|
--- |
|
license: cc-by-sa-4.0 |
|
task_categories: |
|
- text-classification |
|
language: |
|
- mt |
|
tags: |
|
- nlp |
|
- nli |
|
- rte |
|
- maltese |
|
- malti |
|
size_categories: |
|
- 100K<n<1M |
|
configs: |
|
- config_name: snli |
|
data_files: "snli_mt.csv" |
|
sep: ";" |
|
- config_name: mnli |
|
data_files: "mnli_mt.csv" |
|
sep: ";" |
|
--- |
|
|
|
|
|
# Dataset Card for the SMNLI-MT |
|
|
|
The SMNLI-MT datasets are machine-translated versions of the Stanford NLI and MultiNLI datasets in Maltese. |
|
|
|
## Dataset Details |
|
|
|
### Dataset Description |
|
|
|
The datasets were translated using the Google Cloud Translate as part of the initial exploration of NLI in the Maltese language. |
|
|
|
|
|
- **Curated by:** [Matthew Darmanin] |
|
- **Language(s) (NLP):** Maltese |
|
- **License:** [CC 4.0] |
|
|
|
## Dataset Structure |
|
|
|
The datasets are in the form of CSV files, delimited by semi-colons (;) and encoded in UTF-8. |
|
|
|
Each dataset contains the following columns: |
|
- **id** - the index of the sentence pair |
|
- **premise** - the premise sentence |
|
- **hypothesis** - the hypothesis sentence |
|
- **gold label** - the consensus label assigned in the original datasets; either _entailment_, _contradiction_ or _neutral_ |
|
|
|
|
|
## Dataset Card Contact |
|
|
|
E-mail: [email protected] |
