Datasets:

leoli04

/

github-HF-datasets-issues

like 0

https://api.github.com/repos/huggingface/datasets/issues/84

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/84/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/84/comments

https://api.github.com/repos/huggingface/datasets/issues/84/events

https://github.com/huggingface/datasets/pull/84

MDExOlB1bGxSZXF1ZXN0NDE3MjAxODcz

[TedHrLr] add left dummy data

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/84/timeline

https://api.github.com/repos/huggingface/datasets/issues/83

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/83/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/83/comments

https://api.github.com/repos/huggingface/datasets/issues/83/events

https://github.com/huggingface/datasets/pull/83

MDExOlB1bGxSZXF1ZXN0NDE2ODkyOTUz

New datasets

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/83/timeline

https://api.github.com/repos/huggingface/datasets/issues/82

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/82/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/82/comments

https://api.github.com/repos/huggingface/datasets/issues/82/events

https://github.com/huggingface/datasets/pull/82

MDExOlB1bGxSZXF1ZXN0NDE2ODQ1Njc5

[Datasets] add ted_hrlr

closed

CONTRIBUTOR

@thomwolf - After looking at `xnli` I think it's better to leave the translation features and add a `translation` key to make them work in our framework. The result looks like this:  you can see that each split has a `translation` key which value is the nlp.features.Translation object. That's a simple change. If it's ok for you, I will add dummy data for the other configs and treat the other translation scripts in the same way.

https://api.github.com/repos/huggingface/datasets/issues/82/timeline

https://api.github.com/repos/huggingface/datasets/issues/81

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/81/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/81/comments

https://api.github.com/repos/huggingface/datasets/issues/81/events

https://github.com/huggingface/datasets/pull/81

MDExOlB1bGxSZXF1ZXN0NDE2ODM1NzE1

add tests

closed

MEMBER

Tests for py_utils functions and for the BaseReader used to read from arrow and parquet. I also removed unused utils functions.

https://api.github.com/repos/huggingface/datasets/issues/81/timeline

https://api.github.com/repos/huggingface/datasets/issues/80

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/80/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/80/comments

https://api.github.com/repos/huggingface/datasets/issues/80/events

https://github.com/huggingface/datasets/pull/80

MDExOlB1bGxSZXF1ZXN0NDE2ODMwNjk3

Add nbytes + nexamples check

closed

MEMBER

### Save size and number of examples Now when you do `save_checksums`, it also create `cached_sizes.txt` right next to the checksum file. This new file stores the bytes sizes and the number of examples of each split that has been prepared and stored in the cache. Example: ``` # Cached sizes: <full_config_name> <num_bytes> <num_examples> hansards/house/1.0.0/test 22906629 122290 hansards/house/1.0.0/train 191459584 947969 hansards/senate/1.0.0/test 5711686 25553 hansards/senate/1.0.0/train 40324278 182135 ``` ### Check processing output If there is a `caches_sizes.txt`, then each time we run `download_and_prepare` it will make sure that the sizes match. You can set `ignore_checksums=True` if you don't want that to happen. ### Fill Dataset Info All the split infos and the checksums are now stored correctly in DatasetInfo after `download_and_prepare` ### Check space on disk before running `download_and_prepare` Check if the space is lower than the sum of the sizes of the files in `checksums.txt` and `cached_files.txt`. This is not ideal though as it considers the files for all configs. TODO: A better way to do it would be to have save the `DatasetInfo` instead of the `checksums.txt` and `cached_sizes.txt`, in order to have one file per dataset config (and therefore consider only the sizes of the files for one config and not all of them). It can also be the occasion to factorize all the `download_and_prepare` verifications. Maybe next PR ?

https://api.github.com/repos/huggingface/datasets/issues/80/timeline

https://api.github.com/repos/huggingface/datasets/issues/79

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/79/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/79/comments

https://api.github.com/repos/huggingface/datasets/issues/79/events

https://github.com/huggingface/datasets/pull/79

MDExOlB1bGxSZXF1ZXN0NDE2ODI5NzMy

[Convert] add new pattern

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/79/timeline

https://api.github.com/repos/huggingface/datasets/issues/78

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/78/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/78/comments

https://api.github.com/repos/huggingface/datasets/issues/78/events

https://github.com/huggingface/datasets/pull/78

MDExOlB1bGxSZXF1ZXN0NDE2ODIwNzU5

[Tests] skip beam dataset tests for now

closed

CONTRIBUTOR

For now we will skip tests for Beam Datasets

https://api.github.com/repos/huggingface/datasets/issues/78/timeline

https://api.github.com/repos/huggingface/datasets/issues/77

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/77/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/77/comments

https://api.github.com/repos/huggingface/datasets/issues/77/events

https://github.com/huggingface/datasets/pull/77

MDExOlB1bGxSZXF1ZXN0NDE2NzQwMjAz

New datasets

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/77/timeline

https://api.github.com/repos/huggingface/datasets/issues/76

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/76/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/76/comments

https://api.github.com/repos/huggingface/datasets/issues/76/events

https://github.com/huggingface/datasets/pull/76

MDExOlB1bGxSZXF1ZXN0NDE2NjYyMTk2

pin flake 8

closed

CONTRIBUTOR

Flake 8's new version does not like our format. Pinning the version for now.

https://api.github.com/repos/huggingface/datasets/issues/76/timeline

https://api.github.com/repos/huggingface/datasets/issues/75

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/75/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/75/comments

https://api.github.com/repos/huggingface/datasets/issues/75/events

https://github.com/huggingface/datasets/pull/75

MDExOlB1bGxSZXF1ZXN0NDE2NjE0MzU1

WIP adding metrics

closed

MEMBER

Adding the following metrics as identified by @mariamabarham: 1. BLEU: BiLingual Evaluation Understudy: https://github.com/tensorflow/nmt/blob/master/nmt/scripts/bleu.py, https://github.com/chakki-works/sumeval/blob/master/sumeval/metrics/bleu.py (multilingual) 2. GLEU: Google-BLEU: https://github.com/cnap/gec-ranking/blob/master/scripts/compute_gleu 3. Sacrebleu: https://pypi.org/project/sacrebleu/1.4.8/ (pypi package), https://github.com/mjpost/sacrebleu (github implementation) 4. ROUGE: Recall-Oriented Understudy for Gisting Evaluation: https://github.com/google-research/google-research/tree/master/rouge, https://github.com/chakki-works/sumeval/blob/master/sumeval/metrics/rouge.py (multilingual) 5. Seqeval: https://github.com/chakki-works/seqeval (github implementation), https://pypi.org/project/seqeval/0.0.12/ (pypi package) 6. Coval: coreference evaluation package for the CoNLL and ARRAU datasets https://github.com/ns-moosavi/coval 7. SQuAD v1 evaluation script 8. SQuAD V2 evaluation script: https://worksheets.codalab.org/rest/bundles/0x6b567e1cf2e041ec80d7098f031c5c9e/contents/blob/ 9. GLUE 10. XNLI Not now: 1. Perplexity: https://github.com/allenai/allennlp/blob/master/allennlp/training/metrics/perplexity.py 2. Spearman: https://github.com/allenai/allennlp/blob/master/allennlp/training/metrics/spearman_correlation.py 3. F1_measure: https://github.com/allenai/allennlp/blob/master/allennlp/training/metrics/f1_measure.py 4. Pearson_corelation: https://github.com/allenai/allennlp/blob/master/allennlp/training/metrics/pearson_correlation.py 5. AUC: https://github.com/allenai/allennlp/blob/master/allennlp/training/metrics/auc.py 6. Entropy: https://github.com/allenai/allennlp/blob/master/allennlp/training/metrics/entropy.py

https://api.github.com/repos/huggingface/datasets/issues/75/timeline

https://api.github.com/repos/huggingface/datasets/issues/74

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/74/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/74/comments

https://api.github.com/repos/huggingface/datasets/issues/74/events

https://github.com/huggingface/datasets/pull/74

MDExOlB1bGxSZXF1ZXN0NDE2NjA3MDcy

fix overflow check

closed

MEMBER

I did some tests and unfortunately the test ``` pa_array.nbytes > MAX_BATCH_BYTES ``` doesn't work. Indeed for a StructArray, `nbytes` can be less 2GB even if there is an overflow (it loops...). I don't think we can do a proper overflow test for the limit of 2GB... For now I replaced it with a sanity check on the first element.

https://api.github.com/repos/huggingface/datasets/issues/74/timeline

https://api.github.com/repos/huggingface/datasets/issues/73

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/73/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/73/comments

https://api.github.com/repos/huggingface/datasets/issues/73/events

https://github.com/huggingface/datasets/pull/73

MDExOlB1bGxSZXF1ZXN0NDE2NTMyMTg1

JSON script

closed

CONTRIBUTOR

Add a JSONS script to read JSON datasets from files.

https://api.github.com/repos/huggingface/datasets/issues/73/timeline

https://api.github.com/repos/huggingface/datasets/issues/72

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/72/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/72/comments

https://api.github.com/repos/huggingface/datasets/issues/72/events

https://github.com/huggingface/datasets/pull/72

MDExOlB1bGxSZXF1ZXN0NDE2Mzc4Mjg4

[README dummy data tests] README to better understand how the dummy data structure works

closed

CONTRIBUTOR

In this PR a README.md is added to tests to shine more light on how the dummy data structure works. I try to explain the different possible cases. IMO the best way to understand the logic is to checkout the dummy data structure of the different datasets I mention in the README.md since those are the "edge cases". @mariamabarham @thomwolf @lhoestq @jplu - I'd be happy to checkout the dummy data structure and get some feedback on possible improvements.

https://api.github.com/repos/huggingface/datasets/issues/72/timeline

https://api.github.com/repos/huggingface/datasets/issues/71

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/71/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/71/comments

https://api.github.com/repos/huggingface/datasets/issues/71/events

https://github.com/huggingface/datasets/pull/71

MDExOlB1bGxSZXF1ZXN0NDE2MTUxODM4

Fix arrow writer for big datasets using writer_batch_size

closed

MEMBER

This PR fixes Yacine's bug. According to [this](https://github.com/apache/arrow/blob/master/docs/source/cpp/arrays.rst#size-limitations-and-recommendations), it is not recommended to have pyarrow arrays bigger than 2Go. Therefore I set a default batch size of 100 000 examples per batch. In general it shouldn't exceed 2Go. If it does, I reduce the batch_size on the fly, and I notify the user with a warning.

https://api.github.com/repos/huggingface/datasets/issues/71/timeline

https://api.github.com/repos/huggingface/datasets/issues/70

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/70/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/70/comments

https://api.github.com/repos/huggingface/datasets/issues/70/events

https://github.com/huggingface/datasets/pull/70

MDExOlB1bGxSZXF1ZXN0NDE1OTM3NDgw

adding RACE, QASC, Super_glue and Tiny_shakespear datasets

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/70/timeline

https://api.github.com/repos/huggingface/datasets/issues/69

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/69/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/69/comments

https://api.github.com/repos/huggingface/datasets/issues/69/events

https://github.com/huggingface/datasets/pull/69

MDExOlB1bGxSZXF1ZXN0NDE1NzYyNTQ4

fix cache dir in builder tests

closed

MEMBER

minor fix

https://api.github.com/repos/huggingface/datasets/issues/69/timeline

https://api.github.com/repos/huggingface/datasets/issues/68

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/68/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/68/comments

https://api.github.com/repos/huggingface/datasets/issues/68/events

https://github.com/huggingface/datasets/pull/68

MDExOlB1bGxSZXF1ZXN0NDE1MzQ3NTgw

[CSV] re-add csv

closed

CONTRIBUTOR

Re-adding csv under the datasets under construction to keep circle ci happy - will have to see how to include it in the tests. @lhoestq noticed that I accidently deleted it in https://github.com/huggingface/nlp/pull/63#discussion_r422263729.

https://api.github.com/repos/huggingface/datasets/issues/68/timeline

https://api.github.com/repos/huggingface/datasets/issues/67

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/67/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/67/comments

https://api.github.com/repos/huggingface/datasets/issues/67/events

https://github.com/huggingface/datasets/pull/67

MDExOlB1bGxSZXF1ZXN0NDE1Mjc5NjI0

[Tests] Test files locally

closed

CONTRIBUTOR

This PR adds a `aws` and a `local` decorator to the tests so that tests now run on the local datasets. By default, the `aws` is deactivated and `local` is activated and `slow` is deactivated, so that only 1 test per dataset runs on circle ci. **When local is activated all folders in `./datasets` are tested.** **Important** When adding a dataset, we should no longer upload it to AWS. The steps are: 1. Open a PR 2. Add a dataset as described in `datasets/README.md` 3. If all tests pass, push to master Currently we have 49 functional datasets in our code base. We have 6 datasets "under-construction" that don't pass the tests - so I put them in a folder "datasets_under_construction" - it would be nice to open a PR to fix them and put them in the `datasets` folder. **Important** when running tests locally, the datasets are cached so to rerun them delete your local cache via: `rm -r ~/.cache/huggingface/datasets/*` @thomwolf @mariamabarham @lhoestq

https://api.github.com/repos/huggingface/datasets/issues/67/timeline

https://api.github.com/repos/huggingface/datasets/issues/66

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/66/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/66/comments

https://api.github.com/repos/huggingface/datasets/issues/66/events

https://github.com/huggingface/datasets/pull/66

MDExOlB1bGxSZXF1ZXN0NDE1MjM5Njgy

[Datasets] ReadME

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/66/timeline

https://api.github.com/repos/huggingface/datasets/issues/65

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/65/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/65/comments

https://api.github.com/repos/huggingface/datasets/issues/65/events

https://github.com/huggingface/datasets/pull/65

MDExOlB1bGxSZXF1ZXN0NDE1MjM4MDEw

fix math dataset and xcopa

closed

CONTRIBUTOR

- fixes math dataset and xcopa, uploaded both of the to S3

https://api.github.com/repos/huggingface/datasets/issues/65/timeline

https://api.github.com/repos/huggingface/datasets/issues/64

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/64/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/64/comments

https://api.github.com/repos/huggingface/datasets/issues/64/events

https://github.com/huggingface/datasets/pull/64

MDExOlB1bGxSZXF1ZXN0NDE1MjMwMjYy

[Datasets] Make master ready for datasets adding

closed

CONTRIBUTOR

Add all relevant files so that datasets can now be added on master

https://api.github.com/repos/huggingface/datasets/issues/64/timeline

https://api.github.com/repos/huggingface/datasets/issues/63

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/63/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/63/comments

https://api.github.com/repos/huggingface/datasets/issues/63/events

https://github.com/huggingface/datasets/pull/63

MDExOlB1bGxSZXF1ZXN0NDE1MTczODU5

[Dataset scripts] add all datasets scripts

closed

CONTRIBUTOR

As mentioned, we can have the canonical datasets in the master. For now I also want to include all the data as present on S3 to make the synchronization easier when uploading new datastes. @mariamabarham @lhoestq @thomwolf - what do you think? If this is ok for you, I can sync up the master with the `add_dataset` branch: https://github.com/huggingface/nlp/pull/37 so that master is up to date.

https://api.github.com/repos/huggingface/datasets/issues/63/timeline

https://api.github.com/repos/huggingface/datasets/issues/62

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/62/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/62/comments

https://api.github.com/repos/huggingface/datasets/issues/62/events

https://github.com/huggingface/datasets/pull/62

MDExOlB1bGxSZXF1ZXN0NDE1MTQ1NDAx

[Cached Path] Better error message

closed

CONTRIBUTOR

IMO returning `None` in this function only leads to confusion and is never helpful.

https://api.github.com/repos/huggingface/datasets/issues/62/timeline

https://api.github.com/repos/huggingface/datasets/issues/61

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/61/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/61/comments

https://api.github.com/repos/huggingface/datasets/issues/61/events

https://github.com/huggingface/datasets/pull/61

MDExOlB1bGxSZXF1ZXN0NDE1MTI3MTU4

[Load] rename setup_module to prepare_module

closed

CONTRIBUTOR

rename setup_module to prepare_module due to issues with pytests `setup_module` function. See: PR #59.

https://api.github.com/repos/huggingface/datasets/issues/61/timeline

https://api.github.com/repos/huggingface/datasets/issues/60

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/60/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/60/comments

https://api.github.com/repos/huggingface/datasets/issues/60/events

https://github.com/huggingface/datasets/pull/60

MDExOlB1bGxSZXF1ZXN0NDE0OTQyNjEy

Update to simplify some datasets conversion

closed

MEMBER

This PR updates the encoding of `Values` like `integers`, `boolean` and `float` to use python casting and avoid having to cast in the dataset scripts, as mentioned here: https://github.com/huggingface/nlp/pull/37#discussion_r420176626 We could also change (not included in this PR yet): - `supervized_keys` to make them a NamedTuple instead of a dataclass, and - handle specifically the `Translation` features. as mentioned here: https://github.com/huggingface/nlp/pull/37#discussion_r421740236 @patrickvonplaten @mariamabarham tell me if you want these two last changes as well.

https://api.github.com/repos/huggingface/datasets/issues/60/timeline

https://api.github.com/repos/huggingface/datasets/issues/59

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/59/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/59/comments

https://api.github.com/repos/huggingface/datasets/issues/59/events

https://github.com/huggingface/datasets/pull/59

MDExOlB1bGxSZXF1ZXN0NDE0OTM3NTgx

Fix tests

closed

MEMBER

@patrickvonplaten I've broken a bit the tests with #25 while simplifying and re-organizing the `load.py` and `download_manager.py` scripts. I'm trying to fix them here but I have a weird error, do you think you can have a look? ```bash (datasets) MacBook-Pro-de-Thomas:datasets thomwolf$ python -m pytest -sv ./tests/test_dataset_common.py::DatasetTest::test_builder_class_snli ============================================================================= test session starts ============================================================================= platform darwin -- Python 3.7.7, pytest-5.4.1, py-1.8.1, pluggy-0.13.1 -- /Users/thomwolf/miniconda2/envs/datasets/bin/python cachedir: .pytest_cache rootdir: /Users/thomwolf/Documents/GitHub/datasets plugins: xdist-1.31.0, forked-1.1.3 collected 1 item tests/test_dataset_common.py::DatasetTest::test_builder_class_snli ERROR =================================================================================== ERRORS ==================================================================================== ____________________________________________________________ ERROR at setup of DatasetTest.test_builder_class_snli ____________________________________________________________ file_path = <module 'tests.test_dataset_common' from '/Users/thomwolf/Documents/GitHub/datasets/tests/test_dataset_common.py'> download_config = DownloadConfig(cache_dir=None, force_download=False, resume_download=False, local_files_only=False, proxies=None, user_agent=None, extract_compressed_file=True, force_extract=True) download_kwargs = {} def setup_module(file_path: str, download_config: Optional[DownloadConfig] = None, **download_kwargs,) -> DatasetBuilder: r""" Download/extract/cache a dataset to add to the lib from a path or url which can be: - a path to a local directory containing the dataset processing python script - an url to a S3 directory with a dataset processing python script Dataset codes are cached inside the lib to allow easy import (avoid ugly sys.path tweaks) and using cloudpickle (among other things). Return: tuple of the unique id associated to the dataset the local path to the dataset """ if download_config is None: download_config = DownloadConfig(**download_kwargs) download_config.extract_compressed_file = True download_config.force_extract = True > name = list(filter(lambda x: x, file_path.split("/")))[-1] + ".py" E AttributeError: module 'tests.test_dataset_common' has no attribute 'split' src/nlp/load.py:169: AttributeError ============================================================================== warnings summary =============================================================================== /Users/thomwolf/miniconda2/envs/datasets/lib/python3.7/site-packages/tensorflow_core/python/pywrap_tensorflow_internal.py:15 /Users/thomwolf/miniconda2/envs/datasets/lib/python3.7/site-packages/tensorflow_core/python/pywrap_tensorflow_internal.py:15: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses import imp -- Docs: https://docs.pytest.org/en/latest/warnings.html =========================================================================== short test summary info =========================================================================== ERROR tests/test_dataset_common.py::DatasetTest::test_builder_class_snli - AttributeError: module 'tests.test_dataset_common' has no attribute 'split' ========================================================================= 1 warning, 1 error in 3.63s ========================================================================= ```

https://api.github.com/repos/huggingface/datasets/issues/59/timeline

https://api.github.com/repos/huggingface/datasets/issues/58

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/58/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/58/comments

https://api.github.com/repos/huggingface/datasets/issues/58/events

https://github.com/huggingface/datasets/pull/58

MDExOlB1bGxSZXF1ZXN0NDE0OTM0NTY4

Aborted PR - Fix tests

closed

MEMBER

@patrickvonplaten I've broken a bit the tests with #25 while simplifying and re-organizing the `load.py` and `download_manager.py` scripts. I'm trying to fix them here but I have a weird error, do you think you can have a look? ```bash (datasets) MacBook-Pro-de-Thomas:datasets thomwolf$ python -m pytest -sv ./tests/test_dataset_common.py::DatasetTest::test_builder_class_snli ============================================================================= test session starts ============================================================================= platform darwin -- Python 3.7.7, pytest-5.4.1, py-1.8.1, pluggy-0.13.1 -- /Users/thomwolf/miniconda2/envs/datasets/bin/python cachedir: .pytest_cache rootdir: /Users/thomwolf/Documents/GitHub/datasets plugins: xdist-1.31.0, forked-1.1.3 collected 1 item tests/test_dataset_common.py::DatasetTest::test_builder_class_snli ERROR =================================================================================== ERRORS ==================================================================================== ____________________________________________________________ ERROR at setup of DatasetTest.test_builder_class_snli ____________________________________________________________ file_path = <module 'tests.test_dataset_common' from '/Users/thomwolf/Documents/GitHub/datasets/tests/test_dataset_common.py'> download_config = DownloadConfig(cache_dir=None, force_download=False, resume_download=False, local_files_only=False, proxies=None, user_agent=None, extract_compressed_file=True, force_extract=True) download_kwargs = {} def setup_module(file_path: str, download_config: Optional[DownloadConfig] = None, **download_kwargs,) -> DatasetBuilder: r""" Download/extract/cache a dataset to add to the lib from a path or url which can be: - a path to a local directory containing the dataset processing python script - an url to a S3 directory with a dataset processing python script Dataset codes are cached inside the lib to allow easy import (avoid ugly sys.path tweaks) and using cloudpickle (among other things). Return: tuple of the unique id associated to the dataset the local path to the dataset """ if download_config is None: download_config = DownloadConfig(**download_kwargs) download_config.extract_compressed_file = True download_config.force_extract = True > name = list(filter(lambda x: x, file_path.split("/")))[-1] + ".py" E AttributeError: module 'tests.test_dataset_common' has no attribute 'split' src/nlp/load.py:169: AttributeError ============================================================================== warnings summary =============================================================================== /Users/thomwolf/miniconda2/envs/datasets/lib/python3.7/site-packages/tensorflow_core/python/pywrap_tensorflow_internal.py:15 /Users/thomwolf/miniconda2/envs/datasets/lib/python3.7/site-packages/tensorflow_core/python/pywrap_tensorflow_internal.py:15: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses import imp -- Docs: https://docs.pytest.org/en/latest/warnings.html =========================================================================== short test summary info =========================================================================== ERROR tests/test_dataset_common.py::DatasetTest::test_builder_class_snli - AttributeError: module 'tests.test_dataset_common' has no attribute 'split' ========================================================================= 1 warning, 1 error in 3.63s ========================================================================= ```

https://api.github.com/repos/huggingface/datasets/issues/58/timeline

https://api.github.com/repos/huggingface/datasets/issues/57

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/57/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/57/comments

https://api.github.com/repos/huggingface/datasets/issues/57/events

https://github.com/huggingface/datasets/pull/57

MDExOlB1bGxSZXF1ZXN0NDE0ODUzMDM5

Better cached path

closed

MEMBER

### Changes: - The `cached_path` no longer returns None if the file is missing/the url doesn't work. Instead, it can raise `FileNotFoundError` (missing file), `ConnectionError` (no cache and unreachable url) or `ValueError` (parsing error) - Fix requests to firebase API that doesn't handle HEAD requests... - Allow custom download in datasets script: it allows to use `tf.io.gfile.copy` for example, to download from google storage. I added an example: the `boolq` script

https://api.github.com/repos/huggingface/datasets/issues/57/timeline

https://api.github.com/repos/huggingface/datasets/issues/56

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/56/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/56/comments

https://api.github.com/repos/huggingface/datasets/issues/56/events

https://github.com/huggingface/datasets/pull/56

MDExOlB1bGxSZXF1ZXN0NDE0ODMyODY4

[Dataset] Tester add mock function

closed

CONTRIBUTOR

need to add an empty `extract()` function to make `hansard` dataset test work.

https://api.github.com/repos/huggingface/datasets/issues/56/timeline

https://api.github.com/repos/huggingface/datasets/issues/55

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/55/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/55/comments

https://api.github.com/repos/huggingface/datasets/issues/55/events

https://github.com/huggingface/datasets/pull/55

MDExOlB1bGxSZXF1ZXN0NDE0NjE0MjE1

Beam datasets

closed

MEMBER

# Beam datasets ## Intro Beam Datasets are using beam pipelines for preprocessing (basically lots of `.map` over objects called PCollections). The advantage of apache beam is that you can choose which type of runner you want to use to preprocess your data. The main runners are: - the `DirectRunner` to run the pipeline locally (default). However I encountered memory issues for big datasets (like the french or english wikipedia). Small dataset work fine - Google Dataflow. I didn't play with it. - Spark or Flink, two well known data processing frameworks. I tried to use the Spark/Flink local runners provided by apache beam for python and wasn't able to make them work properly though... ## From tfds beam datasets to our own beam datasets Tensorflow datasets used beam and a complicated pipeline to shard the TFRecords files. To allow users to download beam datasets and not having to preprocess them, they also allow to download the already preprocessed datasets from their google storage (the beam pipeline doesn't run in that case). On our side, we replace TFRecords by something else. Arrow or Parquet do the job but I chose Parquet as: 1) there is a builtin apache beam parquet writer that is quite convenient, and 2) reading parquet from the pyarrow library is also simple and effective (there is a mmap option !) Moreover we don't shard datasets in many many files like tfds (they were doing probably doing that mainly because of the limit of 2Gb per TFRecord file). Therefore we have a simpler pipeline that saves each split into one parquet file. We also removed the utilities to use their google storage (for now maybe ? we'll have to discuss it). ## Main changes - Added a BeamWriter to save the output of beam pipelines into parquet files and fill dataset infos - Create a ParquetReader and refactor a bit the arrow_reader.py \> **With this, we can now try to add beam datasets from tfds** I already added the wikipedia one, and I will also try to add the Wiki40b dataset ## Test the wikipedia script You can download and run the beam pipeline for wikipedia (using the `DirectRunner` by default) like this: ``` >>> import nlp >>> nlp.load("datasets/nlp/wikipedia", dataset_config="20200501.frr") ``` This wikipedia dataset (lang: frr, North Frisian) is a small one (~10Mb), but feel free to try bigger ones (and fill 20Gb of swap memory if you try the english one lol) ## Next Should we allow to download preprocessed datasets from the tfds google storage ? Should we try to optimize the beam pipelines to run locally without memory issues ? Should we try other data processing frameworks for big datasets, like spark ? ## About this PR It should be merged after #25 ----------------- I'd be happy to have your feedback and your ideas to improve the processing of big datasets like wikipedia :)

https://api.github.com/repos/huggingface/datasets/issues/55/timeline

https://api.github.com/repos/huggingface/datasets/issues/54

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/54/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/54/comments

https://api.github.com/repos/huggingface/datasets/issues/54/events

https://github.com/huggingface/datasets/pull/54

MDExOlB1bGxSZXF1ZXN0NDE0MjUyODkw

[Tests] Improved Error message for dummy folder structure

closed

CONTRIBUTOR

Improved Error message

https://api.github.com/repos/huggingface/datasets/issues/54/timeline

https://api.github.com/repos/huggingface/datasets/issues/53

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/53/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/53/comments

https://api.github.com/repos/huggingface/datasets/issues/53/events

https://github.com/huggingface/datasets/pull/53

MDExOlB1bGxSZXF1ZXN0NDE0MTkwMzkz

[Features] Typo in generate_from_dict

closed

CONTRIBUTOR

Change `isinstance` test in features when generating features from dict.

https://api.github.com/repos/huggingface/datasets/issues/53/timeline

https://api.github.com/repos/huggingface/datasets/issues/52

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/52/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/52/comments

https://api.github.com/repos/huggingface/datasets/issues/52/events

https://github.com/huggingface/datasets/pull/52

MDExOlB1bGxSZXF1ZXN0NDE0MTEyMDAy

allow dummy folder structure to handle dict of lists

closed

CONTRIBUTOR

`esnli.py` needs that extension of the dummy data testing.

https://api.github.com/repos/huggingface/datasets/issues/52/timeline

https://api.github.com/repos/huggingface/datasets/issues/51

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/51/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/51/comments

https://api.github.com/repos/huggingface/datasets/issues/51/events

https://github.com/huggingface/datasets/pull/51

MDExOlB1bGxSZXF1ZXN0NDE0MDUyOTYw

[Testing] Improved testing structure

closed

CONTRIBUTOR

This PR refactors the test design a bit and puts the mock download manager in the `utils` files as it is just a test helper class. as @mariamabarham pointed out, creating a dummy folder structure can be quite hard to grasp. This PR tries to change that to some extent. It follows the following logic for the `dummy` folder structure now: 1.) The data bulider has no config -> the `dummy` folder structure is: `dummy/<version>/dummy_data.zip` 2) The data builder has >= 1 configs -> the `dummy` folder structure is: `dummy/<config_name_1>/<version>/dummy_data.zip` `dummy/<config_name_2>/<version>/dummy_data.zip` Now, the difficult part is how to create the `dummy_data.zip` file. There are two cases: A) The `data_urs` parameter inserted into the `download_and_extract` fn is a **string**: -> the `dummy_data.zip` file zips the folder: `dummy_data/<relative_path_of_folder_structure_of_url>` B) The `data_urs` parameter inserted into the `download_and_extract` fn is a **dict**: -> the `dummy_data.zip` file zips the folder: `dummy_data/<relative_path_of_folder_structure_of_url_behind _key_1>` `dummy_data/<relative_path_of_folder_structure_of_url_behind _key_2>` By relative folder structure I mean `url_path.split('./')[-1]`. As an example the dataset **xquad** by deepmind has the following url path behind the key `de`: `https://github.com/deepmind/xquad/blob/master/xquad.de.json` -> This means that the relative url path should be `xquad.de.json`. @mariamabarham B) is a change from how is was before and I think is makes more sense. While before the `dummy_data.zip` file for xquad with config `de` looked like: `dummy_data/de` it would now look like `dummy_data/xquad.de.json`. I think this is better and easier to understand. Therefore there are currently 6 tests that would have to have changed their dummy folder structure, but which can easily be done (30min). I also added a function: `print_dummy_data_folder_structure` that prints out the expected structures when testing which should be quite helpful.

https://api.github.com/repos/huggingface/datasets/issues/51/timeline

https://api.github.com/repos/huggingface/datasets/issues/50

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/50/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/50/comments

https://api.github.com/repos/huggingface/datasets/issues/50/events

https://github.com/huggingface/datasets/pull/50

MDExOlB1bGxSZXF1ZXN0NDEzNTAwMjE0

[Tests] test only for fast test as a default

closed

CONTRIBUTOR

Test only for one config on circle ci to speed up testing. Add all config test as a slow test. @mariamabarham @thomwolf

https://api.github.com/repos/huggingface/datasets/issues/50/timeline

https://api.github.com/repos/huggingface/datasets/issues/49

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/49/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/49/comments

https://api.github.com/repos/huggingface/datasets/issues/49/events

https://github.com/huggingface/datasets/pull/49

MDExOlB1bGxSZXF1ZXN0NDEzNDY5ODg0

fix flatten nested

closed

MEMBER

https://api.github.com/repos/huggingface/datasets/issues/49/timeline

https://api.github.com/repos/huggingface/datasets/issues/48

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/48/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/48/comments

https://api.github.com/repos/huggingface/datasets/issues/48/events

https://github.com/huggingface/datasets/pull/48

MDExOlB1bGxSZXF1ZXN0NDEzNDM2MTgz

[Command Convert] remove tensorflow import

closed

CONTRIBUTOR

Remove all tensorflow import statements.

https://api.github.com/repos/huggingface/datasets/issues/48/timeline

https://api.github.com/repos/huggingface/datasets/issues/47

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/47/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/47/comments

https://api.github.com/repos/huggingface/datasets/issues/47/events

https://github.com/huggingface/datasets/pull/47

MDExOlB1bGxSZXF1ZXN0NDEzMzg5MDc1

[PyArrow Feature] fix py arrow bool

closed

CONTRIBUTOR

To me it seems that `bool` can only be accessed with `bool_` when looking at the pyarrow types: https://arrow.apache.org/docs/python/api/datatypes.html.

https://api.github.com/repos/huggingface/datasets/issues/47/timeline

https://api.github.com/repos/huggingface/datasets/issues/46

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/46/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/46/comments

https://api.github.com/repos/huggingface/datasets/issues/46/events

https://github.com/huggingface/datasets/pull/46

MDExOlB1bGxSZXF1ZXN0NDEzMzUxNTY0

[Features] Strip str key before dict look-up

closed

CONTRIBUTOR

The dataset `anli.py` currently fails because it tries to look up a key `1\n` in a dict that only has the key `1`. Added an if statement to strip key if it cannot be found in dict.

https://api.github.com/repos/huggingface/datasets/issues/46/timeline

https://api.github.com/repos/huggingface/datasets/issues/45

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/45/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/45/comments

https://api.github.com/repos/huggingface/datasets/issues/45/events

https://github.com/huggingface/datasets/pull/45

MDExOlB1bGxSZXF1ZXN0NDEzMzQzMjAy

[Load] Separate Module kwargs and builder kwargs.

closed

CONTRIBUTOR

Kwargs for the `load_module` fn should be passed with `module_xxxx` to `builder_kwargs` of `load` fn. This is a follow-up PR of: https://github.com/huggingface/nlp/pull/41

https://api.github.com/repos/huggingface/datasets/issues/45/timeline

https://api.github.com/repos/huggingface/datasets/issues/44

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/44/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/44/comments

https://api.github.com/repos/huggingface/datasets/issues/44/events

https://github.com/huggingface/datasets/pull/44

MDExOlB1bGxSZXF1ZXN0NDEyOTUwMzU1

[Tests] Fix tests for datasets with no config

closed

CONTRIBUTOR

Forgot to fix `None` problem for datasets that have no config this in PR: https://github.com/huggingface/nlp/pull/42

https://api.github.com/repos/huggingface/datasets/issues/44/timeline

https://api.github.com/repos/huggingface/datasets/issues/43

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/43/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/43/comments

https://api.github.com/repos/huggingface/datasets/issues/43/events

https://github.com/huggingface/datasets/pull/43

MDExOlB1bGxSZXF1ZXN0NDEyODcxNTE5

[Checksums] If no configs exist prevent to run over empty list

closed

CONTRIBUTOR

`movie_rationales` e.g. has no configs.

https://api.github.com/repos/huggingface/datasets/issues/43/timeline

https://api.github.com/repos/huggingface/datasets/issues/42

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/42/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/42/comments

https://api.github.com/repos/huggingface/datasets/issues/42/events

https://github.com/huggingface/datasets/pull/42

MDExOlB1bGxSZXF1ZXN0NDEyODU1OTE2

[Tests] allow tests for builders without config

closed

CONTRIBUTOR

Some dataset scripts have no configs - the tests have to be adapted for this case. In this case the dummy data will be saved as: - natural_questions -> dummy -> -> 1.0.0 (version num) -> -> -> dummy_data.zip

https://api.github.com/repos/huggingface/datasets/issues/42/timeline

https://api.github.com/repos/huggingface/datasets/issues/41

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/41/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/41/comments

https://api.github.com/repos/huggingface/datasets/issues/41/events

https://github.com/huggingface/datasets/pull/41

MDExOlB1bGxSZXF1ZXN0NDEyODQzNDQy

[Load module] allow kwargs into load module

closed

CONTRIBUTOR

Currenly it is not possible to force a re-download of the dataset script. This simple change allows to pass ``force_reload=True`` as ``builder_kwargs`` in the ``load.py`` function.

https://api.github.com/repos/huggingface/datasets/issues/41/timeline

https://api.github.com/repos/huggingface/datasets/issues/40

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/40/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/40/comments

https://api.github.com/repos/huggingface/datasets/issues/40/events

https://github.com/huggingface/datasets/pull/40

MDExOlB1bGxSZXF1ZXN0NDEyODI4NzU2

Update remote checksums instead of overwrite

closed

MEMBER

When the user uploads a dataset on S3, checksums are also uploaded with the `--upload_checksums` parameter. If the user uploads the dataset in several steps, then the remote checksums file was previously overwritten. Now it's going to be updated with the new checksums.

https://api.github.com/repos/huggingface/datasets/issues/40/timeline

https://api.github.com/repos/huggingface/datasets/issues/39

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/39/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/39/comments

https://api.github.com/repos/huggingface/datasets/issues/39/events

https://github.com/huggingface/datasets/pull/39

MDExOlB1bGxSZXF1ZXN0NDEyODIxNTA4

[Test] improve slow testing

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/39/timeline

https://api.github.com/repos/huggingface/datasets/issues/38

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/38/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/38/comments

https://api.github.com/repos/huggingface/datasets/issues/38/events

https://github.com/huggingface/datasets/issues/38

MDU6SXNzdWU2MTE2Nzc2NTY=

[Checksums] Error for some datasets

closed

CONTRIBUTOR

The checksums command works very nicely for `squad`. But for `crime_and_punish` and `xnli`, the same bug happens: When running: ``` python nlp-cli nlp-cli test xnli --save_checksums ``` leads to: ``` File "nlp-cli", line 33, in <module> service.run() File "/home/patrick/python_bin/nlp/commands/test.py", line 61, in run ignore_checksums=self._ignore_checksums, File "/home/patrick/python_bin/nlp/builder.py", line 383, in download_and_prepare self._download_and_prepare(dl_manager=dl_manager, download_config=download_config) File "/home/patrick/python_bin/nlp/builder.py", line 627, in _download_and_prepare dl_manager=dl_manager, max_examples_per_split=download_config.max_examples_per_split, File "/home/patrick/python_bin/nlp/builder.py", line 431, in _download_and_prepare split_generators = self._split_generators(dl_manager, **split_generators_kwargs) File "/home/patrick/python_bin/nlp/datasets/xnli/8bf4185a2da1ef2a523186dd660d9adcf0946189e7fa5942ea31c63c07b68a7f/xnli.py", line 95, in _split_generators dl_dir = dl_manager.download_and_extract(_DATA_URL) File "/home/patrick/python_bin/nlp/utils/download_manager.py", line 246, in download_and_extract return self.extract(self.download(url_or_urls)) File "/home/patrick/python_bin/nlp/utils/download_manager.py", line 186, in download self._record_sizes_checksums(url_or_urls, downloaded_path_or_paths) File "/home/patrick/python_bin/nlp/utils/download_manager.py", line 166, in _record_sizes_checksums self._recorded_sizes_checksums[url] = get_size_checksum(path) File "/home/patrick/python_bin/nlp/utils/checksums_utils.py", line 81, in get_size_checksum with open(path, "rb") as f: TypeError: expected str, bytes or os.PathLike object, not tuple ```

https://api.github.com/repos/huggingface/datasets/issues/38/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/37

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/37/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/37/comments

https://api.github.com/repos/huggingface/datasets/issues/37/events

https://github.com/huggingface/datasets/pull/37

MDExOlB1bGxSZXF1ZXN0NDEyNzg5MjQ4

[Datasets ToDo-List] add datasets

closed

CONTRIBUTOR

## Description This PR acts as a dashboard to see which datasets are added to the library and work. Cicle-ci should always be green so that we can be sure that newly added datasets are functional. This PR should not be merged. ## Progress **For the following datasets the test commands**: ``` RUN_SLOW=1 pytest tests/test_dataset_common.py::DatasetTest::test_load_real_dataset_<your-dataset-name> ``` and ``` RUN_SLOW=1 pytest tests/test_dataset_common.py::DatasetTest::test_load_dataset_all_configs_<your-dataset-name> ``` **passes**. - [x] Squad - [x] Sentiment140 - [x] XNLI - [x] Crime_and_Punish - [x] movie_rationales - [x] ai2_arc - [x] anli - [x] event2Mind - [x] Fquad - [x] blimp - [x] empathetic_dialogues - [x] cosmos_qa - [x] xquad - [x] blog_authorship_corpus - [x] SNLI - [x] break_data - [x] SQuAD v2 - [x] cfq - [x] eraser_multi_rc - [x] Glue - [x] Tydiqa - [x] wiki_qa - [x] wikitext - [x] winogrande - [x] wiqa - [x] esnli - [x] civil_comments - [x] commonsense_qa - [x] com_qa - [x] coqa - [x] wiki_split - [x] cos_e - [x] xcopa - [x] quarel - [x] quartz - [x] squad_it - [x] quoref - [x] squad_pt - [x] cornell_movie_dialog - [x] SciQ - [x] Scifact - [x] hellaswag - [x] ted_multi (in translate) - [x] Aeslc (summarization) - [x] drop - [x] gap - [x] hansard - [x] opinosis - [x] MLQA - [x] math_dataset ## How-To-Add a dataset **Before adding a dataset make sure that your branch is up to date**: 1. `git checkout add_datasets` 2. `git pull` **Add a dataset via the `convert_dataset.sh` bash script:** Running `bash convert_dataset.sh <file/to/tfds/datascript.py>` (*e.g.* `bash convert_dataset.sh ../tensorflow-datasets/tensorflow_datasets/text/movie_rationales.py`) will automatically run all the steps mentioned in **Add a dataset manually** below. Make sure that you run `convert_dataset.sh` from the root folder of `nlp`. The conversion script should work almost always for step 1): "convert dataset script from tfds to nlp format" and 2) "create checksum file" and step 3) "make style". It can also sometimes automatically run step 4) "create the correct dummy data from tfds", but this will only work if a) there is either no config name or only one config name and b) the `tfds testing/test_data/fake_example` is in the correct form. Nevertheless, the script should always be run in the beginning until an error occurs to be more efficient. If the conversion script does not work or fails at some step, then you can run the steps manually as follows: **Add a dataset manually** Make sure you run all of the following commands from the root of your `nlp` git clone. Also make sure that you changed to this branch: ``` git checkout add_datasets ``` 1) the tfds datascript file should be converted to `nlp` style: ``` python nlp-cli convert --tfds_path <path/to/tensorflow_datasets/text/your_dataset_name>.py --nlp_directory datasets/nlp ``` This will convert the tdfs script and create a folder with the correct name. 2) the checksum file should be added. Use the command: ``` python nlp-cli test datasets/nlp/<your-dataset-folder> --save_checksums --all_configs ``` A checksums.txt file should be created in your folder and the structure should look as follows: squad/ ├── squad.py/ └── urls_checksums/ ...........└── checksums.txt Delete the created `*.lock` file afterward - it should not be uploaded to AWS. 3) run black and isort on your newly added datascript files so that they look nice: ``` make style ``` 4) the dummy data should be added. For this it might be useful to take a look into the structure of other examples as shown in the PR here and at `<path/to/tensorflow_datasets/testing/test_data/test_data/fake_examples>` whether the same data can be used. 5) the data can be uploaded to AWS using the command ``` aws s3 cp datasets/nlp/<your-dataset-folder> s3://datasets.huggingface.co/nlp/<your-dataset-folder> --recursive ``` 6) check whether all works as expected using: ``` RUN_SLOW=1 pytest tests/test_dataset_common.py::DatasetTest::test_load_real_dataset_<your-dataset-name> ``` and ``` RUN_SLOW=1 pytest tests/test_dataset_common.py::DatasetTest::test_load_dataset_all_configs_<your-dataset-name> ``` 7) push to this PR and rerun the circle ci workflow to check whether circle ci stays green. 8) Edit this commend and tick off your newly added dataset :-) ## TODO-list Maybe we can add a TODO-list here for everybody that feels like adding new datasets so that we will not add the same datasets. Here a link to available datasets: https://docs.google.com/spreadsheets/d/1zOtEqOrnVQwdgkC4nJrTY6d-Av02u0XFzeKAtBM2fUI/edit#gid=0 Patrick: - [ ] boolq - *weird download link* - [ ] c4 - *beam dataset*

https://api.github.com/repos/huggingface/datasets/issues/37/timeline

https://api.github.com/repos/huggingface/datasets/issues/36

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/36/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/36/comments

https://api.github.com/repos/huggingface/datasets/issues/36/events

https://github.com/huggingface/datasets/pull/36

MDExOlB1bGxSZXF1ZXN0NDEyNjgwOTk1

Metrics - refactoring, adding support for download and distributed metrics

closed

MEMBER

Refactoring metrics to have a similar loading API than the datasets and improving the import system. # Import system The import system has ben upgraded. There are now three types of imports allowed: 1. `library` imports (identified as "absolute imports") ```python import seqeval ``` => we'll test all the imports before running the scripts and if one cannot be imported we'll display an error message like this one: `ImportError: To be able to use this metric/dataset, you need to install the following dependencies ['seqeval'] using 'pip install seqeval' for instance'` 2. `internal` imports (identified as "relative imports") ```python import .c4_utils ``` => we'll assume this point to a file in the same directory/S3-directory as the main script and download this file. 2. `external` imports (identified as "relative imports" with a comment starting with `# From:`) ```python from .nmt_bleu import compute_bleu # From: https://github.com/tensorflow/nmt/blob/master/nmt/scripts/bleu.py ``` => we'll assume this point to the URL of a python script (if it's a link to a github file, we'll take the raw file automatically). => the script is downloaded and renamed to the import name (here above renamed from `bleu.py` to `nmt_bleu.py`). Renaming the file can be necessary if the distant file has the same name as the dataset/metric processing script. If you forgot to rename the distant script and it has the same name as the dataset/metric, you'll have an explicit error message asking to rename the import anyway. # Hosting metrics Metrics are hosted on a S3 bucket like the dataset processing scripts. # Metrics scripts Metrics scripts have a lot in common with datasets processing scripts. They also have a `metric.info` including citations, descriptions and links to relevant pages. Metrics have more documentation to supply to ensure they are used well. Four examples are already included for reference in [./metrics](./metrics): BLEU, ROUGE, SacreBLEU and SeqEVAL. # Automatic support for distributed/multi-processing metric computation We've also added support for automatic distributed/multi-processing metric computation (e.g. when using DistributedDataParallel). We leverage our own dataset format for smart caching in this case. Here is a quick gist of a standard use of metrics (the simplest usage): ```python import nlp bleu_metric = nlp.load_metric('bleu') # If you only have a single iteration, you can easily compute the score like this predictions = model(inputs) score = bleu_metric.compute(predictions, references) # If you have a loop, you can "add" your predictions and references at each iteration instead of having to save them yourself (the metric object store them efficiently for you) for batch in dataloader: model_input, targets = batch predictions = model(model_inputs) bleu.add(predictions, targets) score = bleu_metric.compute() # Compute the score from all the stored predictions/references ``` Here is a quick gist of a use in a distributed torch setup (should work for any python multi-process setup actually). It's pretty much identical to the second example above: ```python import nlp # You need to give the total number of parallel python processes (num_process) and the id of each process (process_id) bleu = nlp.load_metric('bleu', process_id=torch.distributed.get_rank(),b num_process=torch.distributed.get_world_size()) for batch in dataloader: model_input, targets = batch predictions = model(model_inputs) bleu.add(predictions, targets) score = bleu_metric.compute() # Compute the score on the first node by default (can be set to compute on each node as well) ```

https://api.github.com/repos/huggingface/datasets/issues/36/timeline

https://api.github.com/repos/huggingface/datasets/issues/35

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/35/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/35/comments

https://api.github.com/repos/huggingface/datasets/issues/35/events

https://github.com/huggingface/datasets/pull/35

MDExOlB1bGxSZXF1ZXN0NDEyNjAyMTc0

[Tests] fix typo

closed

CONTRIBUTOR

@lhoestq - currently the slow test fail with: ``` _____________________________________________________________________________________ DatasetTest.test_load_real_dataset_xnli _____________________________________________________________________________________ self = <tests.test_dataset_common.DatasetTest testMethod=test_load_real_dataset_xnli>, dataset_name = 'xnli' @slow def test_load_real_dataset(self, dataset_name): with tempfile.TemporaryDirectory() as temp_data_dir: > dataset = load(dataset_name, data_dir=temp_data_dir) tests/test_dataset_common.py:153: _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ ../../python_bin/nlp/load.py:497: in load dbuilder.download_and_prepare(**download_and_prepare_kwargs) ../../python_bin/nlp/builder.py:383: in download_and_prepare self._download_and_prepare(dl_manager=dl_manager, download_config=download_config) ../../python_bin/nlp/builder.py:627: in _download_and_prepare dl_manager=dl_manager, max_examples_per_split=download_config.max_examples_per_split, ../../python_bin/nlp/builder.py:431: in _download_and_prepare split_generators = self._split_generators(dl_manager, **split_generators_kwargs) ../../python_bin/nlp/datasets/xnli/8bf4185a2da1ef2a523186dd660d9adcf0946189e7fa5942ea31c63c07b68a7f/xnli.py:95: in _split_generators dl_dir = dl_manager.download_and_extract(_DATA_URL) ../../python_bin/nlp/utils/download_manager.py:246: in download_and_extract return self.extract(self.download(url_or_urls)) ../../python_bin/nlp/utils/download_manager.py:186: in download self._record_sizes_checksums(url_or_urls, downloaded_path_or_paths) ../../python_bin/nlp/utils/download_manager.py:166: in _record_sizes_checksums self._recorded_sizes_checksums[url] = get_size_checksum(path) _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ path = ('', '/tmp/tmpkajlg9yc/downloads/c0f7773c480a3f2d85639d777e0e17e65527460310d80760fd3fc2b2f2960556.c952a63cb17d3d46e412ceb7dbcd656ce2b15cc9ef17f50c28f81c48a7c853b5') def get_size_checksum(path: str) -> Tuple[int, str]: """Compute the file size and the sha256 checksum of a file""" m = sha256() > with open(path, "rb") as f: E TypeError: expected str, bytes or os.PathLike object, not tuple ../../python_bin/nlp/utils/checksums_utils.py:81: TypeError ``` - the checksums probably need to be updated no? And we should also think about how to write a test for the checksums.

https://api.github.com/repos/huggingface/datasets/issues/35/timeline

https://api.github.com/repos/huggingface/datasets/issues/34

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/34/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/34/comments

https://api.github.com/repos/huggingface/datasets/issues/34/events

https://github.com/huggingface/datasets/pull/34

MDExOlB1bGxSZXF1ZXN0NDEyNTg0OTM0

[Tests] add slow tests

closed

CONTRIBUTOR

This PR adds a slow test that downloads the "real" dataset. The test is decorated as "slow" so that it will not automatically run on circle ci. Before uploading a dataset, one should test that this test passes, manually by running ``` RUN_SLOW=1 pytest tests/test_dataset_common.py::DatasetTest::test_load_real_dataset_<your-dataset-script-name> ``` This PR should be merged after PR: #33

https://api.github.com/repos/huggingface/datasets/issues/34/timeline

https://api.github.com/repos/huggingface/datasets/issues/33

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/33/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/33/comments

https://api.github.com/repos/huggingface/datasets/issues/33/events

https://github.com/huggingface/datasets/pull/33

MDExOlB1bGxSZXF1ZXN0NDEyMzU1ODE0

Big cleanup/refactoring for clean serialization

closed

MEMBER

This PR cleans many base classes to re-build them as `dataclasses`. We can thus use a simple serialization workflow for `DatasetInfo`, including it's `Features` and `SplitDict` based on `dataclasses` `asdict()`. The resulting code is a lot shorter, can be easily serialized/deserialized, dataset info are human-readable and we can get rid of the `dataclass_json` dependency. The scripts have breaking changes and the conversion tool is updated. Example of dataset info in SQuAD script now: ```python def _info(self): return nlp.DatasetInfo( description=_DESCRIPTION, features=nlp.Features({ "id": nlp.Value('string'), "title": nlp.Value('string'), "context": nlp.Value('string'), "question": nlp.Value('string'), "answers": nlp.Sequence({ "text": nlp.Value('string'), "answer_start": nlp.Value('int32'), }), }), # No default supervised_keys (as we have to pass both question # and context as input). supervised_keys=None, homepage="https://rajpurkar.github.io/SQuAD-explorer/", citation=_CITATION, ) ``` Example of serialized dataset info: ```bash { "description": "Stanford Question Answering Dataset (SQuAD) is a reading comprehension dataset, consisting of questions posed by crowdworkers on a set of Wikipedia articles, where the answer to every question is a segment of text, or span, from the corresponding reading passage, or the question might be unanswerable.\n", "citation": "@article{2016arXiv160605250R,\n author = {{Rajpurkar}, Pranav and {Zhang}, Jian and {Lopyrev},\n Konstantin and {Liang}, Percy},\n title = \"{SQuAD: 100,000+ Questions for Machine Comprehension of Text}\",\n journal = {arXiv e-prints},\n year = 2016,\n eid = {arXiv:1606.05250},\n pages = {arXiv:1606.05250},\narchivePrefix = {arXiv},\n eprint = {1606.05250},\n}\n", "homepage": "https://rajpurkar.github.io/SQuAD-explorer/", "license": "", "features": { "id": { "dtype": "string", "_type": "Value" }, "title": { "dtype": "string", "_type": "Value" }, "context": { "dtype": "string", "_type": "Value" }, "question": { "dtype": "string", "_type": "Value" }, "answers": { "feature": { "text": { "dtype": "string", "_type": "Value" }, "answer_start": { "dtype": "int32", "_type": "Value" } }, "length": -1, "_type": "Sequence" } }, "supervised_keys": null, "name": "squad", "version": { "version_str": "1.0.0", "description": "New split API (https://tensorflow.org/datasets/splits)", "nlp_version_to_prepare": null, "major": 1, "minor": 0, "patch": 0 }, "splits": { "train": { "name": "train", "num_bytes": 79426386, "num_examples": 87599, "dataset_name": "squad" }, "validation": { "name": "validation", "num_bytes": 10491883, "num_examples": 10570, "dataset_name": "squad" } }, "size_in_bytes": 0, "download_size": 35142551, "download_checksums": [] } ```

https://api.github.com/repos/huggingface/datasets/issues/33/timeline

https://api.github.com/repos/huggingface/datasets/issues/32

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/32/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/32/comments

https://api.github.com/repos/huggingface/datasets/issues/32/events

https://github.com/huggingface/datasets/pull/32

MDExOlB1bGxSZXF1ZXN0NDEyMTAzMzIx

Fix map caching notebooks

closed

MEMBER

Previously, caching results with `.map()` didn't work in notebooks. To reuse a result, `.map()` serializes the functions with `dill.dumps` and then it hashes it. The problem is that when using `dill.dumps` to serialize a function, it also saves its origin (filename + line no.) and the origin of all the `globals` this function needs. However for notebooks and shells, the filename looks like \<ipython-input-13-9ed2afe61d25\> and the line no. changes often. To fix the problem, I added a new dispatch function for code objects that ignore the origin of the code if it comes from a notebook or a python shell. I tested these cases in a notebook: - lambda functions - named functions - methods - classmethods - staticmethods - classes that implement `__call__` The caching now works as expected for all of them :) I also tested the caching in the demo notebook and it works fine !

https://api.github.com/repos/huggingface/datasets/issues/32/timeline

https://api.github.com/repos/huggingface/datasets/issues/31

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/31/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/31/comments

https://api.github.com/repos/huggingface/datasets/issues/31/events

https://github.com/huggingface/datasets/pull/31

MDExOlB1bGxSZXF1ZXN0NDEyMDczNDE4

[Circle ci] Install a virtual env before running tests

closed

CONTRIBUTOR

Install a virtual env before running tests to not running into sudo issues when dynamically downloading files. Same number of tests now pass / fail as on my local computer: 

https://api.github.com/repos/huggingface/datasets/issues/31/timeline

https://api.github.com/repos/huggingface/datasets/issues/30

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/30/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/30/comments

https://api.github.com/repos/huggingface/datasets/issues/30/events

https://github.com/huggingface/datasets/pull/30

MDExOlB1bGxSZXF1ZXN0NDExOTY4Mzk3

add metrics which require download files from github

closed

CONTRIBUTOR

To download files from github, I copied the `load_dataset_module` and its dependencies (without the builder) in `load.py` to `metrics/metric_utils.py`. I made the following changes: - copy the needed files in a folder`metric_name` - delete all other files that are not needed For metrics that require an external import, I first create a `<metric_name>_imports.py` file which contains all external urls. Then I create a `<metric_name>.py` in which I will load the external files using `<metric_name>_imports.py`

https://api.github.com/repos/huggingface/datasets/issues/30/timeline

https://api.github.com/repos/huggingface/datasets/issues/29

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/29/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/29/comments

https://api.github.com/repos/huggingface/datasets/issues/29/events

https://github.com/huggingface/datasets/pull/29

MDExOlB1bGxSZXF1ZXN0NDExNzIwODMx

Hf_api small changes

closed

MEMBER

From Patrick: ```python from nlp import hf_api api = hf_api.HfApi() api.dataset_list() ``` works :-)

https://api.github.com/repos/huggingface/datasets/issues/29/timeline

https://api.github.com/repos/huggingface/datasets/issues/28

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/28/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/28/comments

https://api.github.com/repos/huggingface/datasets/issues/28/events

https://github.com/huggingface/datasets/pull/28

MDExOlB1bGxSZXF1ZXN0NDExNzE5MTQy

[Circle ci] Adds circle ci config

closed

CONTRIBUTOR

@thomwolf can you take a look and set up circle ci on: https://app.circleci.com/projects/project-dashboard/github/huggingface I think for `nlp` only admins can set it up, which I guess is you :-)

https://api.github.com/repos/huggingface/datasets/issues/28/timeline

https://api.github.com/repos/huggingface/datasets/issues/27

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/27/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/27/comments

https://api.github.com/repos/huggingface/datasets/issues/27/events

https://github.com/huggingface/datasets/pull/27

MDExOlB1bGxSZXF1ZXN0NDExNzA5OTc0

[Cleanup] Removes all files in testing except test_dataset_common

closed

CONTRIBUTOR

As far as I know, all files in `tests` were old `tfds test files` so I removed them. We can still look them up on the other library.

https://api.github.com/repos/huggingface/datasets/issues/27/timeline

https://api.github.com/repos/huggingface/datasets/issues/26

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/26/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/26/comments

https://api.github.com/repos/huggingface/datasets/issues/26/events

https://github.com/huggingface/datasets/pull/26

MDExOlB1bGxSZXF1ZXN0NDExNzA2NjA2

[Tests] Clean tests

closed

CONTRIBUTOR

the abseil testing library (https://abseil.io/docs/python/quickstart.html) is better than the one I had before, so I decided to switch to that and changed the `setup.py` config file. Abseil has more support and a cleaner API for parametrized testing I think. I added a list of all dataset scripts that are currently on AWS, but will replace that once the API is integrated into this lib. One can now easily test for just a single function for a single dataset with: `tests/test_dataset_common.py::DatasetTest::test_load_dataset_wikipedia` NOTE: This PR is rebased on PR #29 so should be merged after.

https://api.github.com/repos/huggingface/datasets/issues/26/timeline

https://api.github.com/repos/huggingface/datasets/issues/25

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/25/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/25/comments

https://api.github.com/repos/huggingface/datasets/issues/25/events

https://github.com/huggingface/datasets/pull/25

MDExOlB1bGxSZXF1ZXN0NDExMjQ4Nzg2

Add script csv datasets

closed

CONTRIBUTOR

This is a PR allowing to create datasets from local CSV files. A usage might be: ```python import nlp ds = nlp.load( path="csv", name="bbc", dataset_files={ nlp.Split.TRAIN: ["datasets/dummy_data/csv/train.csv"], nlp.Split.TEST: [""datasets/dummy_data/csv/test.csv""] }, csv_kwargs={ "skip_rows": 0, "delimiter": ",", "quote_char": "\"", "header_as_column_names": True } ) ``` ``` Downloading and preparing dataset bbc/1.0.0 (download: Unknown size, generated: Unknown size, total: Unknown size) to /home/jplu/.cache/huggingface/datasets/bbc/1.0.0... Dataset bbc downloaded and prepared to /home/jplu/.cache/huggingface/datasets/bbc/1.0.0. Subsequent calls will reuse this data. {'test': Dataset(schema: {'category': 'string', 'text': 'string'}, num_rows: 49), 'train': Dataset(schema: {'category': 'string', 'text': 'string'}, num_rows: 99), 'validation': Dataset(schema: {'category': 'string', 'text': 'string'}, num_rows: 0)} ``` How it is read: - `path`: the `csv` word means "I want to create a CSV dataset" - `name`: the name of this dataset is `bbc` - `dataset_files`: this is a dictionary where each key is the list of files corresponding to the key split. - `csv_kwargs`: this is the keywords arguments to "explain" how to read the CSV files * `skip_rows`: number of rows have to be skipped, starting from the beginning of the file * `delimiter`: which delimiter is used to separate the columns * `quote_char`: which quote char is used to represent a column where the delimiter appears in one of them * `header_as_column_names`: will use the first row (header) of the file as name for the features. Otherwise the names will be automatically generated as `f1`, `f2`, etc... Will be applied after the `skip_rows` parameter. **TODO**: for now the `csv.py` is copied each time we create a new dataset as `ds_name.py`, this behavior will be modified to have only the `csv.py` script copied only once and not for all the CSV datasets.

https://api.github.com/repos/huggingface/datasets/issues/25/timeline

https://api.github.com/repos/huggingface/datasets/issues/24

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/24/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/24/comments

https://api.github.com/repos/huggingface/datasets/issues/24/events

https://github.com/huggingface/datasets/pull/24

MDExOlB1bGxSZXF1ZXN0NDEwNzE5MTU0

Add checksums

closed

MEMBER

### Checksums files They are stored next to the dataset script in urls_checksums/checksums.txt. They are used to check the integrity of the datasets downloaded files. I kept the same format as tensorflow-datasets. There is one checksums file for all configs. ### Load a dataset When you do `load("squad")`, it will also download the checksums file and put it next to the script in nlp/datasets/hash/urls_checksums/checksums.txt. It also verifies that the downloaded files checksums match the expected ones. You can ignore checksum tests with `load("squad", ignore_checksums=True)` (under the hood it just adds `ignore_checksums=True` in the `DownloadConfig`) ### Test a dataset There is a new command `nlp-cli test squad` that runs `download_and_prepare` to see if it runs ok, and that verifies that all the checksums match. Allowed arguments are `--name`, `--all_configs`, `--ignore_checksums` and `--register_checksums`. ### Register checksums 1. If the dataset has external dataset files The command `nlp-cli test squad --register_checksums --all_configs` runs `download_and_prepare` on all configs to see if it runs ok, and it creates the checksums file. You can also register one config at a time using `--name` instead ; the checksums file will be completed and not overwritten. If the script is a local script, the checksum file is moved to urls_checksums/checksums.txt next to the local script, to enable the user to upload both the script and the checksums file afterwards with `nlp-cli upload squad`. 2. If the dataset files are all inside the directory of the dataset script The user can directly do `nlp-cli upload squad --register_checksums`, as there is no need to download anything. In this case however, all the dataset must be uploaded at once. -- PS : it doesn't allow to register checksums for canonical datasets, the file has to be added manually on S3 for now (I guess ?) Also I feel like we must be sure that this processes would not constrain too much any user from uploading its dataset. Let me know what you think :)

https://api.github.com/repos/huggingface/datasets/issues/24/timeline

https://api.github.com/repos/huggingface/datasets/issues/23

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/23/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/23/comments

https://api.github.com/repos/huggingface/datasets/issues/23/events

https://github.com/huggingface/datasets/pull/23

MDExOlB1bGxSZXF1ZXN0NDEwMjczOTU2

Add metrics

closed

CONTRIBUTOR

This PR is a draft for adding metrics (sacrebleu and seqeval are added) use case examples: `import nlp` **sacrebleu:** ``` refs = [['The dog bit the man.', 'It was not unexpected.', 'The man bit him first.'], ['The dog had bit the man.', 'No one was surprised.', 'The man had bitten the dog.']] sys = ['The dog bit the man.', "It wasn't surprising.", 'The man had just bitten him.'] sacrebleu = nlp.load_metrics('sacrebleu') print(sacrebleu.score) ``` **seqeval:** ``` y_true = [['O', 'O', 'O', 'B-MISC', 'I-MISC', 'I-MISC', 'O'], ['B-PER', 'I-PER', 'O']] y_pred = [['O', 'O', 'B-MISC', 'I-MISC', 'I-MISC', 'I-MISC', 'O'], ['B-PER', 'I-PER', 'O']] seqeval = nlp.load_metrics('seqeval') print(seqeval.accuracy_score(y_true, y_pred) print(seqeval.f1_score(y_true, y_pred) ``` _examples are taken from the corresponding web page_ your comments and suggestions are more than welcomed

https://api.github.com/repos/huggingface/datasets/issues/23/timeline

https://api.github.com/repos/huggingface/datasets/issues/22

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/22/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/22/comments

https://api.github.com/repos/huggingface/datasets/issues/22/events

https://github.com/huggingface/datasets/pull/22

MDExOlB1bGxSZXF1ZXN0NDEwMTAyMjU3

adding bleu score code

closed

CONTRIBUTOR

this PR add the BLEU score metric to the lib. It can be tested by running the following code. ` from nlp.metrics import bleu hyp1 = "It is a guide to action which ensures that the military always obeys the commands of the party" ref1a = "It is a guide to action that ensures that the military forces always being under the commands of the party " ref1b = "It is the guiding principle which guarantees the military force always being under the command of the Party" ref1c = "It is the practical guide for the army always to heed the directions of the party" list_of_references = [[ref1a, ref1b, ref1c]] hypotheses = [hyp1] bleu = bleu.bleu_score(list_of_references, hypotheses,4, smooth=True) print(bleu) `

https://api.github.com/repos/huggingface/datasets/issues/22/timeline

https://api.github.com/repos/huggingface/datasets/issues/21

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/21/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/21/comments

https://api.github.com/repos/huggingface/datasets/issues/21/events

https://github.com/huggingface/datasets/pull/21

MDExOlB1bGxSZXF1ZXN0NDA5Nzk2MTM4

Cleanup Features - Updating convert command - Fix Download manager

closed

MEMBER

This PR makes a number of changes: # Updating `Features` Features are a complex mechanism provided in `tfds` to be able to modify a dataset on-the-fly when serializing to disk and when loading from disk. We don't really need this because (1) it hides too much from the user and (2) our datatype can be directly mapped to Arrow tables on drive so we usually don't need to change the format before/after serialization. This PR extracts and refactors these features in a single `features.py` files. It still keep a number of features classes for easy compatibility with tfds, namely the `Sequence`, `Tensor`, `ClassLabel` and `Translation` features. Some more complex features involving a pre-processing on-the-fly during serialization are kept: - `ClassLabel` which are able to convert from label strings to integers, - `Translation`which does some check on the languages. # Updating the `convert` command We do a few updates here - following the simplification of the `features` (cf above), conversion are updated - we also makes it simpler to convert a single file - some code need to be fixed manually after conversion (e.g. to remove some encoding processing in former tfds `Text` features. We highlight this code with a "git merge conflict" style syntax for easy manual fixing. # Fix download manager iterator You kept me up quite late on Tuesday night with this `os.scandir` change @lhoestq ;-)

https://api.github.com/repos/huggingface/datasets/issues/21/timeline

https://api.github.com/repos/huggingface/datasets/issues/20

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/20/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/20/comments

https://api.github.com/repos/huggingface/datasets/issues/20/events

https://github.com/huggingface/datasets/pull/20

MDExOlB1bGxSZXF1ZXN0NDA5MzEyMDI1

remove boto3 and promise dependencies

closed

MEMBER

With the new download manager, we don't need `promise` anymore. I also removed `boto3` as in [this pr](https://github.com/huggingface/transformers/pull/3968)

https://api.github.com/repos/huggingface/datasets/issues/20/timeline

https://api.github.com/repos/huggingface/datasets/issues/19

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/19/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/19/comments

https://api.github.com/repos/huggingface/datasets/issues/19/events

https://github.com/huggingface/datasets/pull/19

MDExOlB1bGxSZXF1ZXN0NDA4NjIwMjUw

Replace tf.constant for TF

closed

CONTRIBUTOR

Replace simple tf.constant type of Tensor to tf.ragged.constant which allows to have examples of different size in a tf.data.Dataset. Now the training works with TF. Here the same example than for the PT in collab: ```python import tensorflow as tf import nlp from transformers import BertTokenizerFast, TFBertForQuestionAnswering # Load our training dataset and tokenizer train_dataset = nlp.load('squad', split="train[:1%]") tokenizer = BertTokenizerFast.from_pretrained('bert-base-cased') def get_correct_alignement(context, answer): start_idx = answer['answer_start'][0] text = answer['text'][0] end_idx = start_idx + len(text) if context[start_idx:end_idx] == text: return start_idx, end_idx # When the gold label position is good elif context[start_idx-1:end_idx-1] == text: return start_idx-1, end_idx-1 # When the gold label is off by one character elif context[start_idx-2:end_idx-2] == text: return start_idx-2, end_idx-2 # When the gold label is off by two character else: raise ValueError() # Tokenize our training dataset def convert_to_features(example_batch): # Tokenize contexts and questions (as pairs of inputs) input_pairs = list(zip(example_batch['context'], example_batch['question'])) encodings = tokenizer.batch_encode_plus(input_pairs, pad_to_max_length=True) # Compute start and end tokens for labels using Transformers's fast tokenizers alignement methods. start_positions, end_positions = [], [] for i, (context, answer) in enumerate(zip(example_batch['context'], example_batch['answers'])): start_idx, end_idx = get_correct_alignement(context, answer) start_positions.append([encodings.char_to_token(i, start_idx)]) end_positions.append([encodings.char_to_token(i, end_idx-1)]) if start_positions and end_positions: encodings.update({'start_positions': start_positions, 'end_positions': end_positions}) return encodings train_dataset = train_dataset.map(convert_to_features, batched=True) columns = ['input_ids', 'token_type_ids', 'attention_mask', 'start_positions', 'end_positions'] train_dataset.set_format(type='tensorflow', columns=columns) features = {x: train_dataset[x] for x in columns[:3]} labels = {"output_1": train_dataset["start_positions"]} labels["output_2"] = train_dataset["end_positions"] tfdataset = tf.data.Dataset.from_tensor_slices((features, labels)).batch(8) model = TFBertForQuestionAnswering.from_pretrained("bert-base-cased") loss_fn = tf.keras.losses.SparseCategoricalCrossentropy(reduction=tf.keras.losses.Reduction.NONE, from_logits=True) opt = tf.keras.optimizers.Adam(learning_rate=3e-5) model.compile(optimizer=opt, loss={'output_1': loss_fn, 'output_2': loss_fn}, loss_weights={'output_1': 1., 'output_2': 1.}, metrics=['accuracy']) model.fit(tfdataset, epochs=1, steps_per_epoch=3) ```

https://api.github.com/repos/huggingface/datasets/issues/19/timeline

https://api.github.com/repos/huggingface/datasets/issues/18

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/18/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/18/comments

https://api.github.com/repos/huggingface/datasets/issues/18/events

https://github.com/huggingface/datasets/pull/18

MDExOlB1bGxSZXF1ZXN0NDA4Mzg0MTc3

Updating caching mechanism - Allow dependency in dataset processing scripts - Fix style and quality in the repo

closed

MEMBER

This PR has a lot of content (might be hard to review, sorry, in particular because I fixed the style in the repo at the same time). # Style & quality: You can now install the style and quality tools with `pip install -e .[quality]`. This will install black, the compatible version of sort and flake8. You can then clean the style and check the quality before merging your PR with: ```bash make style make quality ``` # Allow dependencies in dataset processing scripts We can now allow (some level) of imports in dataset processing scripts (in addition to PyPi imports). Namely, you can do the two following things: Import from a relative path to a file in the same folder as the dataset processing script: ```python import .c4_utils ``` Or import from a relative path to a file in a folder/archive/github repo to which you provide an URL after the import state with `# From: [URL]`: ```python import .clicr.dataset_code.build_json_dataset # From: https://github.com/clips/clicr ``` In both these cases, after downloading the main dataset processing script, we will identify the location of these dependencies, download them and copy them in the dataset processing script folder. Note that only direct import in the dataset processing script will be handled. We don't recursively explore the additional import to download further files. Also, when we download from an additional directory (in the second case above), we recursively add `__init__.py` to all the sub-folder so you can import from them. This part is still tested for now. If you've seen datasets which required external utilities, tell me and I can test it. # Update the cache to have a better local structure The local structure in the `src/datasets` folder is now: `src/datasets/DATASET_NAME/DATASET_HASH/*` The hash is computed from the full code of the dataset processing script as well as all the local and downloaded dependencies as mentioned above. This way if you change some code in a utility related to your dataset, a new hash should be computed.

https://api.github.com/repos/huggingface/datasets/issues/18/timeline

https://api.github.com/repos/huggingface/datasets/issues/17

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/17/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/17/comments

https://api.github.com/repos/huggingface/datasets/issues/17/events

https://github.com/huggingface/datasets/pull/17

MDExOlB1bGxSZXF1ZXN0NDA4MDk3NjM0

Add Pandas as format type

closed

CONTRIBUTOR

As detailed in the title ^^

https://api.github.com/repos/huggingface/datasets/issues/17/timeline

https://api.github.com/repos/huggingface/datasets/issues/16

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/16/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/16/comments

https://api.github.com/repos/huggingface/datasets/issues/16/events

https://github.com/huggingface/datasets/pull/16

MDExOlB1bGxSZXF1ZXN0NDA4MDIyMTUz

create our own DownloadManager

closed

MEMBER

I tried to create our own - and way simpler - download manager, by replacing all the complicated stuff with our own `cached_path` solution. With this implementation, I tried `dataset = nlp.load('squad')` and it seems to work fine. For the implementation, what I did exactly: - I copied the old download manager - I removed all the dependences to the old `download` files - I replaced all the download + extract calls by calls to `cached_path` - I removed unused parameters (extract_dir, compute_stats) (maybe compute_stats could be re-added later if we want to compute stats...) - I left some functions unimplemented for now. We will probably have to implement them because they are used by some datasets scripts (download_kaggle_data, iter_archive) or because we may need them at some point (download_checksums, _record_sizes_checksums) Let me know if you think that this is going the right direction or if you have remarks. Note: I didn't write any test yet as I wanted to read your remarks first

https://api.github.com/repos/huggingface/datasets/issues/16/timeline

https://api.github.com/repos/huggingface/datasets/issues/15

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/15/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/15/comments

https://api.github.com/repos/huggingface/datasets/issues/15/events

https://github.com/huggingface/datasets/pull/15

MDExOlB1bGxSZXF1ZXN0NDA3NDEwOTk3

[Tests] General Test Design for all dataset scripts

closed

CONTRIBUTOR

The general idea is similar to how testing is done in `transformers`. There is one general `test_dataset_common.py` file which has a `DatasetTesterMixin` class. This class implements all of the logic that can be used in a generic way for all dataset classes. The idea is to keep each individual dataset test file as minimal as possible. In order to test whether the specific data set class can download the data and generate the examples **without** downloading the actual data all the time, a MockDataLoaderManager class is used which receives a `mock_folder_structure_fn` function from each individual dataset test file that create "fake" data and which returns the same folder structure that would have been created when using the real data downloader.

https://api.github.com/repos/huggingface/datasets/issues/15/timeline

https://api.github.com/repos/huggingface/datasets/issues/14

https://api.github.com/repos/huggingface/datasets

https://api.github.com/repos/huggingface/datasets/issues/14/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/14/comments

https://api.github.com/repos/huggingface/datasets/issues/14/events

https://github.com/huggingface/datasets/pull/14

MDExOlB1bGxSZXF1ZXN0NDA3MjkzNjU5

[Download] Only create dir if not already exist

closed