.0 |
| base_model: HPLT/translate-en-vi-v2.0-hplt_opus |
| library_name: ctranslate2 |
| pipeline_tag: translation |
| tags: [translation, marian, ctranslate2, int8, en, vi] |
| --- |
| |
| # translate-windy-hplt-en — en to vi |
|
|
| Bilingual machine translation in **CTranslate2 INT8** for fast CPU inference. |
| Fine-tuned by **Windstorm Labs** from [`HPLT/translate-en-vi-v2.0-hplt_opus`](https://huggingface.co/HPLT/translate-en-vi-v2.0-hplt_opus). |
|
|
| ## Attribution |
|
|
| Derived from [`HPLT/translate-en-vi-v2.0-hplt_opus`](https://huggingface.co/HPLT/translate-en-vi-v2.0-hplt_opus), from the **HPLT** project, licensed under |
| **CC-BY-4.0**. **Modified by Windstorm Labs.** Attribution is retained as the licence requires. |
|
|
| ## What we did |
|
|
| Converted from the upstream Marian release, LoRA fine-tuned on parallel data for this |
| language pair, merged into the base weights, then quantized to INT8. |
|
|
| | | | |
| |---|---| |
| | Rank / alpha | 8 / 16 | |
| | Learning rate / steps | 2.5e-06 / 50 | |
| | Target modules | `q_proj`, `v_proj` | |
| | Seed | 42 | |
| | Training data | opus100_en-vi (clinic cache), 2,000 sentence pairs | |
| | Tensors modified | 36 | |
| |
| ## Provenance |
| |
| ``` |
| base model.bin sha256 0c94dc9fcf527ddab4e9df98dee176b401213aa3fbaf9c4feed5623e439b9c99 |
| this model.bin sha256 d2108b1d5cf28a080d3ca20b17a17f313ed6f731a24a9bb9e155a0c05d08b1a9 |
| ``` |
| |
| Verified after INT8 quantization, on the published artifact. |
| |
| ## Evaluation |
| |
| FLORES-200 devtest, 200 sentences, beam 4, CTranslate2 INT8 on CPU. |
| |
| | | spBLEU | chrF | |
| |---|---:|---:| |
| | this model | **40.35** | 57.7 | |
| |
| ## Notes |
| |
| - Single language pair: en to vi. |
| - FLORES-200 is news and encyclopedic prose. |
| - No human evaluation was performed. |
| |
