ja-en / README.md
AmerAlreyahi's picture
Add OptimaMT ja-en
c17ea10 verified
|
Raw History Blame Contribute Delete
2.54 kB
---
license: cc-by-4.0
language:
- ja
- en
pipeline_tag: translation
library_name: transformers.js
base_model: Xenova/opus-mt-ja-en
tags:
- translation
- marian
- opus-mt
- onnx
---
# OptimaMT ja → en (Japanese to English)
Drop-in copy of [Xenova/opus-mt-ja-en](https://huggingface.co/Xenova/opus-mt-ja-en) at revision `1a906cfaaf7c` with a lighter merged decoder: ONNX Runtime's fast mode (`disable_quant_qdq = 1`, 2.3–2.5× faster) at half the memory. Same weights, same tokens.
Part of [OptimaMT](https://github.com/AmerAlreyahi/OptimaMT): [all 51 pairs](https://huggingface.co/collections/OptimaMT/optimamt-lighter-opus-mt-onnx-models-6af228dc9cd91f015e56e869), named `OptimaMT/<src>-<tgt>`.
**Used in production:**
- [OpenCircle](https://github.com/AmerAlreyahi/OpenCircle), a desktop screen translator for Windows and Linux, ships these models since v0.1.4.
| Memory after load | Linux | Windows | Identical tokens |
|---|---:|---:|---|
| `disable_quant_qdq = 1` | 513 → 276 MiB (-46 %) | 488 → 253 MiB (-48 %) | yes |
| ONNX Runtime defaults | 274 → 214 MiB (-22 %) | 194 → 164 MiB (-16 %) | yes |
FLORES-200 devtest (1,012 sentences): the same translations as the original, BLEU 17.68, chrF 47.7 (greedy).
ONNX Runtime 1.31.1, CPU, 4 threads (Linux; Windows in a VM). Interactive results: [OptimaMT/results](https://huggingface.co/spaces/OptimaMT/results) ·
method: [REPORT.md](https://github.com/AmerAlreyahi/OptimaMT/blob/main/REPORT.md).
## Use
```js
import { pipeline } from "@huggingface/transformers";
const translator = await pipeline("translation", "OptimaMT/ja-en", {
dtype: "q8",
session_options: { extra: { session: { disable_quant_qdq: "1" } } },
});
const [out] = await translator("灯台は海岸に立っています。");
console.log(out.translation_text);
```
Python ([translate.py](https://github.com/AmerAlreyahi/OptimaMT/blob/main/scripts/translate.py): onnxruntime, sentencepiece, numpy):
```bash
python translate.py OptimaMT/ja-en "灯台は海岸に立っています。"
```
## Files
Only `onnx/decoder_model_merged_quantized.onnx` is rebuilt; every other file is the original, byte for byte.
Vocabulary: 60,716 tokens.
## Licence and credits
- Weights: [OPUS-MT](https://huggingface.co/Helsinki-NLP/opus-mt-ja-en) by the University of Helsinki, CC BY 4.0.
- ONNX export: [Xenova/opus-mt-ja-en](https://huggingface.co/Xenova/opus-mt-ja-en) at revision `1a906cfaaf7c`.
- Modified: the decoder's output projection was moved out of its two branches (graph change only). CC BY 4.0.