tednmt
load_data
¶
Load and return the neural machine translation dataset from TED talks.
Parameters:
Name | Type | Description | Default |
---|---|---|---|
root_dir
|
Optional[str]
|
The path to store the downloaded data. When |
None
|
translate_option
|
str
|
Options for translation languages. Available options are: "az_to_en", "az_tr_to_en", "be_ru_to_en", "be_to_en", "es_to_pt", "fr_to_pt", "gl_pt_to_en", "gl_to_en", "he_to_pt", "it_to_pt", "pt_to_en", "ru_to_en", "ru_to_pt", and "tr_to_en". |
'az_to_en'
|
Returns:
Type | Description |
---|---|
Tuple[NumpyDataset, NumpyDataset, NumpyDataset]
|
(train_data, eval_data, test_data) |