deu-ara

• source group: 德语

• target group: 阿拉伯语

• OPUS readme: deu-ara

• model: transformer-align

• source language(s): 德语

• target language(s): afb apc 阿拉伯语 阿拉伯语_拉丁字母 arq arz

• model: transformer-align

• pre-processing: 规范化 + SentencePiece (spm32k,spm32k)

• 需要以形式为 >>id<< (id = 有效的目标语言ID) 的句子初始语言标记

• 下载原始权重: opus-2020-07-03.zip

• 测试集翻译结果: opus-2020-07-03.test.txt

• 测试集分数: opus-2020-07-03.eval.txt

基准性能

系统信息:

• hf_name: deu-ara

• source_languages: deu

• target_languages: 阿拉伯语

• opus_readme_url: https://github.com/Helsinki-NLP/Tatoeba-Challenge/tree/master/models/deu-ara/README.md

• original_repo: Tatoeba-Challenge

• tags: ['translation']

• languages: ['de', 'ar']

• src_constituents: {'deu'}

• tgt_constituents: {'apc', '阿拉伯语', 'arq_拉丁字母', 'arq', 'afb', '阿拉伯语_拉丁字母', 'apc_拉丁字母', 'arz'}

• src_multilingual: False

• tgt_multilingual: False

• prepro: 规范化 + SentencePiece (spm32k,spm32k)

• url_model: https://object.pouta.csc.fi/Tatoeba-MT-models/deu-ara/opus-2020-07-03.zip

• url_test_set: https://object.pouta.csc.fi/Tatoeba-MT-models/deu-ara/opus-2020-07-03.test.txt

• src_alpha3: deu

• tgt_alpha3: 阿拉伯语

• short_pair: de-ar

• chrF2_score: 0.486

• bleu: 19.7

• brevity_penalty: 0.993

• ref_len: 6324.0

• src_name: 德语

• tgt_name: 阿拉伯语

• train_date: 2020-07-03

• src_alpha2: de

• tgt_alpha2: ar

• prefer_old: False

• long_pair: deu-ara

• helsinki_git_sha: 480fcbe0ee1bf4774bcbe6226ad9f58e63f6c535

• transformers_git_sha: 2207e5d8cb224e954a7cba69fa4ac2309e9ff30b

• port_machine: brutasse

• port_time: 2020-08-21-14:41