Closes #512 | Add/Update Dataloader QED

[patrickamadeus]

Closes #512

Checkbox

• [x] Confirm that this PR is linked to the dataset issue.
• [x] Create the dataloader script seacrowd/sea_datasets/{my_dataset}/{my_dataset}.py (please use only lowercase and underscore for dataset folder naming, as mentioned in dataset issue) and its __init__.py within {my_dataset} folder.
• [x] Provide values for the _CITATION, _DATASETNAME, _DESCRIPTION, _HOMEPAGE, _LICENSE, _LOCAL, _URLs, _SUPPORTED_TASKS, _SOURCE_VERSION, and _SEACROWD_VERSION variables.
• [x] Implement _info(), _split_generators() and _generate_examples() in dataloader script.
• [x] Make sure that the BUILDER_CONFIGS class attribute is a list with at least one SEACrowdConfig for the source schema and one for a seacrowd schema.
• [x] Confirm dataloader script works with datasets.load_dataset function.
• [x] Confirm that your dataloader script passes the test suite run with python -m tests.test_seacrowd seacrowd/sea_datasets/<my_dataset>/<my_dataset>.py or python -m tests.test_seacrowd seacrowd/sea_datasets/<my_dataset>/<my_dataset>.py --subset_id {subset_name_without_source_or_seacrowd_suffix}.
• [.] If my dataset is local, I have provided an output of the unit-tests in the PR (please copy paste). This is OPTIONAL for public datasets, as we can test these without access to the data files.

T2T

malay-thai

malay-burmese

english-javanese

indonesian-javanese

SSP

malay-burmese

indonesian-javanese

NB:

• 4 samples tested out of 28 lang pairs for T2T, and 2 samples out of 56 cases for SSP
• For pairs with en or id, first line is outlier because it shows the document title (might be misinterpreted as an error as depicted from last 2 examples)
• ~Currently involving only Tasks.MACHINE_TRANSLATION, will add task addition is needed for language modeling.~

[patrickamadeus]

will add ssp schema within this week after discussing with @SamuelCahyawijaya

[patrickamadeus]

Updated SSP approach with (some) retests!

cc @ljvmiranda921 @holylovenia @SamuelCahyawijaya

MT jv-ms

SSP jv

SSP vi

SSP tl

[holylovenia]

Checking the data source, we can add the ff. SEA languages to the dataloader:

• Cebuano
• Filipino (fil): Normally we consider tl and fil to be the same, but the monolingual documents of fil and tl here are different.
• Khmer
• Lao
• Madurese
• Pampanga

@holylovenia Please also update the datasheet accordingly, thanks!

I'd like to confirm before I rectify the language list.

So the complete list of the languages is ["eng", "vie", "tha", "mya", "jav", "ind", "tgl", "zlm", "ceb", "fil", "khm", "lao", "mad", "pam"], right? I'm assuming Pampanga == Kapampangan (pam).

cc: @elyanah-aco @ljvmiranda921

[elyanah-aco]

So the complete list of the languages is ["eng", "vie", "tha", "mya", "jav", "ind", "tgl", "zlm", "ceb", "fil", "khm", "lao", "mad", "pam"], right? I'm assuming Pampanga == Kapampangan (pam).

Yes this is correct, thanks!

[holylovenia]

So the complete list of the languages is ["eng", "vie", "tha", "mya", "jav", "ind", "tgl", "zlm", "ceb", "fil", "khm", "lao", "mad", "pam"], right? I'm assuming Pampanga == Kapampangan (pam).

Yes this is correct, thanks!

Done!

[holylovenia]

Hi @patrickamadeus, a friendly reminder to address @elyanah-aco's suggestions.

[patrickamadeus]

Thank you for the review! Updated the lang list. @elyanah-aco @holylovenia

[patrickamadeus]

Thank you for the nitpick! It's done in the latest commit! @elyanah-aco

[holylovenia]

Thank you for the nitpick! It's done in the latest commit! @elyanah-aco

Hi @elyanah-aco, may I know if you're ready to approve this PR? If all your concerns have been addressed by @patrickamadeus, I'm inclined to get this PR merged.