đ iir-eng Translation Model
This project focuses on a translation model that bridges the Indo - Iranian languages and English. It provides a reliable solution for translating between these language groups, with a well - trained model and detailed evaluation metrics.
đ Quick Start
The model details and related resources are as follows:
- Source Group: Indo - Iranian languages
- Target Group: English
- OPUS Readme: [iir - eng](https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/iir - eng/README.md)
- Model: Transformer
- Source Languages: asm, awa, ben, bho, gom, guj, hif_Latn, hin, jdt_Cyrl, kur_Arab, kur_Latn, mai, mar, npi, ori, oss, pan_Guru, pes, pes_Latn, pes_Thaa, pnb, pus, rom, san_Deva, sin, snd_Arab, tgk_Cyrl, tly_Latn, urd, zza
- Target Language: eng
- Pre - processing: normalization + SentencePiece (spm32k, spm32k)
- Download Original Weights: [opus2m - 2020 - 08 - 01.zip](https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.zip)
- Test Set Translations: [opus2m - 2020 - 08 - 01.test.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.test.txt)
- Test Set Scores: [opus2m - 2020 - 08 - 01.eval.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.eval.txt)
đ Documentation
Benchmarks
The following table shows the BLEU and chr - F scores of the model on different test sets:
| Testset |
BLEU |
chr - F |
| newsdev2014 - hineng.hin.eng |
8.1 |
0.324 |
| newsdev2019 - engu - gujeng.guj.eng |
8.1 |
0.309 |
| newstest2014 - hien - hineng.hin.eng |
12.1 |
0.380 |
| newstest2019 - guen - gujeng.guj.eng |
6.0 |
0.280 |
| Tatoeba - test.asm - eng.asm.eng |
13.9 |
0.327 |
| Tatoeba - test.awa - eng.awa.eng |
7.0 |
0.219 |
| Tatoeba - test.ben - eng.ben.eng |
42.5 |
0.576 |
| Tatoeba - test.bho - eng.bho.eng |
27.3 |
0.452 |
| Tatoeba - test.fas - eng.fas.eng |
5.6 |
0.262 |
| Tatoeba - test.guj - eng.guj.eng |
15.9 |
0.350 |
| Tatoeba - test.hif - eng.hif.eng |
10.1 |
0.247 |
| Tatoeba - test.hin - eng.hin.eng |
36.5 |
0.544 |
| Tatoeba - test.jdt - eng.jdt.eng |
11.4 |
0.094 |
| Tatoeba - test.kok - eng.kok.eng |
6.6 |
0.256 |
| Tatoeba - test.kur - eng.kur.eng |
3.4 |
0.149 |
| Tatoeba - test.lah - eng.lah.eng |
17.4 |
0.301 |
| Tatoeba - test.mai - eng.mai.eng |
65.4 |
0.703 |
| Tatoeba - test.mar - eng.mar.eng |
22.5 |
0.468 |
| Tatoeba - test.multi.eng |
21.3 |
0.424 |
| Tatoeba - test.nep - eng.nep.eng |
3.4 |
0.185 |
| Tatoeba - test.ori - eng.ori.eng |
4.8 |
0.244 |
| Tatoeba - test.oss - eng.oss.eng |
1.6 |
0.173 |
| Tatoeba - test.pan - eng.pan.eng |
14.8 |
0.348 |
| Tatoeba - test.pus - eng.pus.eng |
1.1 |
0.182 |
| Tatoeba - test.rom - eng.rom.eng |
2.8 |
0.185 |
| Tatoeba - test.san - eng.san.eng |
2.8 |
0.185 |
| Tatoeba - test.sin - eng.sin.eng |
22.8 |
0.474 |
| Tatoeba - test.snd - eng.snd.eng |
8.2 |
0.287 |
| Tatoeba - test.tgk - eng.tgk.eng |
11.9 |
0.321 |
| Tatoeba - test.tly - eng.tly.eng |
0.9 |
0.076 |
| Tatoeba - test.urd - eng.urd.eng |
23.9 |
0.438 |
| Tatoeba - test.zza - eng.zza.eng |
0.6 |
0.098 |
System Info
| Property |
Details |
| hf_name |
iir - eng |
| source_languages |
iir |
| target_languages |
eng |
| opus_readme_url |
[https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/iir - eng/README.md](https://github.com/Helsinki - NLP/Tatoeba - Challenge/tree/master/models/iir - eng/README.md) |
| original_repo |
Tatoeba - Challenge |
| tags |
['translation'] |
| languages |
['bn', 'or', 'gu', 'mr', 'ur', 'hi', 'ps', 'os', 'as', 'si', 'iir', 'en'] |
| src_constituents |
{'pnb', 'gom', 'ben', 'hif_Latn', 'ori', 'guj', 'pan_Guru', 'snd_Arab', 'npi', 'mar', 'urd', 'pes', 'bho', 'kur_Arab', 'tgk_Cyrl', 'hin', 'kur_Latn', 'pes_Thaa', 'pus', 'san_Deva', 'oss', 'tly_Latn', 'jdt_Cyrl', 'asm', 'zza', 'rom', 'mai', 'pes_Latn', 'awa', 'sin'} |
| tgt_constituents |
{'eng'} |
| src_multilingual |
True |
| tgt_multilingual |
False |
| prepro |
normalization + SentencePiece (spm32k, spm32k) |
| url_model |
[https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.zip](https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.zip) |
| url_test_set |
[https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.test.txt](https://object.pouta.csc.fi/Tatoeba - MT - models/iir - eng/opus2m - 2020 - 08 - 01.test.txt) |
| src_alpha3 |
iir |
| tgt_alpha3 |
eng |
| short_pair |
iir - en |
| chrF2_score |
0.424 |
| bleu |
21.3 |
| brevity_penalty |
1.0 |
| ref_len |
67026.0 |
| src_name |
Indo - Iranian languages |
| tgt_name |
English |
| train_date |
2020 - 08 - 01 |
| src_alpha2 |
iir |
| tgt_alpha2 |
en |
| prefer_old |
False |
| long_pair |
iir - eng |
| helsinki_git_sha |
480fcbe0ee1bf4774bcbe6226ad9f58e63f6c535 |
| transformers_git_sha |
2207e5d8cb224e954a7cba69fa4ac2309e9ff30b |
| port_machine |
brutasse |
| port_time |
2020 - 08 - 21 - 14:41 |
đ License
This project is licensed under the Apache - 2.0 license.
%20--%3e%3cdefs%3e%3cstyle%3e%20.st0%20{%20fill:%20%23061b40;%20}%20.st1%20{%20fill:%20%23306af1;%20}%20.st2%20{%20fill:%20%235ce5cf;%20}%20%3c/style%3e%3c/defs%3e%3cg%3e%3cpath%20class='st0'%20d='M55,10.5h9v3h-9v9h12V7.5h-12v3ZM64,19.5h-6v-3h6v3Z'/%3e%3cpolygon%20class='st0'%20points='69%2016.5%2078%2016.5%2078%2019.5%2069%2019.5%2069%2022.5%2081%2022.5%2081%2013.5%2072%2013.5%2072%2010.5%2081%2010.5%2081%207.5%2069%207.5%2069%2016.5'/%3e%3cpolygon%20class='st0'%20points='95%2010.5%2095%207.5%2083%207.5%2083%2022.5%2095%2022.5%2095%2019.5%2086%2019.5%2086%2016.5%2095%2016.5%2095%2013.5%2086%2013.5%2086%2010.5%2095%2010.5'/%3e%3cpath%20class='st0'%20d='M40,1.5v21h11.6l1.4-1.4v-7.6h0c0,0-1.4-1.5-1.4-1.5l1.4-1.4V2.9l-1.4-1.4h-11.6ZM50,19.5h-7v-6h7v6ZM50,10.5h-7v-6h7v6Z'/%3e%3c/g%3e%3cpath%20class='st1'%20d='M23.1,24L14.7,4.8l-4.9,11.2h3.8l-1.8,4H3.3L12.1,0H2C.9,0,0,.9,0,2v20c0,1.1.9,2,2,2h21.1Z'/%3e%3cpath%20class='st2'%20d='M34,0h-16.8l10.6,24h6.2c1.1,0,2-.9,2-2V2C36,.9,35.1,0,34,0ZM32.5,20h-4V4h4v16Z'/%3e%3c/svg%3e)
