Datasets:
Merge in GATITOS README. Wow, what long tables!
Browse files
README.md
CHANGED
|
@@ -2628,20 +2628,55 @@ configs:
|
|
| 2628 |
# SMOL
|
| 2629 |
|
| 2630 |
SMOL (Set for Maximal Overall Leverage) is a collection professional
|
| 2631 |
-
translations into
|
| 2632 |
translation models, and otherwise increasing the representations of said
|
| 2633 |
languages in NLP and technology.
|
| 2634 |
|
| 2635 |
-
Please read
|
| 2636 |
-
|
| 2637 |
-
## Resources in this repository
|
| 2638 |
|
| 2639 |
There are three resources in this directory:
|
| 2640 |
|
| 2641 |
-
* **
|
| 2642 |
-
* **
|
| 2643 |
-
* **SmolDoc
|
| 2644 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2645 |
|
| 2646 |
## Per-language information
|
| 2647 |
|
|
@@ -2650,127 +2685,236 @@ The following table only includes information for the
|
|
| 2650 |
in both directions, so e.g. this table lists only `en_ay`, but reversed `ay_en`
|
| 2651 |
data is also released.
|
| 2652 |
|
| 2653 |
-
LP | Target Language Name | ISO 15924 Script | Continent | N SmolSent | N
|
| 2654 |
-
|
| 2655 |
-
am_ar
|
| 2656 |
-
am_zh
|
| 2657 |
-
en_aa
|
| 2658 |
-
|
| 2659 |
-
|
| 2660 |
-
|
| 2661 |
-
|
| 2662 |
-
|
| 2663 |
-
|
| 2664 |
-
|
| 2665 |
-
|
| 2666 |
-
|
| 2667 |
-
|
| 2668 |
-
|
| 2669 |
-
|
| 2670 |
-
|
| 2671 |
-
|
| 2672 |
-
|
| 2673 |
-
|
| 2674 |
-
|
| 2675 |
-
|
| 2676 |
-
|
| 2677 |
-
|
| 2678 |
-
|
| 2679 |
-
|
| 2680 |
-
|
| 2681 |
-
|
| 2682 |
-
|
| 2683 |
-
|
| 2684 |
-
|
| 2685 |
-
|
| 2686 |
-
|
| 2687 |
-
|
| 2688 |
-
|
| 2689 |
-
|
| 2690 |
-
|
| 2691 |
-
|
| 2692 |
-
|
| 2693 |
-
|
| 2694 |
-
|
| 2695 |
-
|
| 2696 |
-
|
| 2697 |
-
|
| 2698 |
-
|
| 2699 |
-
|
| 2700 |
-
|
| 2701 |
-
|
| 2702 |
-
|
| 2703 |
-
|
| 2704 |
-
|
| 2705 |
-
|
| 2706 |
-
|
| 2707 |
-
|
| 2708 |
-
|
| 2709 |
-
|
| 2710 |
-
|
| 2711 |
-
|
| 2712 |
-
|
| 2713 |
-
|
| 2714 |
-
|
| 2715 |
-
|
| 2716 |
-
|
| 2717 |
-
|
| 2718 |
-
|
| 2719 |
-
|
| 2720 |
-
|
| 2721 |
-
|
| 2722 |
-
|
| 2723 |
-
|
| 2724 |
-
|
| 2725 |
-
|
| 2726 |
-
|
| 2727 |
-
|
| 2728 |
-
|
| 2729 |
-
|
| 2730 |
-
|
| 2731 |
-
|
| 2732 |
-
|
| 2733 |
-
|
| 2734 |
-
|
| 2735 |
-
|
| 2736 |
-
|
| 2737 |
-
|
| 2738 |
-
|
| 2739 |
-
|
| 2740 |
-
|
| 2741 |
-
|
| 2742 |
-
|
| 2743 |
-
|
| 2744 |
-
|
| 2745 |
-
|
| 2746 |
-
|
| 2747 |
-
|
| 2748 |
-
|
| 2749 |
-
|
| 2750 |
-
|
| 2751 |
-
|
| 2752 |
-
|
| 2753 |
-
|
| 2754 |
-
|
| 2755 |
-
|
| 2756 |
-
|
| 2757 |
-
|
| 2758 |
-
|
| 2759 |
-
|
| 2760 |
-
|
| 2761 |
-
|
| 2762 |
-
|
| 2763 |
-
|
| 2764 |
-
|
| 2765 |
-
|
| 2766 |
-
|
| 2767 |
-
|
| 2768 |
-
|
| 2769 |
-
|
| 2770 |
-
|
| 2771 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2772 |
|
| 2773 |
-
## Notes on language codes
|
| 2774 |
|
| 2775 |
We use the backwards-compatible BCP-47 language code convention. Several of
|
| 2776 |
these languages have various alternate code also in use, including:
|
|
@@ -2784,9 +2928,86 @@ these languages have various alternate code also in use, including:
|
|
| 2784 |
* **qu** Another supercode issue. The Quechua variety this data is closest to
|
| 2785 |
is likely `quy`, though a language expert should opine
|
| 2786 |
* **pa-Arab** is also referred to as `lah`
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2787 |
|
| 2788 |
For convenience, here is a mapping from codes used by Flores and SMOL:
|
| 2789 |
|
|
|
|
| 2790 |
Flores code | BCP-47 code
|
| 2791 |
----------- | -----------
|
| 2792 |
eng_Latn | en
|
|
@@ -3006,11 +3227,23 @@ There are several known issues of this dataset:
|
|
| 3006 |
Tibetan (`bo`) caused notable regressions in `ChrF`, suggesting that they
|
| 3007 |
are low-quality, though this could also be an issue with training or
|
| 3008 |
evaluation.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3009 |
|
| 3010 |
-
|
|
|
|
|
|
|
|
|
|
| 3011 |
Another especially useful piece of information is whether the correct
|
| 3012 |
-
orthographies are used in these deliveries.
|
|
|
|
|
|
|
| 3013 |
|
|
|
|
| 3014 |
## License
|
| 3015 |
|
| 3016 |
This data is released under cc-by-4.0
|
|
|
|
| 2628 |
# SMOL
|
| 2629 |
|
| 2630 |
SMOL (Set for Maximal Overall Leverage) is a collection professional
|
| 2631 |
+
translations into 221 Low-Resource Languages, for the purpose of training
|
| 2632 |
translation models, and otherwise increasing the representations of said
|
| 2633 |
languages in NLP and technology.
|
| 2634 |
|
| 2635 |
+
Please read the [SMOL Paper](...) and the [GATITOS Paper](https://arxiv.org/abs/2303.15265) for a much more thorough description!
|
|
|
|
|
|
|
| 2636 |
|
| 2637 |
There are three resources in this directory:
|
| 2638 |
|
| 2639 |
+
* **GATITOS:** token-level translations into 172 languages
|
| 2640 |
+
* **SmolSent:** sentence-level translations into 81 languages
|
| 2641 |
+
* **SmolDoc:** document-level translations into 100 languages
|
| 2642 |
+
* **SmolDoc-factuality-annotations:** factuality annotations and rationales
|
| 2643 |
+
for 661 documents from `SmolDoc`
|
| 2644 |
+
|
| 2645 |
+
## Notes on the GATITOS multilingual Lexicon
|
| 2646 |
+
|
| 2647 |
+
The GATITOS (Google's Additional Translations Into Tail-languages: Often Short)
|
| 2648 |
+
dataset is a high-quality, multi-way parallel dataset of tokens and short
|
| 2649 |
+
phrases, intended for training and improving machine translation models.
|
| 2650 |
+
Experiments on this dataset and Panlex focusing on unsupervised translation in
|
| 2651 |
+
a 208-language model can be found in
|
| 2652 |
+
[BiLex Rx: Lexical Data Augmentation for Massively Multilingual Machine Translation](https://arxiv.org/abs/2303.15265).
|
| 2653 |
+
|
| 2654 |
+
This dataset consists in 4,000 English segments (4,500 tokens) that have been
|
| 2655 |
+
translated into each of 173 languages, 170 of which are low-resource, and three
|
| 2656 |
+
of which are mid-high resource (es, fr, hi). All translations were made
|
| 2657 |
+
directly from English, with the exception of Aymara, which was translated from
|
| 2658 |
+
the Spanish.
|
| 2659 |
+
|
| 2660 |
+
This dataset contains primarily short segments: 93% single tokens, and only 23
|
| 2661 |
+
sentences (0.6%) have over 5 tokens. As such it is best thought of as a
|
| 2662 |
+
multilingual lexicon, rather than a parallel training corpus. The source text is
|
| 2663 |
+
frequent words in the English Language, along with some common phrases and short
|
| 2664 |
+
sentences. Care has been taken to ensure that they include good coverage of
|
| 2665 |
+
numbers, months, days of the week, swadesh words, and names of the languages
|
| 2666 |
+
themselves (including the endonym).
|
| 2667 |
+
|
| 2668 |
+
Single tokens are remarkably fickle to translate because of the common issue of
|
| 2669 |
+
polysemy. Some languages in this dataset only provide one translation per
|
| 2670 |
+
source token; others provide multiple translations. When multiple translations
|
| 2671 |
+
are given, they are broken up into multiple lines. Therefore, you can get the
|
| 2672 |
+
average number of translations per token by looking at the number of lines in
|
| 2673 |
+
the file versus the number of deduplicated English lines in the file. The three
|
| 2674 |
+
languages with the most average translations per English token are Betawi,
|
| 2675 |
+
Kachin, and Occitan, each with over 1.5 on average.
|
| 2676 |
+
|
| 2677 |
+
Sometimes translators have left annotations like "pike (verb)", "او (مونث)", "фамили(я)", "أسوأ(repeated)" etc..
|
| 2678 |
+
Wolof is particularly noticeable in this regard, with notes like "xont (verb) which means to feed an animal". Several languages also give definitions instead of translations for some words, e.g. Tok Pisin translation of mole" to "wanpela liklik animal we i save stap long hol long graun".
|
| 2679 |
+
|
| 2680 |
|
| 2681 |
## Per-language information
|
| 2682 |
|
|
|
|
| 2685 |
in both directions, so e.g. this table lists only `en_ay`, but reversed `ay_en`
|
| 2686 |
data is also released.
|
| 2687 |
|
| 2688 |
+
LP | Target Language Name | ISO 15924 Script | Continent | Endonym | Alternate names | N Smoldoc | N SmolSent | N Gatitos |
|
| 2689 |
+
---|----------------------|------------------|-----------|---------|-----------------|-----------|------------|-----------|
|
| 2690 |
+
am_ar | Arabic | Arab | Asia | | | 329 | 0 | 0 |
|
| 2691 |
+
am_zh | Mandarin Chinese | Hans | Asia | | | 329 | 0 | 0 |
|
| 2692 |
+
en_aa | Afar | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2693 |
+
en_ab | Abkhaz | Cyrl | Asia | | | 0 | 0 | 4001 |
|
| 2694 |
+
en_ace | Achinese | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2695 |
+
en_ach | Acoli | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2696 |
+
en_ady | Adyghe | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2697 |
+
en_aeb | Tunisian Arabic | Arab | Africa | | | 66 | 862 | 4001 |
|
| 2698 |
+
en_af | Afrikaans | Latn | Africa | | | 130 | 863 | 0 |
|
| 2699 |
+
en_ahr | Ahirani | Deva | Asia | | | 457 | 0 | 0 |
|
| 2700 |
+
en_aii | Assyrian Neo-Aramaic | Syrc | Asia | | | 0 | 0 | 4001 |
|
| 2701 |
+
en_ak | Twi | Latn | Africa | | | 260 | 863 | 4000 |
|
| 2702 |
+
en_alz | Alur | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2703 |
+
en_am | Amharic | Ethi | Africa | | | 584 | 863 | 0 |
|
| 2704 |
+
en_apc | North Levantine Arabic | Arab | Asia | | | 0 | 0 | 4001 |
|
| 2705 |
+
en_apd | Sudanese Arabic | Arab | Africa | | | 0 | 0 | 0 |
|
| 2706 |
+
en_ar-MA | Morrocan Arabic | Arab | Africa | | | 260 | 863 | 4001 |
|
| 2707 |
+
en_arn | Mapudungun | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2708 |
+
en_arz | Egyptian Arabic | Arab | Africa | | | 260 | 863 | 4001 |
|
| 2709 |
+
en_as | Assamese | Beng | Asia | | | 0 | 0 | 4000 |
|
| 2710 |
+
en_av | Avar | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2711 |
+
en_awa | Awadhi | Deva | Asia | | | 0 | 0 | 4001 |
|
| 2712 |
+
en_ay | Aymara | Latn | Americas | | | 0 | 863 | 4000 |
|
| 2713 |
+
en_ayl | Libyan Arabic | Arab | Africa | | | 66 | 863 | 4001 |
|
| 2714 |
+
en_ba | Bashkir | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2715 |
+
en_bal | Baluchi | Arab | Asia | | | 0 | 0 | 4001 |
|
| 2716 |
+
en_ban | Balinese | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2717 |
+
en_bbc | Batak Toba | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2718 |
+
en_bci | Baoulé | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2719 |
+
en_bem | Bemba (Zambia) | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2720 |
+
en_ber | Tamazight (Tifinagh Script) | Tfng | Africa | | | 130 | 862 | 4001 |
|
| 2721 |
+
en_ber-Latn | Tamazight (Latin Script) | Latn | Africa | | | 130 | 862 | 4001 |
|
| 2722 |
+
en_bew | Betawi | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2723 |
+
en_bfq | Badaga | Taml | Asia | | | 457 | 0 | 0 |
|
| 2724 |
+
en_bfy | Bagheli | Deva | Asia | | | 457 | 0 | 0 |
|
| 2725 |
+
en_bgq | Bagri | Deva | Asia | | | 457 | 0 | 0 |
|
| 2726 |
+
en_bho | Bhojpuri | Deva | Asia | | | 0 | 0 | 4000 |
|
| 2727 |
+
en_bik | Central Bikol | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2728 |
+
en_bjn | Banjar | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2729 |
+
en_bjn-Arab | Banjar (Arabic script) | Arab | Asia | | | 0 | 0 | 4001 |
|
| 2730 |
+
en_bm | Bambara | Latn | Africa | | | 260 | 863 | 4000 |
|
| 2731 |
+
en_bns | Bundeli | Deva | Asia | | | 456 | 0 | 0 |
|
| 2732 |
+
en_bo | Tibetan | Tibt | Asia | | | 0 | 863 | 4001 |
|
| 2733 |
+
en_br | Breton | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2734 |
+
en_bra | Braj | Deva | Asia | | | 457 | 0 | 0 |
|
| 2735 |
+
en_brx | Bodo (India) | Deva | Asia | | | 0 | 863 | 4001 |
|
| 2736 |
+
en_bts | Batak Simalungun | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2737 |
+
en_btx | Batak Karo | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2738 |
+
en_bua | Buryat | Cyrl | Asia | | | 0 | 0 | 4001 |
|
| 2739 |
+
en_bug | Buginese | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2740 |
+
en_ccp-Latn | Chakma (Latin script) | Latn | Asia | | | 457 | 0 | 0 |
|
| 2741 |
+
en_ce | Chechen | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2742 |
+
en_cgg | Chiga | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2743 |
+
en_ch | Chamorro | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2744 |
+
en_chk | Chuukese | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2745 |
+
en_chm | Meadow Mari | Cyrl | Asia | | | 0 | 0 | 4001 |
|
| 2746 |
+
en_ckb | Kurdish (Sorani) | Arab | Asia | | | 0 | 0 | 4000 |
|
| 2747 |
+
en_cnh | Hakha Chin | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2748 |
+
en_crh | Crimean Tatar (Cyrillic script) | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2749 |
+
en_crh-Latn | Crimean Tatar (Latin script) | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2750 |
+
en_crs | Seselwa Creole French | Latn | Africa | | | 0 | 0 | 4001 |
|
| 2751 |
+
en_ctg | Chittagonian | Beng | Asia | | | 0 | 0 | 4001 |
|
| 2752 |
+
en_cv | Chuvash | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2753 |
+
en_dhd | Dhundari | Deva | Asia | | | 456 | 0 | 0 |
|
| 2754 |
+
en_din | Dinka | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2755 |
+
en_doi | Dogri | Deva | Asia | | | 454 | 0 | 4000 |
|
| 2756 |
+
en_dov | Dombe | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2757 |
+
en_dv | Dhivehi | Thaa | Asia | | | 0 | 0 | 4000 |
|
| 2758 |
+
en_dyu | Dyula | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2759 |
+
en_dz | Dzongkha | Tibt | Asia | | | 0 | 0 | 4001 |
|
| 2760 |
+
en_ee | Ewe | Latn | Africa | | | 130 | 863 | 4000 |
|
| 2761 |
+
en_efi | Efik | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2762 |
+
en_es | Spanish | Latn | Europe | | | 0 | 863 | 4000 |
|
| 2763 |
+
en_fa-AF | Dari | Arab | Asia | | | 0 | 0 | 4001 |
|
| 2764 |
+
en_ff | Fulfulde | Latn | Africa | | | 260 | 862 | 4000 |
|
| 2765 |
+
en_fj | Fijian | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2766 |
+
en_fo | Faroese | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2767 |
+
en_fon | Fon | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2768 |
+
en_fr | French | Latn | Europe | | | 0 | 0 | 4000 |
|
| 2769 |
+
en_fr-CA | French (Canada) | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2770 |
+
en_fur | Friulian | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2771 |
+
en_gaa | Ga | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2772 |
+
en_gn | Guarani | Latn | Americas | | | 0 | 863 | 4000 |
|
| 2773 |
+
en_gom | Goan Konkani | Deva | Asia | | | 0 | 0 | 4000 |
|
| 2774 |
+
en_grt-Latn | Garo (Latin script) | Latn | Asia | | | 457 | 0 | 0 |
|
| 2775 |
+
en_ha | Hausa | Latn | Africa | | | 584 | 863 | 0 |
|
| 2776 |
+
en_hi | Hindi | Deva | Asia | | | 0 | 0 | 4000 |
|
| 2777 |
+
en_hil | Hiligaynon | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2778 |
+
en_hne | Chhattisgarhi | Deva | Asia | | | 0 | 0 | 4001 |
|
| 2779 |
+
en_hoc-Wara | Ho (Warang Chiti script) | Wara | Asia | | | 457 | 0 | 0 |
|
| 2780 |
+
en_hrx | Hunsrik | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2781 |
+
en_iba | Iban | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2782 |
+
en_ig | Igbo | Latn | Africa | | | 391 | 863 | 0 |
|
| 2783 |
+
en_ilo | Ilocano | Latn | Asia | | | 0 | 0 | 4000 |
|
| 2784 |
+
en_iso | Isoko | Latn | Africa | | | 0 | 0 | 4001 |
|
| 2785 |
+
en_iu | Inuktut (Syllabics) | Cans | Americas | | | 0 | 0 | 4001 |
|
| 2786 |
+
en_iu-Latn | Inuktut (Latin) | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2787 |
+
en_jam | Jamaican Creole English | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2788 |
+
en_kaa | Kara-Kalpak | Cyrl | Asia | | | 0 | 0 | 4001 |
|
| 2789 |
+
en_kac | Kachin | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2790 |
+
en_kbd | Kabardian | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2791 |
+
en_kek | Kekchí | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2792 |
+
en_kfy | Kumaoni | Deva | Asia | | | 457 | 0 | 0 |
|
| 2793 |
+
en_kg | Kongo | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2794 |
+
en_kha | Khasi | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2795 |
+
en_ki | Kikuyu | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2796 |
+
en_kl | Kalaallisut | Latn | Americas | | | 0 | 863 | 4000 |
|
| 2797 |
+
en_kr | Kanuri | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2798 |
+
en_kri | Krio | Latn | Africa | | | 130 | 863 | 4026 |
|
| 2799 |
+
en_kru | Kurukh | Deva | Asia | | | 457 | 0 | 0 |
|
| 2800 |
+
en_ks | Kashmiri | Arab | Asia | | | 0 | 863 | 4001 |
|
| 2801 |
+
en_ks-Deva | Kashmiri (Devanagari script) | Deva | Asia | | | 0 | 863 | 4001 |
|
| 2802 |
+
en_ktu | Kituba (DRC) | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2803 |
+
en_kv | Komi | Cyrl | Asia | | | 0 | 0 | 4001 |
|
| 2804 |
+
en_lep | Lepcha | Lepc | Asia | | | 456 | 0 | 0 |
|
| 2805 |
+
en_lg | Luganda | Latn | Africa | | | 260 | 863 | 4000 |
|
| 2806 |
+
en_li | Limburgan | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2807 |
+
en_lif-Limb | Limbu (Limbu script) | Limb | Asia | | | 457 | 0 | 0 |
|
| 2808 |
+
en_lij | Ligurian | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2809 |
+
en_lmo | Lombard | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2810 |
+
en_ln | Lingala | Latn | Africa | | | 260 | 863 | 4000 |
|
| 2811 |
+
en_ltg | Latgalian | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2812 |
+
en_lu | Kiluba (Luba-Katanga) | Latn | Africa | | | 66 | 863 | 0 |
|
| 2813 |
+
en_lua | Tshiluba (Luba-Lulua) | Latn | Africa | | | 0 | 0 | 4001 |
|
| 2814 |
+
en_luo | Luo | Latn | Africa | | | 260 | 863 | 4001 |
|
| 2815 |
+
en_lus | Mizo | Latn | Asia | | | 0 | 863 | 4000 |
|
| 2816 |
+
en_mad | Madurese | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2817 |
+
en_mag | Magahi | Deva | Asia | | | 456 | 0 | 4001 |
|
| 2818 |
+
en_mai | Maithili | Deva | Asia | | | 0 | 0 | 4000 |
|
| 2819 |
+
en_mak | Makasar | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2820 |
+
en_mam | Mam | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2821 |
+
en_meo | Kedah Malay | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2822 |
+
en_mfe | Morisien | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2823 |
+
en_mg | Malagasy | Latn | Africa | | | 391 | 863 | 0 |
|
| 2824 |
+
en_mh | Marshallese | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2825 |
+
en_min | Minangkabau | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2826 |
+
en_mjl | Mandeali | Deva | Asia | | | 457 | 0 | 0 |
|
| 2827 |
+
en_mni-Mtei | Meiteilon (Manipuri) | Mtei | Asia | | | 0 | 863 | 4000 |
|
| 2828 |
+
en_mos | Mossi | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2829 |
+
en_ms-Arab | Malay (Jawi Script) | Arab | Asia | | | 0 | 0 | 4001 |
|
| 2830 |
+
en_mtr | Mewari | Deva | Asia | | | 457 | 0 | 0 |
|
| 2831 |
+
en_mwr | Marwari | Deva | Asia | | | 0 | 0 | 4001 |
|
| 2832 |
+
en_nd | North Ndebele | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2833 |
+
en_ndc-ZW | Ndau | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2834 |
+
en_ne | Nepali | Deva | Asia | | | 456 | 0 | 0 |
|
| 2835 |
+
en_new | Newari | Deva | Asia | | | 0 | 0 | 4001 |
|
| 2836 |
+
en_nhe | Eastern Huasteca Nahuatl | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2837 |
+
en_noe | Nimadi | Deva | Asia | | | 315 | 0 | 0 |
|
| 2838 |
+
en_nqo | Bambara (N'Ko script) | Nkoo | Africa | | | 0 | 0 | 0 |
|
| 2839 |
+
en_nr | South Ndebele | Latn | Africa | isiNdebele seSewula | | 66 | 863 | 4001 |
|
| 2840 |
+
en_nso | Sepedi | Latn | Africa | | | 130 | 863 | 4000 |
|
| 2841 |
+
en_nus | Nuer | Latn | Africa | | | 0 | 0 | 4001 |
|
| 2842 |
+
en_nv | Navajo | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2843 |
+
en_ny | Chichewa | Latn | Africa | | | 260 | 863 | 0 |
|
| 2844 |
+
en_oc | Occitan | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2845 |
+
en_om | Oromo | Latn | Africa | | | 391 | 863 | 4000 |
|
| 2846 |
+
en_os | Ossetian | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2847 |
+
en_pa-Arab | Lahnda Punjabi (Pakistan) | Arab | Asia | | | 0 | 863 | 4001 |
|
| 2848 |
+
en_pag | Pangasinan | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2849 |
+
en_pam | Pampanga | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2850 |
+
en_pap | Papiamento | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2851 |
+
en_pcm | Nigerian Pidgin | Latn | Africa | | | 130 | 863 | 4001 |
|
| 2852 |
+
en_pt-PT | Portuguese (Portugal) | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2853 |
+
en_qu | Quechua | Latn | Americas | | | 0 | 863 | 4000 |
|
| 2854 |
+
en_quc | K'iche' | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2855 |
+
en_rhg-Latn | Rohingya (Latin script) | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2856 |
+
en_rn | Rundi | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2857 |
+
en_rom | Romani | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2858 |
+
en_rw | Kinyarwanda | Latn | Africa | | | 260 | 863 | 0 |
|
| 2859 |
+
en_sa | Sanskrit | Deva | Asia | | | 0 | 863 | 4000 |
|
| 2860 |
+
en_sah | Yakut | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2861 |
+
en_sat | Santali (Ol Chiki script) | Olck | Asia | ᱥᱟᱱᱛᱟᱲᱤ | | 0 | 863 | 4001 |
|
| 2862 |
+
en_sat-Latn | Santali (Latin Script) | Latn | Asia | | | 0 | 863 | 4001 |
|
| 2863 |
+
en_scl | Shina | Arab | Asia | | | 457 | 0 | 0 |
|
| 2864 |
+
en_scn | Sicilian | Latn | Europe | | | 100 | 0 | 4001 |
|
| 2865 |
+
en_sd-Deva | Sindhi (Devanagari script) | Deva | Asia | | | 456 | 0 | 0 |
|
| 2866 |
+
en_se | Northern Sami | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2867 |
+
en_sg | Sango | Latn | Africa | | | 0 | 0 | 4001 |
|
| 2868 |
+
en_sgj | Surgujia | Deva | Asia | | | 356 | 0 | 0 |
|
| 2869 |
+
en_shn | Shan | Mymr | Asia | | | 0 | 0 | 4001 |
|
| 2870 |
+
en_sjp | Surjapuri | Deva | Asia | | | 299 | 0 | 0 |
|
| 2871 |
+
en_skr | Saraiki | Arab | Asia | | | 0 | 0 | 4001 |
|
| 2872 |
+
en_sn | Shona | Latn | Africa | | | 260 | 863 | 0 |
|
| 2873 |
+
en_so | Somali | Latn | Africa | | | 260 | 862 | 0 |
|
| 2874 |
+
en_spv | Sambalpuri | Orya | Asia | | | 457 | 0 | 0 |
|
| 2875 |
+
en_ss | Swati | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2876 |
+
en_st | Sesotho | Latn | Africa | | | 260 | 863 | 0 |
|
| 2877 |
+
en_sus | Susu | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2878 |
+
en_sw | Swahili | Latn | Africa | | | 584 | 863 | 0 |
|
| 2879 |
+
en_syl | Sylheti | Beng | Asia | | | 0 | 0 | 4001 |
|
| 2880 |
+
en_szl | Silesian | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2881 |
+
en_tcy | Tulu | Knda | Asia | | | 451 | 0 | 4001 |
|
| 2882 |
+
en_tet | Tetum | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2883 |
+
en_ti | Tigrinya | Ethi | Africa | | | 260 | 863 | 4000 |
|
| 2884 |
+
en_tiv | Tiv | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2885 |
+
en_tn | Tswana | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2886 |
+
en_to | Tonga (Tonga Islands) | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2887 |
+
en_tpi | Tok Pisin | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2888 |
+
en_trp | Kok Borok | Latn | Asia | | | 457 | 0 | 4001 |
|
| 2889 |
+
en_ts | Tsonga | Latn | Africa | | | 66 | 863 | 4000 |
|
| 2890 |
+
en_tum | Tumbuka | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2891 |
+
en_ty | Tahitian | Latn | Oceania | | | 0 | 0 | 4001 |
|
| 2892 |
+
en_tyv | Tuvinian | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2893 |
+
en_udm | Udmurt | Cyrl | Europe | | | 0 | 0 | 4001 |
|
| 2894 |
+
en_unr-Deva | Mundari (Devanagari script) | Deva | Asia | | | 457 | 0 | 0 |
|
| 2895 |
+
en_ve | Venda | Latn | Africa | | | 66 | 863 | 4001 |
|
| 2896 |
+
en_vec | Venetian | Latn | Europe | | | 0 | 0 | 4001 |
|
| 2897 |
+
en_war | Waray (Philippines) | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2898 |
+
en_wbr | Wagdi | Deva | Asia | | | 455 | 0 | 0 |
|
| 2899 |
+
en_wo | Wolof | Latn | Africa | | | 260 | 863 | 4001 |
|
| 2900 |
+
en_xh | Xhosa | Latn | Africa | | | 260 | 863 | 0 |
|
| 2901 |
+
en_xnr | Kangri | Deva | Asia | | | 457 | 0 | 0 |
|
| 2902 |
+
en_xsr-Tibt | Sherpa (Tibetan script) | Tibt | Asia | | | 457 | 0 | 0 |
|
| 2903 |
+
en_yo | Yoruba | Latn | Africa | | | 584 | 863 | 0 |
|
| 2904 |
+
en_yua | Yucateco | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2905 |
+
en_yue | Cantonese | Hant | Asia | | | 584 | 863 | 4201 |
|
| 2906 |
+
en_zap | Zapotec | Latn | Americas | | | 0 | 0 | 4001 |
|
| 2907 |
+
en_zu | Zulu | Latn | Africa | | | 260 | 863 | 0 |
|
| 2908 |
+
en_zza | Zaza | Latn | Asia | | | 0 | 0 | 4001 |
|
| 2909 |
+
sw_ar | Arabic | Arab | Asia | | | 330 | 0 | 0 |
|
| 2910 |
+
sw_zh | Mandarin Chinese | Hans | Asia | | | 330 | 0 | 0 |
|
| 2911 |
+
yue_en | English | Latn | Europe | | | 584 | 863 | 3983 |
|
| 2912 |
+
yue_zh | Mandarin Chinese | Hans | Asia | | | 0 | 0 | 4201 |
|
| 2913 |
+
yue_zh-Hant | Taiwanese Mandarin | Hant | Asia | | | 0 | 0 | 4201 |
|
| 2914 |
+
zh-Hant_yue | Cantonese | Hant | Asia | | | 0 | 0 | 4031 |
|
| 2915 |
+
zh_yue | Cantonese | Hant | Asia | | | 0 | 0 | 4031 |
|
| 2916 |
|
| 2917 |
+
## Notes on language codes and names
|
| 2918 |
|
| 2919 |
We use the backwards-compatible BCP-47 language code convention. Several of
|
| 2920 |
these languages have various alternate code also in use, including:
|
|
|
|
| 2928 |
* **qu** Another supercode issue. The Quechua variety this data is closest to
|
| 2929 |
is likely `quy`, though a language expert should opine
|
| 2930 |
* **pa-Arab** is also referred to as `lah`
|
| 2931 |
+
* **chm**: we use `chm` for Meadow Mari, although `chm` is properly the superset, and `mhr` is the individual code for Meadow Mari.
|
| 2932 |
+
* **rom**: Similarly, we use `rom` for all Romani dialects, `ak` for Twi (see note above), and `ff` for all Fulfulde dialects.
|
| 2933 |
+
* **fa-AF,ar-MA**: We use `fa-AF` and `ar-MA` for Dari and Darija respectively. The specific codes `prs` and `ary` are probably more precise.
|
| 2934 |
+
* **ndc-ZW**: `ndc-ZW` is probably redundant; `ndc` alone refers to Ndau.
|
| 2935 |
+
|
| 2936 |
+
|
| 2937 |
+
Please note that the language name can be a point of considerable tension, and
|
| 2938 |
+
there is rarely a "right" name for a language! This said, here are some
|
| 2939 |
+
non-exhaustive notes that went into the names in the second column above, based
|
| 2940 |
+
on conversations with speakers. We welcome more comments from the community:
|
| 2941 |
+
|
| 2942 |
+
* **ak:** Calling this language "Twi" instead of "Akan" here is reflective of
|
| 2943 |
+
the corresponding translation models active on Google Translate. These are
|
| 2944 |
+
likely to mix different Akan dialects, including Twi proper, Fante, Akuapem Twi, etc.. Therefore, from a linguistic perspective, one would call this "Akan" and use the corresponding code "ak". However, this is a more academic word for the language; according to native speaker consultants, "Akan" is usually used to describe the ethnic group rather than the language, so they suggested simply calling this "Twi", which is most likely accurate for the large majority of model outputs.
|
| 2945 |
+
* **ckb:** *Central Kurdish* is a more academic name that may not be as widely
|
| 2946 |
+
known to speakers of the language, so *Sorani* is preferred -- though they
|
| 2947 |
+
are both acceptable.
|
| 2948 |
+
* **dv:** There is not a consistent romanization of this language, and both
|
| 2949 |
+
*Divehi* and *Dhivehi* are common. The first consonant in this language is a
|
| 2950 |
+
dental, non-aspirated *d*. When transliterating Indian language names there is
|
| 2951 |
+
no standard about how to represent this sound. In the south, "dh" usually
|
| 2952 |
+
refers to the dental consonant; in the North, it just as often refers to the
|
| 2953 |
+
aspirated one. However, in both cases, it is common that "d" alone would
|
| 2954 |
+
represent the retroflex consonant, so it is possible that *Divehi* would be
|
| 2955 |
+
mispronounced more often.
|
| 2956 |
+
* **gom:** Technically, `gom` is the code for *Goan* Konkani only. Other
|
| 2957 |
+
varieties of Konkani, like Mangalore Konkani (which oddly enough does not
|
| 2958 |
+
have an ISO code), are mutually unintelligible! There is certainly dispute
|
| 2959 |
+
about whether to treat Konkani as one language or not.
|
| 2960 |
+
* **ilo:** Native speakers give a variety of different answers around the
|
| 2961 |
+
correct name. Some say that *Ilocano* is more formal, whereas *Ilokano* is
|
| 2962 |
+
more informal; some say that both of these actually refer to the people, and
|
| 2963 |
+
that *Iloko* is the only correct term for the language.
|
| 2964 |
+
* **lg:** Speakers usually refer to this language as "Luganda"; linguists sometimes call it
|
| 2965 |
+
"Ganda", but like Sepedi, this is not usually used by people
|
| 2966 |
+
familiar with the language.
|
| 2967 |
+
* **lus:** There is agreement that *Lushai* is an obsolete colonial name and
|
| 2968 |
+
should not be used. *Mizo* is the preferred name.
|
| 2969 |
+
* **mni-Mtei:** The conventional and most common name for this language is
|
| 2970 |
+
*Manipuri*. However, in reality, this is only the language of one of the
|
| 2971 |
+
many ethnicities in Manipur, so that name can be seen as offensive to some.
|
| 2972 |
+
Therefore, we use the endonym *Meiteilon*, with the conventional name
|
| 2973 |
+
(Manipuri) in parentheses.
|
| 2974 |
+
* **nso:** Although a linguist might say that Sepedi is a dialect of Northern
|
| 2975 |
+
Sotho, speakers often say that they call the whole language Sepedi, and that
|
| 2976 |
+
*Northern Sotho* has offensive colonial overtones. Others, however, may
|
| 2977 |
+
disagree with this. The endonym *Sesotho sa Leboa* is not uncommon. There is
|
| 2978 |
+
agreement that no-one uses *Pedi*.
|
| 2979 |
+
* **qu:** These translations are technically in *Southern* Quechua. However,
|
| 2980 |
+
this is a point of some contention. Some will opine that "Quechua" is not a
|
| 2981 |
+
language but a family of languages. Others, however, say that they all feel
|
| 2982 |
+
like one language, and that the regional differences are slight, and it's
|
| 2983 |
+
easy for Peruvians, Bolivians, and even Ecuadorians to understand each
|
| 2984 |
+
other. The majority opinion among the naive speakers we talked to seemed to
|
| 2985 |
+
be closer to the latter, so we have opted for the name "Quechua".
|
| 2986 |
+
* **ti:** In Tigray *Tigrigna* is the most common Romanization, and in Eritrea
|
| 2987 |
+
*Tigrinya* is.
|
| 2988 |
+
* **zap:** Linguists usually talk about the Zapotec *languages*, rather than
|
| 2989 |
+
considering Zapotec a single language. However, just as with Quechua above,
|
| 2990 |
+
conversations with a native speaker indicated that they perceived Zapotec
|
| 2991 |
+
as one language, with a variety of local variants that are all easy enough
|
| 2992 |
+
to understand (excepting one or two). A broader survey would help resolve this question.
|
| 2993 |
+
* Those languages in this dataset that are not on Google Translate have yet to
|
| 2994 |
+
undergo a more careful name review; if you see a name in the list above that
|
| 2995 |
+
you want to correct, please contact us!
|
| 2996 |
+
|
| 2997 |
+
Special note on Bantu languages: Bantu language names have a special noun-class
|
| 2998 |
+
prefix for languages, like <i>xi</i>Tsonga, <i>se</i>Pedi, <i>ki</i>Swahili, <i>isi</i>Zulu,
|
| 2999 |
+
<i>lu</i>Ganda, <i>si</i>Swati, <i>isi</i>Ndebele, <i>tshi</i>Venda, and so on. However, there is no
|
| 3000 |
+
pattern as to whether this prefix is used in the conventional English form of
|
| 3001 |
+
the name. For instance, Xhosa usually leaves off the <i>isi-</i> in English, whereas
|
| 3002 |
+
Lingala almost never leaves off the <i>li-</i>. Others, like (xi)Tsonga, seem to be
|
| 3003 |
+
frequently used in both forms. Linguistic resources occasionally strip the
|
| 3004 |
+
prefixes, yielding names like "Ganda", "Pedi" and so on, perhaps out of a desire
|
| 3005 |
+
for consistent handling of the noun class prefix. (Even the *ba* in *Bantu* is
|
| 3006 |
+
such a prefix, so these languages are sometimes called the *Ntu* languages!)
|
| 3007 |
|
| 3008 |
For convenience, here is a mapping from codes used by Flores and SMOL:
|
| 3009 |
|
| 3010 |
+
|
| 3011 |
Flores code | BCP-47 code
|
| 3012 |
----------- | -----------
|
| 3013 |
eng_Latn | en
|
|
|
|
| 3227 |
Tibetan (`bo`) caused notable regressions in `ChrF`, suggesting that they
|
| 3228 |
are low-quality, though this could also be an issue with training or
|
| 3229 |
evaluation.
|
| 3230 |
+
* The Manx (`gv`) split of GATITOS that was originally released was spurious.
|
| 3231 |
+
It has been removed.
|
| 3232 |
+
|
| 3233 |
+
|
| 3234 |
+
|
| 3235 |
+
### Contact
|
| 3236 |
|
| 3237 |
+
We welcome contributions, corrections, and suggestions, especially from speakers
|
| 3238 |
+
of these languages. If you speak any of these languages and notice quality
|
| 3239 |
+
issues, please file bugs on this repo or contact
|
| 3240 |
+
{icaswell, eknielsen}꩜google.com.
|
| 3241 |
Another especially useful piece of information is whether the correct
|
| 3242 |
+
orthographies are used in these deliveries. This
|
| 3243 |
+
repository may be updated sporadically with corrections or additional
|
| 3244 |
+
translations.
|
| 3245 |
|
| 3246 |
+
|
| 3247 |
## License
|
| 3248 |
|
| 3249 |
This data is released under cc-by-4.0
|