A small (~75M parameters, 6+6 layers, d=512) MarianMT model for one language pair. In CrispASR these are the fastest translators for live transcription + translation (
--live-translate
).
Licence and attribution
CC-BY-4.0.
The weights are the
opus-2020-02-26
release of the
OPUS-MT project
(Jörg Tiedemann and Santhosh Thottingal, University of Helsinki;
OPUS-MT — Building open translation services for the World
, EAMT 2020), trained on
OPUS
data. The project distributes its pre-trained models under CC-BY 4.0 (
Opus-MT README
,
OPUS-MT-train README
); that statement, not the tag on an individual model card, is what this repository follows. This is a format conversion: the weights are unchanged at f16 and quantized at q8_0. Redistribution requires attribution to the OPUS-MT project.
Files
File
Size
Output vs. the reference implementation
opus-mt-fr-en-f16.gguf
155 MB
8/8 sentences identical greedy, 8/8 with beam 4
opus-mt-fr-en-q8_0.gguf
85 MB
8/8 sentences identical greedy — any others differ in wording. Fastest; recommended.
"Reference" is
MarianMTModel.generate
from Hugging Face
transformers
on the original checkpoint, on 8 test sentences; input token ids were identical for 8/8.
Quick start
git clone https://github.com/CrispStrobe/CrispASR && cd CrispASR
cmake -B build -DCMAKE_BUILD_TYPE=Release && cmake --build build -j
# Text → text. -bs 1 is greedy; without it the checkpoint's own beam size is used.
./build/bin/crispasr --backend marian -m opus-mt-fr-en -sl fr -tl en -bs 1 \
--text "Bonjour et bienvenue à la réunion d'aujourd'hui."# Hello and welcome to today's meeting.# Live: microphone in, transcript + translation out, sentence by sentence
./build/bin/crispasr --live-translate -l fr --tr-tl en \
-m auto --backend parakeet --translate-backend marian
-m opus-mt-fr-en
downloads
opus-mt-fr-en-q8_0.gguf
on first use. The recogniser in the live example must support French.
Notes
One model per direction; the opposite direction, where one exists, is
cstr/opus-mt-en-fr-GGUF
.
Live mode always decodes greedy.
Literal
</s>
,
<unk>
,
<pad>
in the input are treated as ordinary text here (the reference treats them as special tokens).
opus-mt-fr-en-GGUF huggingface.co is an AI model on huggingface.co that provides opus-mt-fr-en-GGUF's model effect (), which can be used instantly with this cstr opus-mt-fr-en-GGUF model. huggingface.co supports a free trial of the opus-mt-fr-en-GGUF model, and also provides paid use of the opus-mt-fr-en-GGUF. Support call opus-mt-fr-en-GGUF model through api, including Node.js, Python, http.
opus-mt-fr-en-GGUF huggingface.co is an online trial and call api platform, which integrates opus-mt-fr-en-GGUF's modeling effects, including api services, and provides a free online trial of opus-mt-fr-en-GGUF, you can try opus-mt-fr-en-GGUF online for free by clicking the link below.
cstr opus-mt-fr-en-GGUF online free url in huggingface.co:
opus-mt-fr-en-GGUF is an open source model from GitHub that offers a free installation service, and any user can find opus-mt-fr-en-GGUF on GitHub to install. At the same time, huggingface.co provides the effect of opus-mt-fr-en-GGUF install, users can directly use opus-mt-fr-en-GGUF installed effect in huggingface.co for debugging and trial. It also supports api for free installation.