OptimaMT it β†’ es (Italian to Spanish)

Drop-in copy of Xenova/opus-mt-it-es at revision 0eba27fe6da2 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 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 = 1 501 β†’ 271 MiB (-46 %) 478 β†’ 246 MiB (-49 %) yes
ONNX Runtime defaults 269 β†’ 211 MiB (-21 %) 190 β†’ 161 MiB (-15 %) yes

FLORES-200 devtest (1,012 sentences): the same translations as the original, BLEU 21.72, chrF 51.17 (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/it-es", {
  dtype: "q8",
  session_options: { extra: { session: { disable_quant_qdq: "1" } } },
});
const [out] = await translator("Il faro si trova sulla costa.");
console.log(out.translation_text);

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

python translate.py OptimaMT/it-es "Il faro si trova sulla costa."

Files

Only onnx/decoder_model_merged_quantized.onnx is rebuilt; every other file is the original, byte for byte. Vocabulary: 59,031 tokens.

Licence and credits

  • Weights: OPUS-MT by the University of Helsinki, CC BY 4.0.
  • ONNX export: Xenova/opus-mt-it-es at revision 0eba27fe6da2.
  • Modified: the decoder's output projection was moved out of its two branches (graph change only). CC BY 4.0.
Downloads last month
100
Inference Providers NEW
This model isn't deployed by any Inference Provider. πŸ™‹ Ask for provider support

Model tree for OptimaMT/it-es

Quantized
(1)
this model

Space using OptimaMT/it-es 1

Collection including OptimaMT/it-es