Downloads · 30 days
29
1% of all-time downloads
opus-mt-zlw-fiu
Helsinki-NLP/opus-mt-zlw-fiu
opus-mt-zlw-fiu is a translation model from Helsinki-NLP. Use it when you need text moved from one language to another. It is set up for transformers. The card lists the license as apache-2.0.
source language name: West Slavic languages target language name: Finno-Ugrian languages OPUS readme: README.md model: transformer source language codes: dsb, cs, csbLatn, hsb, pl, zlw target language codes: hu, vro,…
All-time downloads
2K
Public
Parameters
74.8M
1 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.h5302 MB · 50%
At a glance
- Task
- Translation
- Library
- transformers
- License
- apache-2.0
- Model type
- marian
- Access
- Public
- Created
- Mar 2, 2022
- Updated
- Oct 10, 2023
- SHA
- b1f1df76
Model card
From the Hugging Face model README
zlw-fiu
- source language name: West Slavic languages
- target language name: Finno-Ugrian languages
- OPUS readme: README.md
- model: transformer
- source language codes: dsb, cs, csb_Latn, hsb, pl, zlw
- target language codes: hu, vro, fi, liv_Latn, mdf, krl, fkv_Latn, mhr, et, sma, udm, vep, myv, kpv, se, izh, fiu
- dataset: opus
- release date: 2021-02-18
- pre-processing: normalization + SentencePiece (spm32k,spm32k)
- download original weights: opus-2021-02-18.zip
- a sentence-initial language token is required in the form of >>id<<(id = valid, usually three-letter target language ID)
- Training data:
- ces-fin: Tatoeba-train (1000000)
- ces-hun: Tatoeba-train (1000000)
- pol-est: Tatoeba-train (1000000)
- pol-fin: Tatoeba-train (1000000)
- pol-hun: Tatoeba-train (1000000)
- Validation data:
- ces-fin: Tatoeba-dev, 1000
- ces-hun: Tatoeba-dev, 1000
- est-pol: Tatoeba-dev, 1000
- fin-pol: Tatoeba-dev, 1000
- hun-pol: Tatoeba-dev, 1000
- mhr-pol: Tatoeba-dev, 461
- total-size-shuffled: 5426
- devset-selected: top 5000 lines of Tatoeba-dev.src.shuffled!
- Test data:
- newssyscomb2009.ces-hun: 502/9733
- newstest2009.ces-hun: 2525/54965
- Tatoeba-test.ces-fin: 88/408
- Tatoeba-test.ces-hun: 1911/10336
- Tatoeba-test.multi-multi: 4562/25497
- Tatoeba-test.pol-chm: 5/36
- Tatoeba-test.pol-est: 15/98
- Tatoeba-test.pol-fin: 609/3293
- Tatoeba-test.pol-hun: 1934/11285
- test set translations file: test.txt
- test set scores file: eval.txt
- BLEU-scores |Test set|score| |---|---| |Tatoeba-test.ces-fin|57.2| |Tatoeba-test.ces-hun|42.6| |Tatoeba-test.multi-multi|39.4| |Tatoeba-test.pol-hun|36.6| |Tatoeba-test.pol-fin|36.1| |Tatoeba-test.pol-est|20.9| |newssyscomb2009.ces-hun|13.9| |newstest2009.ces-hun|13.9| |Tatoeba-test.pol-chm|2.0|
- chr-F-scores |Test set|score| |---|---| |Tatoeba-test.ces-fin|0.71| |Tatoeba-test.ces-hun|0.637| |Tatoeba-test.multi-multi|0.616| |Tatoeba-test.pol-hun|0.605| |Tatoeba-test.pol-fin|0.592| |newssyscomb2009.ces-hun|0.449| |newstest2009.ces-hun|0.443| |Tatoeba-test.pol-est|0.372| |Tatoeba-test.pol-chm|0.007|
System Info:
- hf_name: zlw-fiu
- source_languages: dsb,cs,csb_Latn,hsb,pl,zlw
- target_languages: hu,vro,fi,liv_Latn,mdf,krl,fkv_Latn,mhr,et,sma,udm,vep,myv,kpv,se,izh,fiu
- opus_readme_url: https://object.pouta.csc.fi/Tatoeba-MT-models/zlw-fiu/opus-2021-02-18.zip/README.md
- original_repo: Tatoeba-Challenge
- tags: ['translation']
- languages: ['dsb', 'cs', 'csb_Latn', 'hsb', 'pl', 'zlw', 'hu', 'vro', 'fi', 'liv_Latn', 'mdf', 'krl', 'fkv_Latn', 'mhr', 'et', 'sma', 'udm', 'vep', 'myv', 'kpv', 'se', 'izh', 'fiu']
- src_constituents: ['dsb', 'ces', 'csb_Latn', 'hsb', 'pol']
- tgt_constituents: ['hun', 'vro', 'fin', 'liv_Latn', 'mdf', 'krl', 'fkv_Latn', 'mhr', 'est', 'sma', 'udm', 'vep', 'myv', 'kpv', 'sme', 'izh']
- src_multilingual: True
- tgt_multilingual: True
- helsinki_git_sha: a0966db6db0ae616a28471ff0faf461b36fec07d
- transformers_git_sha: 3857f2b4e34912c942694489c2b667d9476e55f5
- port_machine: bungle
- port_time: 2021-06-29-15:24
- Task
- Translation
- Library
- transformers
- Type
- marian
- License
- apache-2.0
- Created
- Mar 2, 2022
- Updated
- Oct 10, 2023