How to use from the
Use from the
Transformers.js library
// npm i @huggingface/transformers
import { pipeline } from '@huggingface/transformers';

// Allocate pipeline
const pipe = await pipeline('translation', 'OptimaMT/ar-de');

OptimaMT ar โ†’ de (Arabic to German)

Drop-in copy of an ONNX export of Helsinki-NLP/opus-mt-ar-de at revision 8e6cbcce68fe (same layout as the Xenova exports) with a lighter merged decoder: ONNX Runtime's fast mode (disable_quant_qdq = 1, 2.3โ€“2.5ร— faster) at half the memory. Same weights, same tokens. Part of OptimaMT: all 51 pairs, named OptimaMT/<src>-<tgt>.

Used in production:

  • 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 = 1 530 โ†’ 283 MiB (-47 %) 506 โ†’ 259 MiB (-49 %) yes
ONNX Runtime defaults 283 โ†’ 221 MiB (-22 %) 199 โ†’ 167 MiB (-16 %) yes

FLORES-200 devtest (1,012 sentences): the same translations as the original, BLEU 17.43, chrF 47.48 (greedy).

ONNX Runtime 1.31.1, CPU, 4 threads (Linux; Windows in a VM). Interactive results: OptimaMT/results ยท method: REPORT.md.

Use

import { pipeline } from "@huggingface/transformers";

const translator = await pipeline("translation", "OptimaMT/ar-de", {
  dtype: "q8",
  session_options: { extra: { session: { disable_quant_qdq: "1" } } },
});
const [out] = await translator("ุงู„ู…ู†ุงุฑุฉ ุชู‚ู ุนู„ู‰ ุงู„ุณุงุญู„.");
console.log(out.translation_text);

Python (translate.py: onnxruntime, sentencepiece, numpy):

python translate.py OptimaMT/ar-de "ุงู„ู…ู†ุงุฑุฉ ุชู‚ู ุนู„ู‰ ุงู„ุณุงุญู„."

Files

onnx/decoder_model_merged_quantized.onnx is rebuilt; tokenizer.json, special_tokens_map.json and tokenizer_config.json were written in the Xenova layout; every other file is the original, byte for byte. Vocabulary: 63,079 tokens.

Licence and credits

  • Weights: OPUS-MT by the University of Helsinki, CC BY 4.0.
  • ONNX export: an ONNX export of Helsinki-NLP/opus-mt-ar-de at revision 8e6cbcce68fe (same layout as the Xenova exports).
  • Modified: the decoder's output projection was moved out of its two branches (graph change only). CC BY 4.0.
Downloads last month
122
Inference Providers NEW
This model isn't deployed by any Inference Provider. ๐Ÿ™‹ Ask for provider support

Model tree for OptimaMT/ar-de

Quantized
(2)
this model

Space using OptimaMT/ar-de 1

Collection including OptimaMT/ar-de