Datasets
Dataset
last-translation-benchmark
by zouhar
Last Translation Benchmark Abstract: For scientific progress, we need benchmarks that test the limits of state-of-the-art models, and evaluation methods that inform us about failure cases. Standard benchmarks for machine translation evaluation are often eit…
CC-BY-4.0translation 1 01K < n < 10K rows
published 31 Aug 2026
Preview
First 5 rows of default / train, from the Hugging Face dataset viewer.
| id | source_text | source_lang | target_lang | source_lang_iso | target_lang_iso | source_instructions | source_media | attribution | translations | verification_rules | linguistics | tags |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 8 | what's the difference between jail and prison? | English | Czech | eng | ces | [{"model":"Lara","translation":"jaký je rozdíl mezi vězením a věznicí?","verified":[false,true]},{"model":"Google Translate","translation":"jaký je rozdíl mezi vězením a vězením?","verified":[false,true]},{"model":"Gemini 2.5 Flash","translation":"Jaký je rozdíl mezi vězením a káznicí?","verified":… | ["The two main words that refer to jail/prison should not be entirely identical, such as \"vězením\" a \"věznicí\" which are pure synonyms. However, \"vazební věznicí\" and \"věznicí\" are distinct.","The words should refer to jail/prison, so věznice, káznice, žalář, vazební cela, vazební věznice, … | ["Lexical Gradation","Semantic/Lexical"] | ["LTBv1"] | |||
| 9 | ما در پیاله عکس رخ یار دیده ایم | Persian | English | fas | eng | [{"model":"Google Translate","translation":"We have seen his face in the photo cup","verified":[false,true,false]},{"model":"Gemini 2.5 Flash","translation":"ما در پیاله عکس رخ یار دیدهایم","verified":[true,false,false]},{"model":"Gemma 4","translation":"We have seen the reflection of the beloved'… | ["Translation can't contain exactly \"cup\".","The translation should be in English","The translation should contain \"reflection\" or \"picture.\""] | ["Poetry","Polysemy","Style Preservation","Semantic/Lexical"] | ["LTBv1"] | |||
| 10 | Il fattone si stava facendo una spada nel parco | Italian | English | ita | eng | [{"model":"Lara","translation":"The big guy was making a sword in the park","verified":[false]},{"model":"Google Translate","translation":"The stoner was having a sword in the park","verified":[false]},{"model":"Gemini 2.5 Flash","translation":"The stoner was having a blowjob in the park","verified… | ["Should contain 'syringe' or 'drugs', not 'sword'"] | ["Slang","Metaphor","Collocation","Semantic/Lexical"] | ["LTBv1","LTBv1-eval"] | |||
| 12 | مدیریت بحران منطقه | Farsi | English | fas | eng | [{"model":"Gemini 2.5 Flash","translation":"Crisis Management Region","verified":[true,true,false]},{"model":"Gemma 4","translation":"Regional crisis management","verified":[false,true,true]},{"model":"Llama 4 Maverick","translation":"Crisis management in the region","verified":[true,false,true]},{… | ["Translation can't contain \"regional\".","Translation can't contain \"in\" or \"for\".","In the translation, \"management\" should be the main subject."] | ["Target Gap: Morph to Words","Collocation","Syntactic","Morphological","Semantic/Lexical"] | ["LTBv1"] | |||
| 13 | صورت تو ملک جمال و جمال ملک است | Farsi | English | fas | eng | [{"model":"Gemini 2.5 Flash","translation":"Your face is the land of beauty and beauty is the land.","verified":[false,true]},{"model":"Gemma 4","translation":"Your face is the kingdom of beauty and the beauty of the kingdom.","verified":[true,true]},{"model":"Llama 4 Scout","translation":" face is… | ["Translation can't contain exactly \"beauty is\".","Translation has to contain exactly \"Your face\"."] | ["Gardenpath","Poetry","Syntactic","Orthography"] | ["LTBv1"] |
Cite this
CC-BY-4.0textCite
Tags
task_categories:translationtask_categories:text-generationlanguage:enlanguage:zhlanguage:delanguage:frlanguage:eslanguage:jalanguage:rulanguage:itlanguage:ptlanguage:nllanguage:pllanguage:arlanguage:kolanguage:vilanguage:trlanguage:cslanguage:svlanguage:uklanguage:falanguage:idlanguage:rolanguage:calanguage:filanguage:hulanguage:dalanguage:nolanguage:helanguage:ellanguage:thlanguage:hilanguage:sklanguage:hrlanguage:ltlanguage:etlanguage:bnlanguage:urlanguage:talanguage:te