README.md

---
dataset_info:
  features:
  - name: th
    dtype: string
  - name: en
    dtype: string
  - name: sco
    dtype: float64
  splits:
  - name: train
    num_bytes: 8681293302
    num_examples: 26934996
  download_size: 4582682834
  dataset_size: 8681293302
configs:
- config_name: default
  data_files:
  - split: train
    path: data/train-*
task_categories:
- translation
language:
- th
- en
---

# `quickmt` th-en Training Corpus

Contains the following datasets downloaded with `mtdata` after deduplication and [basic filtering](https://github.com/quickmt/quickmt/blob/main/quickmt/scripts/clean.py) with `quickmt`:

* Statmt-ccaligned-1-eng-tha_TH
* Neulab-tedtalks_train-1-eng-tha
* Neulab-tedtalks_test-1-eng-tha
* Neulab-tedtalks_dev-1-eng-tha
* ELRC-wikipedia_health-1-eng-tha
* ELRC-hrw_dataset_v1-1-eng-tha
* OPUS-ccaligned-v1-eng-tha
* OPUS-elrc_3048_wikipedia_health-v1-eng-tha
* OPUS-elrc_wikipedia_health-v1-eng-tha
* OPUS-elrc_2922-v1-eng-tha
* OPUS-gnome-v1-eng-tha
* OPUS-hplt-v2-eng-tha
* OPUS-kde4-v2-eng-tha
* OPUS-neulab_tedtalks-v1-eng-tha
* OPUS-opensubtitles-v2016-eng-tha
* OPUS-opensubtitles-v2018-eng-tha
* OPUS-opensubtitles-v2024-eng-tha
* OPUS-paracrawl_bonus-v9-eng-tha
* OPUS-qed-v2.0a-eng-tha
* OPUS-scb_mt_en_th-v1.0-eng-tha
* OPUS-ted2020-v1-eng-tha
* OPUS-tanzil-v1-eng-tha
* OPUS-tatoeba-v2-eng-tha
* OPUS-tatoeba-v20190709-eng-tha
* OPUS-tatoeba-v20200531-eng-tha
* OPUS-tatoeba-v20201109-eng-tha
* OPUS-tatoeba-v20210310-eng-tha
* OPUS-tatoeba-v20210722-eng-tha
* OPUS-tatoeba-v20220303-eng-tha
* OPUS-tatoeba-v20230412-eng-tha
* OPUS-ubuntu-v14.10-eng-tha
* OPUS-xlent-v1-eng-tha
* OPUS-xlent-v1.1-eng-tha
* OPUS-xlent-v1.2-eng-tha
* OPUS-bible_uedin-v1-eng-tha
* OPUS-tldr_pages-v20230829-eng-tha
* OPUS-wikimedia-v20210402-eng-tha
* OPUS-wikimedia-v20230407-eng-tha
* OPUS-opus100_train-1-eng-tha
* OPUS-opus100_dev-1-eng-tha
* OPUS-opus100_test-1-eng-tha
* Google-wmt24pp-1-eng-tha_TH