mteb icon indicating copy to clipboard operation
mteb copied to clipboard

Add a Benchmark for Asian Languages

Open KennethEnevoldsen opened this issue 1 year ago • 3 comments

Linguistic Families and Proposed Languages:

East Asian Languages

  • [x] Chinese (Mandarin) - cmn
  • [x] Cantonese - yue (#370)
  • [x] Japanese - jpn
  • [x] Korean - kor
  • [ ] Mongolian - mon

South Asian Languages

Indic Languages:

  • [x] Hindi - hin

  • [x] Bengali - ben

  • [x] Punjabi - pan

  • [x] Marathi - mar

  • [x] Gujarati - guj

  • [x] Urdu - urd

  • [x] Nepali - nep

  • [x] Sinhala - sin

  • [x] Tamil - tam

  • [x] Telugu - tel

  • [x] Kannada - kan

  • [x] Malayalam - mal

  • Dravidian Languages:

    • Included above (Tamil, Telugu, Kannada, Malayalam)

Southeast Asian Languages

  • Austronesian Languages:
    • [x] Indonesian - ind
    • [x] Filipino - fil (#472 )
    • [ ] Malay - msa
    • [x] Javanese - jav
  • Tai-Kadai Languages:
    • [x] Thai - tha
    • [x] Lao - lao
  • Austroasiatic Languages:
    • [x] Vietnamese - vie (see #364)
    • [x] Khmer - khm
  • [ ] Burmese - mya

Central Asian Languages

  • Turkic Languages:
    • [x] Kazakh - kaz
    • [ ] Uzbek - uzb
    • [ ] Turkmen - tkm
    • [x] Kyrgyz - kir
    • [x] Uighur - uig

West Asian (Middle Eastern) Languages

  • Semitic Languages:
    • [x] Arabic - ara
    • [x] Hebrew - heb
  • Iranian Languages:
    • [x] Persian - fas
    • [x] Kurdish - kur
    • [ ] Pashto - pus
    • [x] Dari - prs

Note this list does not claim to be comprehensive, do feel free to add to the list.

KennethEnevoldsen avatar Apr 14 '24 10:04 KennethEnevoldsen

I will take a stab at a Bengali benchmark together with a colleague of mine 👍

rasdani avatar Apr 15 '24 07:04 rasdani

Wonderful @rasdani feel free to create an issue on this as well so that others can see that you are working on it.

KennethEnevoldsen avatar Apr 15 '24 08:04 KennethEnevoldsen

I created PRs for Indonesian languages (at least 10+ additions from 2 corpus) and African language. Once, they are approved, I can add the languages to the list.

gentaiscool avatar Jun 13 '24 22:06 gentaiscool