OptimaMT es β†’ ar (Spanish to Arabic)

Drop-in copy of an ONNX export of Helsinki-NLP/opus-mt-es-ar at revision 76efb7da47ad (same layout as the Xenova exports) with a lighter merged decoder: ONNX Runtime's fast mode (disable_quant_qdq = 1, 2.3–2.5Γ— faster) at half the memory. Same weights, same tokens. Part of OptimaMT: all 49 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 = 1 526 β†’ 282 MiB (-46 %) 502 β†’ 260 MiB (-48 %) yes
ONNX Runtime defaults 281 β†’ 220 MiB (-22 %) 197 β†’ 166 MiB (-16 %) yes

FLORES-200 devtest (1,012 sentences): the same translations as the original, BLEU 10.52, chrF 43.09 (greedy).

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

Use

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

const translator = await pipeline("translation", "OptimaMT/es-ar", {
  dtype: "q8",
  session_options: { extra: { session: { disable_quant_qdq: "1" } } },
});
const [out] = await translator(">>ara<< El faro estΓ‘ en la costa.");
console.log(out.translation_text);

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

python translate.py OptimaMT/es-ar "El faro estΓ‘ en la costa." --lang ">>ara<<"

Start every input with >>ara<< (the target variant). Use transformers.js 3.x: 4.3.2 splits this token, with the original export too.

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: 62,518 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-es-ar at revision 76efb7da47ad (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
47
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for OptimaMT/es-ar

Quantized
(1)
this model

Space using OptimaMT/es-ar 1

Collection including OptimaMT/es-ar