Datasets:

Okyx

/

github-issues

like 0

https://api.github.com/repos/huggingface/datasets/issues/4773

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

https://api.github.com/repos/huggingface/datasets/issues/4773/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4773/comments

https://api.github.com/repos/huggingface/datasets/issues/4773/events

https://github.com/huggingface/datasets/pull/4773

PR_kwDODunzps48WNV3

Document loading from relative path

open

MEMBER

This PR describes loading a dataset from the Hub by specifying a relative path in `data_dir` or `data_files` in `load_dataset` (see #4757).

https://api.github.com/repos/huggingface/datasets/issues/4773/timeline

https://api.github.com/repos/huggingface/datasets/issues/4772

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

https://api.github.com/repos/huggingface/datasets/issues/4772/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4772/comments

https://api.github.com/repos/huggingface/datasets/issues/4772/events

https://github.com/huggingface/datasets/issues/4772

I_kwDODunzps5O1rID

AssertionError when using label_cols in to_tf_dataset

open

NONE

## Describe the bug An incorrect `AssertionError` is raised when using `label_cols` in `to_tf_dataset` and the label's key name is `label`. The assertion is in this line: https://github.com/huggingface/datasets/blob/2.4.0/src/datasets/arrow_dataset.py#L475 ## Steps to reproduce the bug ```python from datasets import load_dataset from transformers import DefaultDataCollator dataset = load_dataset('glue', 'mrpc', split='train') tf_dataset = dataset.to_tf_dataset( columns=["sentence1", "sentence2", "idx"], label_cols=["label"], batch_size=16, collate_fn=DefaultDataCollator(return_tensors="tf"), ) ``` ## Expected results No assertion error. ## Actual results ``` AssertionError: in user code: File "/opt/conda/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 475, in split_features_and_labels * assert set(features.keys()).union(labels.keys()) == set(input_batch.keys()) ``` ## Environment info <!-- You can run the command `datasets-cli env` and copy-and-paste its output below. --> - `datasets` version: 2.4.0 - Platform: Linux-4.18.0-305.45.1.el8_4.ppc64le-ppc64le-with-glibc2.17 - Python version: 3.8.13 - PyArrow version: 7.0.0 - Pandas version: 1.4.3

https://api.github.com/repos/huggingface/datasets/issues/4772/timeline

https://api.github.com/repos/huggingface/datasets/issues/4771

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

https://api.github.com/repos/huggingface/datasets/issues/4771/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4771/comments

https://api.github.com/repos/huggingface/datasets/issues/4771/events

https://github.com/huggingface/datasets/pull/4771

PR_kwDODunzps48VjWx

Remove dummy data generation docs

open

MEMBER

This PR removes instructions to generate dummy data since that is no longer necessary for datasets that are uploaded to the Hub instead of our GitHub repo (see #4744).

https://api.github.com/repos/huggingface/datasets/issues/4771/timeline

https://api.github.com/repos/huggingface/datasets/issues/4770

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

https://api.github.com/repos/huggingface/datasets/issues/4770/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4770/comments

https://api.github.com/repos/huggingface/datasets/issues/4770/events

https://github.com/huggingface/datasets/pull/4770

PR_kwDODunzps48UEBT

fix typo

closed

CONTRIBUTOR

By defaul -> By default

https://api.github.com/repos/huggingface/datasets/issues/4770/timeline

https://api.github.com/repos/huggingface/datasets/issues/4769

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

https://api.github.com/repos/huggingface/datasets/issues/4769/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4769/comments

https://api.github.com/repos/huggingface/datasets/issues/4769/events

https://github.com/huggingface/datasets/issues/4769

I_kwDODunzps5OzflS

Fail to process SQuADv1.1 datasets with max_seq_length=128, doc_stride=96.

open

NONE

## Describe the bug datasets fail to process SQuADv1.1 with max_seq_length=128, doc_stride=96 when calling datasets["train"].train_dataset.map(). ## Steps to reproduce the bug I used huggingface[ TF2 question-answering examples](https://github.com/huggingface/transformers/tree/main/examples/tensorflow/question-answering). And my scripts are as follows: ``` python run_qa.py \ --model_name_or_path $BERT_DIR \ --dataset_name $SQUAD_DIR \ --do_train \ --do_eval \ --per_device_train_batch_size 12 \ --learning_rate 3e-5 \ --num_train_epochs 2 \ --max_seq_length 128 \ --doc_stride 96 \ --output_dir $OUTPUT \ --save_steps 10000 \ --overwrite_cache \ --overwrite_output_dir \ ``` ## Expected results Normally process SQuADv1.1 datasets with max_seq_length=128, doc_stride=96. ## Actual results ``` INFO:__main__:Padding all batches to max length because argument was set or we're on TPU. WARNING:datasets.fingerprint:Parameter 'function'=<function main.<locals>.prepare_train_features at 0x7f15bc2d07a0> of the transform datasets.arrow_dataset.Dataset._map_single couldn't be hashed properly, a random hash was used instead. Make sure your transforms and parameters are serializable with pickle or dill for the dataset fingerprinting and caching to work. If you reuse this transform, the caching mechanism will consider it to be different from the previous calls and recompute everything. This warning is only showed once. Subsequent hashing failures won't be showed. 0%| | 0/88 [00:00<?, ?ba/s]thread '<unnamed>' panicked at 'assertion failed: stride < max_len', /__w/tokenizers/tokenizers/tokenizers/src/tokenizer/encoding.rs:311:9 note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace 0%| | 0/88 [00:00<?, ?ba/s] Traceback (most recent call last): File "run_qa.py", line 743, in <module> main() File "run_qa.py", line 485, in main load_from_cache_file=not data_args.overwrite_cache, File "/anaconda3/envs/py37/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 2394, in map desc=desc, File "/anaconda3/envs/py37/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 551, in wrapper out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs) File "/anaconda3/envs/py37/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 518, in wrapper out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs) File "/anaconda3/envs/py37/lib/python3.7/site-packages/datasets/fingerprint.py", line 458, in wrapper out = func(self, *args, **kwargs) File "anaconda3/envs/py37/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 2768, in _map_single offset=offset, File "anaconda3/envs/py37/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 2644, in apply_function_on_filtered_inputs processed_inputs = function(*fn_args, *additional_args, **fn_kwargs) File "anaconda3/envs/py37/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 2336, in decorated result = f(decorated_item, *args, **kwargs) File "run_qa.py", line 410, in prepare_train_features padding=padding, File "anaconda3/envs/py37/lib/python3.7/site-packages/transformers/tokenization_utils_base.py", line 2512, in __call__ **kwargs, File "anaconda3/envs/py37/lib/python3.7/site-packages/transformers/tokenization_utils_base.py", line 2703, in batch_encode_plus **kwargs, File "anaconda3/envs/py37/lib/python3.7/site-packages/transformers/tokenization_utils_fast.py", line 429, in _batch_encode_plus is_pretokenized=is_split_into_words, pyo3_runtime.PanicException: assertion failed: stride < max_len Traceback (most recent call last): File "./data/SQuADv1.1/evaluate-v1.1.py", line 92, in <module> with open(args.prediction_file) as prediction_file: FileNotFoundError: [Errno 2] No such file or directory: './output/bert_base_squadv1.1_tf2/eval_predictions.json' ``` ## Environment info <!-- You can run the command `datasets-cli env` and copy-and-paste its output below. --> - `datasets` version: 2.3.2 - Platform: Ubuntu, pytorch=1.11.0, tensorflow-gpu=2.9.1 - Python version: 2.7 - PyArrow version: 8.0.0

https://api.github.com/repos/huggingface/datasets/issues/4769/timeline

https://api.github.com/repos/huggingface/datasets/issues/4768

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

https://api.github.com/repos/huggingface/datasets/issues/4768/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4768/comments

https://api.github.com/repos/huggingface/datasets/issues/4768/events

https://github.com/huggingface/datasets/pull/4768

PR_kwDODunzps48TRUH

Unpin rouge_score test dependency

closed

MEMBER

Once `rouge-score` has made the 0.1.2 release to fix their issue https://github.com/google-research/google-research/issues/1212, we can unpin it. Related to: - #4735

https://api.github.com/repos/huggingface/datasets/issues/4768/timeline

https://api.github.com/repos/huggingface/datasets/issues/4767

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

https://api.github.com/repos/huggingface/datasets/issues/4767/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4767/comments

https://api.github.com/repos/huggingface/datasets/issues/4767/events

https://github.com/huggingface/datasets/pull/4767

PR_kwDODunzps48TCpI

Add 2.4.0 version added to docstrings

closed

MEMBER

https://api.github.com/repos/huggingface/datasets/issues/4767/timeline

https://api.github.com/repos/huggingface/datasets/issues/4766

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

https://api.github.com/repos/huggingface/datasets/issues/4766/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4766/comments

https://api.github.com/repos/huggingface/datasets/issues/4766/events

https://github.com/huggingface/datasets/issues/4766

I_kwDODunzps5OyTXk

Dataset Viewer issue for openclimatefix/goes-mrms

open

NONE

### Link _No response_ ### Description _No response_ ### Owner _No response_

https://api.github.com/repos/huggingface/datasets/issues/4766/timeline

https://api.github.com/repos/huggingface/datasets/issues/4765

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

https://api.github.com/repos/huggingface/datasets/issues/4765/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4765/comments

https://api.github.com/repos/huggingface/datasets/issues/4765/events

https://github.com/huggingface/datasets/pull/4765

PR_kwDODunzps48S2rM

Fix version in map_nested docstring

closed

MEMBER

After latest release, `map_nested` docstring needs being updated with the right version for versionchanged and versionadded.

https://api.github.com/repos/huggingface/datasets/issues/4765/timeline

https://api.github.com/repos/huggingface/datasets/issues/4764

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

https://api.github.com/repos/huggingface/datasets/issues/4764/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4764/comments

https://api.github.com/repos/huggingface/datasets/issues/4764/events

https://github.com/huggingface/datasets/pull/4764

PR_kwDODunzps48RMLu

Update CI badge

closed

CONTRIBUTOR

Replace the old CircleCI badge with a new one for GH Actions.

https://api.github.com/repos/huggingface/datasets/issues/4764/timeline

https://api.github.com/repos/huggingface/datasets/issues/4763

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

https://api.github.com/repos/huggingface/datasets/issues/4763/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4763/comments

https://api.github.com/repos/huggingface/datasets/issues/4763/events

https://github.com/huggingface/datasets/pull/4763

PR_kwDODunzps48RMKi

More rigorous shape inference in to_tf_dataset

open

MEMBER

`tf.data` needs to know the shape of tensors emitted from a `tf.data.Dataset`. Although `None` dimensions are possible, overusing them can cause problems - Keras uses the dataset tensor spec at compile-time, and so saying that a dimension is `None` when it's actually constant can hurt performance, or even cause training to fail for dimensions that are needed to determine the shape of weight tensors! The compromise I used here was to sample several batches from the underlying dataset and apply the `collate_fn` to them, and then to see which dimensions were "empirically variable". There's an obvious problem here, though - if you sample 10 batches and they all have the same shape on a certain dimension, there's still a small chance that the 11th batch will be different, and Keras will throw an error if a dataset tries to emit a tensor whose shape doesn't match the spec. I encountered this bug in practice once or twice for datasets that were mostly-but-not-totally constant on a given dimension, and I still don't have a perfect solution, but this PR should greatly reduce the risk. It samples many more batches, and also samples very small batches (size 2) - this increases the variability, making it more likely that a few outlier samples will be detected. Ideally, of course, we'd determine the full output shape analytically, but that's surprisingly tricky when the `collate_fn` can be any arbitrary Python code!

https://api.github.com/repos/huggingface/datasets/issues/4763/timeline

https://api.github.com/repos/huggingface/datasets/issues/4762

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

https://api.github.com/repos/huggingface/datasets/issues/4762/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4762/comments

https://api.github.com/repos/huggingface/datasets/issues/4762/events

https://github.com/huggingface/datasets/pull/4762

PR_kwDODunzps48RE56

Improve features resolution in streaming

open

MEMBER

`IterableDataset._resolve_features` was returning the features sorted alphabetically by column name, which is not consistent with non-streaming. I changed this and used the order of columns from the data themselves. It was causing some inconsistencies in the dataset viewer as well. I also fixed `interleave_datasets` that was not filling missing columns with None, because it was not using the columns from `IterableDataset._resolve_features` cc @severo

https://api.github.com/repos/huggingface/datasets/issues/4762/timeline

https://api.github.com/repos/huggingface/datasets/issues/4761

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

https://api.github.com/repos/huggingface/datasets/issues/4761/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4761/comments

https://api.github.com/repos/huggingface/datasets/issues/4761/events

https://github.com/huggingface/datasets/issues/4761

I_kwDODunzps5Oved7

parallel searching in multi-gpu setting using faiss

open

CONTRIBUTOR

While I notice that `add_faiss_index` has supported assigning multiple GPUs, I am still confused about how it works. Does the `search-batch` function automatically parallelizes the input queries to different gpus?https://github.com/huggingface/datasets/blob/d76599bdd4d186b2e7c4f468b05766016055a0a5/src/datasets/search.py#L360

https://api.github.com/repos/huggingface/datasets/issues/4761/timeline

https://api.github.com/repos/huggingface/datasets/issues/4760

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

https://api.github.com/repos/huggingface/datasets/issues/4760/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4760/comments

https://api.github.com/repos/huggingface/datasets/issues/4760/events

https://github.com/huggingface/datasets/issues/4760

I_kwDODunzps5OuwCP

Issue with offline mode

open

NONE

## Describe the bug I can't retrieve a cached dataset with offline mode enabled ## Steps to reproduce the bug To reproduce my issue, first, you'll need to run a script that will cache the dataset ```python import os os.environ["HF_DATASETS_OFFLINE"] = "0" import datasets datasets.logging.set_verbosity_info() ds_name = "SaulLu/toy_struc_dataset" ds = datasets.load_dataset(ds_name) print(ds) ``` then, you can try to reload it in offline mode: ```python import os os.environ["HF_DATASETS_OFFLINE"] = "1" import datasets datasets.logging.set_verbosity_info() ds_name = "SaulLu/toy_struc_dataset" ds = datasets.load_dataset(ds_name) print(ds) ``` ## Expected results I would have expected the 2nd snippet not to return any errors ## Actual results The 2nd snippet returns: ``` Traceback (most recent call last): File "/home/lucile_huggingface_co/sandbox/evaluate/test_cache_datasets.py", line 8, in <module> ds = datasets.load_dataset(ds_name) File "/home/lucile_huggingface_co/anaconda3/envs/evaluate-dev/lib/python3.8/site-packages/datasets/load.py", line 1723, in load_dataset builder_instance = load_dataset_builder( File "/home/lucile_huggingface_co/anaconda3/envs/evaluate-dev/lib/python3.8/site-packages/datasets/load.py", line 1500, in load_dataset_builder dataset_module = dataset_module_factory( File "/home/lucile_huggingface_co/anaconda3/envs/evaluate-dev/lib/python3.8/site-packages/datasets/load.py", line 1241, in dataset_module_factory raise ConnectionError(f"Couln't reach the Hugging Face Hub for dataset '{path}': {e1}") from None ConnectionError: Couln't reach the Hugging Face Hub for dataset 'SaulLu/toy_struc_dataset': Offline mode is enabled. ``` ## Environment info <!-- You can run the command `datasets-cli env` and copy-and-paste its output below. --> - `datasets` version: 2.4.0 - Platform: Linux-4.19.0-21-cloud-amd64-x86_64-with-glibc2.17 - Python version: 3.8.13 - PyArrow version: 8.0.0 - Pandas version: 1.4.3 Maybe I'm misunderstanding something in the use of the offline mode (see [doc](https://huggingface.co/docs/datasets/v2.4.0/en/loading#offline)), is that the case?

https://api.github.com/repos/huggingface/datasets/issues/4760/timeline

https://api.github.com/repos/huggingface/datasets/issues/4759

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

https://api.github.com/repos/huggingface/datasets/issues/4759/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4759/comments

https://api.github.com/repos/huggingface/datasets/issues/4759/events

https://github.com/huggingface/datasets/issues/4759

I_kwDODunzps5OuY3E

Dataset Viewer issue for Toygar/turkish-offensive-language-detection

closed

NONE

### Link https://huggingface.co/datasets/Toygar/turkish-offensive-language-detection ### Description Status code: 400 Exception: Status400Error Message: The dataset does not exist. Hi, I provided train.csv, test.csv and valid.csv files. However, viewer says dataset does not exist. Should I need to do anything else? ### Owner Yes

https://api.github.com/repos/huggingface/datasets/issues/4759/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4757

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

https://api.github.com/repos/huggingface/datasets/issues/4757/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4757/comments

https://api.github.com/repos/huggingface/datasets/issues/4757/events

https://github.com/huggingface/datasets/issues/4757

I_kwDODunzps5Otsuk

Document better when relative paths are transformed to URLs

open

MEMBER

As discussed with @ydshieh, when passing a relative path as `data_dir` to `load_dataset` of a dataset hosted on the Hub, the relative path is transformed to the corresponding URL of the Hub dataset. Currently, we mention this in our docs here: [Create a dataset loading script > Download data files and organize splits](https://huggingface.co/docs/datasets/v2.4.0/en/dataset_script#download-data-files-and-organize-splits) > If the data files live in the same folder or repository of the dataset script, you can just pass the relative paths to the files instead of URLs. Maybe we should document better how relative paths are handled, not only when creating a dataset loading script, but also when passing to `load_dataset`: - `data_dir` - `data_files` CC: @stevhliu

https://api.github.com/repos/huggingface/datasets/issues/4757/timeline

https://api.github.com/repos/huggingface/datasets/issues/4755

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

https://api.github.com/repos/huggingface/datasets/issues/4755/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4755/comments

https://api.github.com/repos/huggingface/datasets/issues/4755/events

https://github.com/huggingface/datasets/issues/4755

I_kwDODunzps5OqNOE

Datasets.map causes incorrect overflow_to_sample_mapping when used with tokenizers and small batch size

open

NONE

## Describe the bug When using `tokenizer`, we can retrieve the field `overflow_to_sample_mapping`, since long samples will be overflown into multiple token sequences. However, when tokenizing is done via `Dataset.map`, with `n_proc > 1`, the `overflow_to_sample_mapping` field is wrong. This seems to be because each tokenizer only looks at its share of the samples, and maps to the index _within its share_, but then `Dataset.map` collates them together. ## Steps to reproduce the bug 1. Make a dataset of 3 strings. 2. Tokenize via Dataset.map with n_proc = 8 3. Inspect the `overflow_to_sample_mapping` field ## Expected results `[0, 1, 2]` ## Actual results `[0, 0, 0]` Notes: 1. I have not yet extracted a minimal example, but the above works reliably 2. If the dataset is large, I've yet to determine if this bug still happens a. not at all b. always c. on the small, leftover batch at the end.

https://api.github.com/repos/huggingface/datasets/issues/4755/timeline

https://api.github.com/repos/huggingface/datasets/issues/4754

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

https://api.github.com/repos/huggingface/datasets/issues/4754/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4754/comments

https://api.github.com/repos/huggingface/datasets/issues/4754/events

https://github.com/huggingface/datasets/pull/4754

PR_kwDODunzps48L9p6

Remove "unkown" language tags

closed

MEMBER

Following https://github.com/huggingface/datasets/pull/4753 there was still a "unknown" langauge tag in `wikipedia` so the job at https://github.com/huggingface/datasets/runs/7542567336?check_suite_focus=true failed for wikipedia

https://api.github.com/repos/huggingface/datasets/issues/4754/timeline

https://api.github.com/repos/huggingface/datasets/issues/4753

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

https://api.github.com/repos/huggingface/datasets/issues/4753/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4753/comments

https://api.github.com/repos/huggingface/datasets/issues/4753/events

https://github.com/huggingface/datasets/pull/4753

PR_kwDODunzps48Ll8G

Add `language_bcp47` tag

closed

MEMBER

Following (internal) https://github.com/huggingface/moon-landing/pull/3509, we need to move the bcp47 tags to `language_bcp47` and keep the `language` tag for iso 639 1-2-3 codes. In particular I made sure that all the tags in `languages` are not longer than 3 characters. I moved the rest to `language_bcp47` and fixed some of them. After this PR is merged I think we can simplify the language validation from the DatasetMetadata class (and keep it bare-bone just for the tagging app) PS: the CI is failing because of missing content in dataset cards that are unrelated to this PR

https://api.github.com/repos/huggingface/datasets/issues/4753/timeline

https://api.github.com/repos/huggingface/datasets/issues/4752

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

https://api.github.com/repos/huggingface/datasets/issues/4752/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4752/comments

https://api.github.com/repos/huggingface/datasets/issues/4752/events

https://github.com/huggingface/datasets/issues/4752

I_kwDODunzps5OpW3Z

DatasetInfo issue when testing multiple configs: mixed task_templates

open

CONTRIBUTOR

## Describe the bug When running the `datasets-cli test` it would seem that some config properties in a DatasetInfo get mangled, leading to issues, e.g., about the ClassLabel. ## Steps to reproduce the bug In summary, what I want to do is create three configs: - unfiltered: no classlabel, no tasks. Gets data from unfiltered.json.gz (I'd want this without splits, just one chunk of data, but that does not seem possible?) - filtered_sentiment: `review_sentiment` as ClassLabel, TextClassification task with `review_sentiment` as label. Gets train/test split from respective json.gz files - filtered_rating: `review_rating0` as ClassLabel, TextClassification task with `review_rating0` as label. Gets train/test split from respective json.gz files This might be a bit tedious to reproduce, so I am sorry, but these are the steps: - Clone datasets -> `datasets/` and install it - Clone `https://huggingface.co/datasets/BramVanroy/hebban-reviews` into `datasets/datasets` so that you have a new folder `datasets/datasets/hebban-reviews/`. - Replace the HebbanReviews class with this new one: ```python class HebbanReviews(datasets.GeneratorBasedBuilder): """The Hebban book reviews dataset.""" BUILDER_CONFIGS = [ HebbanReviewsConfig( name="unfiltered", description=_HEBBAN_REVIEWS_UNFILTERED_DESCRIPTION, version=datasets.Version(_HEBBAN_VERSION) ), HebbanReviewsConfig( name="filtered_sentiment", description=f"This config has the negative, neutral, and positive sentiment scores as ClassLabel in the 'review_sentiment' column.\n{_HEBBAN_REVIEWS_FILTERED_DESCRIPTION}", version=datasets.Version(_HEBBAN_VERSION) ), HebbanReviewsConfig( name="filtered_rating", description=f"This config has the 5-class ratings as ClassLabel in the 'review_rating0' column (which is a variant of 'review_rating' that starts counting from 0 instead of 1).\n{_HEBBAN_REVIEWS_FILTERED_DESCRIPTION}", version=datasets.Version(_HEBBAN_VERSION) ) ] DEFAULT_CONFIG_NAME = "filtered_sentiment" _URLS = { "train": "train.jsonl.gz", "test": "test.jsonl.gz", "unfiltered": "unfiltered.jsonl.gz", } def _info(self): features = { "review_title": datasets.Value("string"), "review_text": datasets.Value("string"), "review_text_without_quotes": datasets.Value("string"), "review_n_quotes": datasets.Value("int32"), "review_n_tokens": datasets.Value("int32"), "review_rating": datasets.Value("int32"), "review_rating0": datasets.Value("int32"), "review_author_url": datasets.Value("string"), "review_author_type": datasets.Value("string"), "review_n_likes": datasets.Value("int32"), "review_n_comments": datasets.Value("int32"), "review_url": datasets.Value("string"), "review_published_date": datasets.Value("string"), "review_crawl_date": datasets.Value("string"), "lid": datasets.Value("string"), "lid_probability": datasets.Value("float32"), "review_sentiment": datasets.features.ClassLabel(names=["negative", "neutral", "positive"]), "review_sentiment_label": datasets.Value("string"), "book_id": datasets.Value("int32"), } if self.config.name == "filtered_sentiment": task_templates = [datasets.TextClassification(text_column="review_text_without_quotes", label_column="review_sentiment")] elif self.config.name == "filtered_rating": # For CrossEntropy, our classes need to start at index 0 -- not 1 features["review_rating0"] = datasets.features.ClassLabel(names=["1", "2", "3", "4", "5"]) features["review_sentiment"] = datasets.Value("int32") task_templates = [datasets.TextClassification(text_column="review_text_without_quotes", label_column="review_rating0")] elif self.config.name == "unfiltered": # no ClassLabels in unfiltered features["review_sentiment"] = datasets.Value("int32") task_templates = None else: raise ValueError(f"Unsupported config {self.config.name}. Expected one of 'filtered_sentiment' (default)," f" 'filtered_rating', or 'unfiltered'") print("AT INFO", self.config.name, task_templates) return datasets.DatasetInfo( description=self.config.description, features=datasets.Features(features), homepage="https://huggingface.co/datasets/BramVanroy/hebban-reviews", citation=_HEBBAN_REVIEWS_CITATION, task_templates=task_templates, license="cc-by-4.0" ) def _split_generators(self, dl_manager): if self.config.name.startswith("filtered"): files = dl_manager.download_and_extract({"train": "train.jsonl.gz", "test": "test.jsonl.gz"}) return [ datasets.SplitGenerator( name=datasets.Split.TRAIN, gen_kwargs={ "data_file": files["train"] }, ), datasets.SplitGenerator( name=datasets.Split.TEST, gen_kwargs={ "data_file": files["test"] }, ), ] elif self.config.name == "unfiltered": files = dl_manager.download_and_extract({"train": "unfiltered.jsonl.gz"}) return [ datasets.SplitGenerator( name=datasets.Split.TRAIN, gen_kwargs={ "data_file": files["train"] }, ), ] else: raise ValueError(f"Unsupported config {self.config.name}. Expected one of 'filtered_sentiment' (default)," f" 'filtered_rating', or 'unfiltered'") def _generate_examples(self, data_file): lines = Path(data_file).open(encoding="utf-8").readlines() for line_idx, line in enumerate(lines): row = json.loads(line) yield line_idx, row ``` - finally, run `datasets-cli test ./datasets/hebban-reviews/ --save_infos --all_configs` from within the topmost `datasets` directory ## Expected results Succeeding tests for three different configs. ## Actual results I printed out the values that are given to `DatasetInfo` for config name and task_templates, as you can see. There, as expected, I get `unfiltered None`. I also modified datasets/info.py and added this line [at L.170](https://github.com/huggingface/datasets/blob/f5847a304aa1b38b3a3c54a8318b4df60f1299bc/src/datasets/info.py#L170): ```python print("INTERNALLY AT INFO.PY", self.config_name, self.task_templates) ``` to my surprise, here I get `unfiltered [TextClassification(task='text-classification', text_column='review_text_without_quotes', label_column='review_sentiment')]`. So one way or another, here I suddenly see that `unfiltered` now does have a task_template -- even though that is not what is written in the data loading script, as the first print statement correctly shows. I do not quite understand how, but it seems that the config name and task_templates get mixed. This ultimately leads to the following error, but this trace may not be very useful in itself: ``` Traceback (most recent call last): File "C:\Users\bramv\.virtualenvs\hebban-U6poXNQd\Scripts\datasets-cli-script.py", line 33, in <module> sys.exit(load_entry_point('datasets', 'console_scripts', 'datasets-cli')()) File "c:\dev\python\hebban\datasets\src\datasets\commands\datasets_cli.py", line 39, in main service.run() File "c:\dev\python\hebban\datasets\src\datasets\commands\test.py", line 144, in run builder.as_dataset() File "c:\dev\python\hebban\datasets\src\datasets\builder.py", line 899, in as_dataset datasets = map_nested( File "c:\dev\python\hebban\datasets\src\datasets\utils\py_utils.py", line 393, in map_nested mapped = [ File "c:\dev\python\hebban\datasets\src\datasets\utils\py_utils.py", line 394, in <listcomp> _single_map_nested((function, obj, types, None, True, None)) File "c:\dev\python\hebban\datasets\src\datasets\utils\py_utils.py", line 330, in _single_map_nested return function(data_struct) File "c:\dev\python\hebban\datasets\src\datasets\builder.py", line 930, in _build_single_dataset ds = self._as_dataset( File "c:\dev\python\hebban\datasets\src\datasets\builder.py", line 1006, in _as_dataset return Dataset(fingerprint=fingerprint, **dataset_kwargs) File "c:\dev\python\hebban\datasets\src\datasets\arrow_dataset.py", line 661, in __init__ info = info.copy() if info is not None else DatasetInfo() File "c:\dev\python\hebban\datasets\src\datasets\info.py", line 286, in copy return self.__class__(**{k: copy.deepcopy(v) for k, v in self.__dict__.items()}) File "<string>", line 20, in __init__ File "c:\dev\python\hebban\datasets\src\datasets\info.py", line 176, in __post_init__ self.task_templates = [ File "c:\dev\python\hebban\datasets\src\datasets\info.py", line 177, in <listcomp> template.align_with_features(self.features) for template in (self.task_templates) File "c:\dev\python\hebban\datasets\src\datasets\tasks\text_classification.py", line 22, in align_with_features raise ValueError(f"Column {self.label_column} is not a ClassLabel.") ValueError: Column review_sentiment is not a ClassLabel. ``` ## Environment info - `datasets` version: 2.4.1.dev0 - Platform: Windows-10-10.0.19041-SP0 - Python version: 3.8.8 - PyArrow version: 8.0.0 - Pandas version: 1.4.3

https://api.github.com/repos/huggingface/datasets/issues/4752/timeline

https://api.github.com/repos/huggingface/datasets/issues/4751

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

https://api.github.com/repos/huggingface/datasets/issues/4751/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4751/comments

https://api.github.com/repos/huggingface/datasets/issues/4751/events

https://github.com/huggingface/datasets/pull/4751

PR_kwDODunzps48LJ7U

Added dataset information in clinic oos dataset card

closed

CONTRIBUTOR

This PR aims to add relevant information like the Description, Language and citation information of the clinic oos dataset card.

https://api.github.com/repos/huggingface/datasets/issues/4751/timeline

https://api.github.com/repos/huggingface/datasets/issues/4750

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

https://api.github.com/repos/huggingface/datasets/issues/4750/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4750/comments

https://api.github.com/repos/huggingface/datasets/issues/4750/events

https://github.com/huggingface/datasets/issues/4750

I_kwDODunzps5Oo28N

Easily create loading script for benchmark comprising multiple huggingface datasets

closed

CONTRIBUTOR

Hi, I would like to create a loading script for a benchmark comprising multiple huggingface datasets. The function _split_generators needs to return the files for the respective dataset. However, the files are not always in the same location for each dataset. I want to just make a wrapper dataset that provides a single interface to all the underlying datasets. I thought about downloading the files with the load_dataset function and then providing the link to the cached file. But this seems a bit inelegant to me. What approach would you propose to do this? Please let me know if you have any questions. Cheers, Joel

https://api.github.com/repos/huggingface/datasets/issues/4750/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4748

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

https://api.github.com/repos/huggingface/datasets/issues/4748/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4748/comments

https://api.github.com/repos/huggingface/datasets/issues/4748/events

https://github.com/huggingface/datasets/pull/4748

PR_kwDODunzps48JTEb

Add image classification processing guide

closed

MEMBER

This PR follows up on #4710 to separate the object detection and image classification guides. It expands a little more on the original guide to include a more complete example of loading and transforming a whole dataset.

https://api.github.com/repos/huggingface/datasets/issues/4748/timeline

https://api.github.com/repos/huggingface/datasets/issues/4747

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

https://api.github.com/repos/huggingface/datasets/issues/4747/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4747/comments

https://api.github.com/repos/huggingface/datasets/issues/4747/events

https://github.com/huggingface/datasets/pull/4747

PR_kwDODunzps48IWKj

Shard parquet in `download_and_prepare`

open

MEMBER

Following https://github.com/huggingface/datasets/pull/4724 (needs to be merged first) It's good practice to shard parquet files to enable parallelism with spark/dask/etc. I added the `max_shard_size` parameter to `download_and_prepare` (default to 500MB for parquet, and None for arrow). ```python from datasets import * cache_dir = "s3://..." builder = load_dataset_builder("squad", cache_dir=cache_dir) builder.download_and_prepare(file_format="parquet", max_shard_size="5MB") ``` ### Implementation details The examples are written to a parquet file until `ParquetWriter._num_bytes > max_shard_size`. When this happens, a new writer is instantiated to start writing the next shard. At the end, all the shards are renamed to include the total number of shards in their names: `{builder.name}-{split}-{shard_id:05d}-of-{num_shards:05d}.parquet` TODO: - [x] docstrings - [x] docs - [x] tests cc @severo

https://api.github.com/repos/huggingface/datasets/issues/4747/timeline

https://api.github.com/repos/huggingface/datasets/issues/4746

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

https://api.github.com/repos/huggingface/datasets/issues/4746/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4746/comments

https://api.github.com/repos/huggingface/datasets/issues/4746/events

https://github.com/huggingface/datasets/issues/4746

I_kwDODunzps5OloJH

Dataset Viewer issue for yanekyuk/wikikey

open

NONE

### Link _No response_ ### Description _No response_ ### Owner _No response_

https://api.github.com/repos/huggingface/datasets/issues/4746/timeline

https://api.github.com/repos/huggingface/datasets/issues/4745

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

https://api.github.com/repos/huggingface/datasets/issues/4745/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4745/comments

https://api.github.com/repos/huggingface/datasets/issues/4745/events

https://github.com/huggingface/datasets/issues/4745

I_kwDODunzps5Oj1aP

Allow `list_datasets` to include private datasets

open

NONE

I am working with a large collection of private datasets, it would be convenient for me to be able to list them. I would envision extending the convention of using `use_auth_token` keyword argument to `list_datasets` function, then calling: ``` list_datasets(use_auth_token="my_token") ``` would return the list of all datasets I have permissions to view, including private ones. The only current alternative I see is to use the hub website to manually obtain the list of dataset names - this is in the context of BigScience where respective private spaces contain hundreds of datasets, so not very convenient to list manually.

https://api.github.com/repos/huggingface/datasets/issues/4745/timeline

https://api.github.com/repos/huggingface/datasets/issues/4744

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

https://api.github.com/repos/huggingface/datasets/issues/4744/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4744/comments

https://api.github.com/repos/huggingface/datasets/issues/4744/events

https://github.com/huggingface/datasets/issues/4744

I_kwDODunzps5OjF-J

Remove instructions to generate dummy data from our docs

open

MEMBER

In our docs, we indicate to generate the dummy data: https://huggingface.co/docs/datasets/dataset_script#testing-data-and-checksum-metadata However: - dummy data makes sense only for datasets in our GitHub repo: so that we can test their loading with our CI - for datasets on the Hub: - they do not pass any CI test requiring dummy data - there are no instructions on how they can test their dataset locally using the dummy data - the generation of the dummy data assumes our GitHub directory structure: - the dummy data will be generated under `./datasets/<dataset_name>/dummy` even if locally there is no `./datasets` directory (which is the usual case). See issue: - #4742 CC: @stevhliu

https://api.github.com/repos/huggingface/datasets/issues/4744/timeline

https://api.github.com/repos/huggingface/datasets/issues/4743

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

https://api.github.com/repos/huggingface/datasets/issues/4743/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4743/comments

https://api.github.com/repos/huggingface/datasets/issues/4743/events

https://github.com/huggingface/datasets/pull/4743

PR_kwDODunzps48EUFs

Update map docs

closed

MEMBER

This PR updates the `map` docs for processing text to include `return_tensors="np"` to make it run faster (see #4676).

https://api.github.com/repos/huggingface/datasets/issues/4743/timeline

https://api.github.com/repos/huggingface/datasets/issues/4742

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

https://api.github.com/repos/huggingface/datasets/issues/4742/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4742/comments

https://api.github.com/repos/huggingface/datasets/issues/4742/events

https://github.com/huggingface/datasets/issues/4742

I_kwDODunzps5Og813

Dummy data nowhere to be found

open

CONTRIBUTOR

## Describe the bug To finalize my dataset, I wanted to create dummy data as per the guide and I ran ```shell datasets-cli dummy_data datasets/hebban-reviews --auto_generate ``` where hebban-reviews is [this repo](https://huggingface.co/datasets/BramVanroy/hebban-reviews). And even though the scripts runs and shows a message at the end that it succeeded, I cannot find the dummy data anywhere. Where is it? ## Expected results To see the dummy data in the datasets' folder or in the folder where I ran the command. ## Actual results I see the following message but I cannot find the dummy data anywhere. ``` Dummy data generation done and dummy data test succeeded for config 'filtered''. Automatic dummy data generation succeeded for all configs of '.\datasets\hebban-reviews\' ``` ## Environment info - `datasets` version: 2.4.1.dev0 - Platform: Windows-10-10.0.19041-SP0 - Python version: 3.8.8 - PyArrow version: 8.0.0 - Pandas version: 1.4.3

https://api.github.com/repos/huggingface/datasets/issues/4742/timeline

https://api.github.com/repos/huggingface/datasets/issues/4741

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

https://api.github.com/repos/huggingface/datasets/issues/4741/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4741/comments

https://api.github.com/repos/huggingface/datasets/issues/4741/events

https://github.com/huggingface/datasets/pull/4741

PR_kwDODunzps48B2fl

Fix to dict conversion of `DatasetInfo`/`Features`

closed

CONTRIBUTOR

Fix #4681

https://api.github.com/repos/huggingface/datasets/issues/4741/timeline

https://api.github.com/repos/huggingface/datasets/issues/4740

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

https://api.github.com/repos/huggingface/datasets/issues/4740/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4740/comments

https://api.github.com/repos/huggingface/datasets/issues/4740/events

https://github.com/huggingface/datasets/pull/4740

PR_kwDODunzps48BX5l

Fix multiprocessing in map_nested

closed

MEMBER

As previously discussed: Before, multiprocessing was not used in `map_nested` if `num_proc` was greater than or equal to `len(iterable)`. - Multiprocessing was not used e.g. when passing `num_proc=20` but having 19 files to download - As by default, `DownloadManager` sets `num_proc=16`, before multiprocessing was only used when `len(iterable)>16` by default Now, if `num_proc` is greater than or equal to ``len(iterable)``, `num_proc` is set to ``len(iterable)`` and multiprocessing is used. - We pass the variable `parallel_min_length=16`, so that multiprocessing is only used if at least 16 files to be downloaded - ~As by default, `DownloadManager` sets `num_proc=16`, now multiprocessing is used when `len(iterable)>1` by default~ See discussion below. ~After having had to fix some tests (87602ac), I am wondering:~ - ~do we want to have multiprocessing by default?~ - ~please note that `DownloadManager.download` sets `num_proc=16` by default~ - ~or would it be better to ask the user to set it explicitly if they want multiprocessing (and default to `num_proc=1`)?~ Fix #4636. CC: @nateraw

https://api.github.com/repos/huggingface/datasets/issues/4740/timeline

https://api.github.com/repos/huggingface/datasets/issues/4739

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

https://api.github.com/repos/huggingface/datasets/issues/4739/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4739/comments

https://api.github.com/repos/huggingface/datasets/issues/4739/events

https://github.com/huggingface/datasets/pull/4739

PR_kwDODunzps48BHdE

Deprecate metrics

closed

MEMBER

Deprecate metrics: - deprecate public functions: `load_metric`, `list_metrics` and `inspect_metric`: docstring and warning - test deprecation warnings are issues - deprecate metrics in all docs - remove mentions to metrics in docs and README - deprecate internal functions/classes Maybe we should also stop testing metrics?

https://api.github.com/repos/huggingface/datasets/issues/4739/timeline

https://api.github.com/repos/huggingface/datasets/issues/4738

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

https://api.github.com/repos/huggingface/datasets/issues/4738/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4738/comments

https://api.github.com/repos/huggingface/datasets/issues/4738/events

https://github.com/huggingface/datasets/pull/4738

PR_kwDODunzps479hq4

Use CI unit/integration tests

closed

MEMBER

This PR: - Implements separate unit/integration tests - A fail in integration tests does not cancel the rest of the jobs - We should implement more robust integration tests: work in progress in a subsequent PR - For the moment, test involving network requests are marked as integration: to be evolved

https://api.github.com/repos/huggingface/datasets/issues/4738/timeline

https://api.github.com/repos/huggingface/datasets/issues/4737

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

https://api.github.com/repos/huggingface/datasets/issues/4737/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4737/comments

https://api.github.com/repos/huggingface/datasets/issues/4737/events

https://github.com/huggingface/datasets/issues/4737

I_kwDODunzps5OYXm8

Download error on scene_parse_150

open

NONE

``` from datasets import load_dataset dataset = load_dataset("scene_parse_150", "scene_parsing") FileNotFoundError: Couldn't find file at http://data.csail.mit.edu/places/ADEchallenge/ADEChallengeData2016.zip ```

https://api.github.com/repos/huggingface/datasets/issues/4737/timeline

https://api.github.com/repos/huggingface/datasets/issues/4736

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

https://api.github.com/repos/huggingface/datasets/issues/4736/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4736/comments

https://api.github.com/repos/huggingface/datasets/issues/4736/events

https://github.com/huggingface/datasets/issues/4736

I_kwDODunzps5OYEUc

Dataset Viewer issue for deepklarity/huggingface-spaces-dataset

closed

NONE

### Link https://huggingface.co/datasets/deepklarity/huggingface-spaces-dataset/viewer/deepklarity--huggingface-spaces-dataset/train ### Description Hi Team, I'm getting the following error on a uploaded dataset. I'm getting the same status for a couple of hours now. The dataset size is `<1MB` and the format is csv, so I'm not sure if it's supposed to take this much time or not. ``` Status code: 400 Exception: Status400Error Message: The split is being processed. Retry later. ``` Is there any explicit step to be taken to get the viewer to work? ### Owner Yes

https://api.github.com/repos/huggingface/datasets/issues/4736/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4735

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

https://api.github.com/repos/huggingface/datasets/issues/4735/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4735/comments

https://api.github.com/repos/huggingface/datasets/issues/4735/events

https://github.com/huggingface/datasets/pull/4735

PR_kwDODunzps477CuP

Pin rouge_score test dependency

closed

MEMBER

Temporarily pin `rouge_score` (to avoid latest version 0.7.0) until the issue is fixed. Fix #4734

https://api.github.com/repos/huggingface/datasets/issues/4735/timeline

https://api.github.com/repos/huggingface/datasets/issues/4734

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

https://api.github.com/repos/huggingface/datasets/issues/4734/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4734/comments

https://api.github.com/repos/huggingface/datasets/issues/4734/events

https://github.com/huggingface/datasets/issues/4734

I_kwDODunzps5OWZuW

Package rouge-score cannot be imported

closed

MEMBER

## Describe the bug After the today release of `rouge_score-0.0.7` it seems no longer importable. Our CI fails: https://github.com/huggingface/datasets/runs/7463218591?check_suite_focus=true ``` FAILED tests/test_dataset_common.py::LocalDatasetTest::test_builder_class_bigbench FAILED tests/test_dataset_common.py::LocalDatasetTest::test_builder_configs_bigbench FAILED tests/test_dataset_common.py::LocalDatasetTest::test_load_dataset_bigbench FAILED tests/test_metric_common.py::LocalMetricTest::test_load_metric_rouge ``` with errors: ``` > from rouge_score import rouge_scorer E ModuleNotFoundError: No module named 'rouge_score' ``` ``` E ImportError: To be able to use rouge, you need to install the following dependency: rouge_score. E Please install it using 'pip install rouge_score' for instance' ```

https://api.github.com/repos/huggingface/datasets/issues/4734/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4733

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

https://api.github.com/repos/huggingface/datasets/issues/4733/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4733/comments

https://api.github.com/repos/huggingface/datasets/issues/4733/events

https://github.com/huggingface/datasets/issues/4733

I_kwDODunzps5OWV4A

rouge metric

closed

NONE

## Describe the bug A clear and concise description of what the bug is. Loading Rouge metric gives error after latest rouge-score==0.0.7 release. Downgrading rougemetric==0.0.4 works fine. ## Steps to reproduce the bug ```python # Sample code to reproduce the bug ``` ## Expected results A clear and concise description of the expected results. from rouge_score import rouge_scorer, scoring should run ## Actual results Specify the actual results or traceback. File "/root/.cache/huggingface/modules/datasets_modules/metrics/rouge/0ffdb60f436bdb8884d5e4d608d53dbe108e82dac4f494a66f80ef3f647c104f/rouge.py", line 21, in <module> from rouge_score import rouge_scorer, scoring ImportError: cannot import name 'rouge_scorer' from 'rouge_score' (unknown location) ## Environment info <!-- You can run the command `datasets-cli env` and copy-and-paste its output below. --> - `datasets` version: - Platform: Linux - Python version:3.9 - PyArrow version:

https://api.github.com/repos/huggingface/datasets/issues/4733/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4732

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

https://api.github.com/repos/huggingface/datasets/issues/4732/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4732/comments

https://api.github.com/repos/huggingface/datasets/issues/4732/events

https://github.com/huggingface/datasets/issues/4732

I_kwDODunzps5OV7fu

Document better that loading a dataset passing its name does not use the local script

open

MEMBER

As reported by @TrentBrick here https://github.com/huggingface/datasets/issues/4725#issuecomment-1191858596, it could be more clear that loading a dataset by passing its name does not use the (modified) local script of it. What he did: - he installed `datasets` from source - he modified locally `datasets/the_pile/the_pile.py` loading script - he tried to load it but using `load_dataset("the_pile")` instead of `load_dataset("datasets/the_pile")` - as explained here https://github.com/huggingface/datasets/issues/4725#issuecomment-1191040245: - the former does not use the local script, but instead it downloads a copy of `the_pile.py` from our GitHub, caches it locally (inside `~/.cache/huggingface/modules`) and uses that. He suggests adding a more clear explanation about this. He suggests adding it maybe in [Installation > source](https://huggingface.co/docs/datasets/installation)) CC: @stevhliu

https://api.github.com/repos/huggingface/datasets/issues/4732/timeline

https://api.github.com/repos/huggingface/datasets/issues/4731

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

https://api.github.com/repos/huggingface/datasets/issues/4731/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4731/comments

https://api.github.com/repos/huggingface/datasets/issues/4731/events

https://github.com/huggingface/datasets/pull/4731

PR_kwDODunzps474dlZ

docs: ✏️ fix TranslationVariableLanguages example

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/4731/timeline

https://api.github.com/repos/huggingface/datasets/issues/4730

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

https://api.github.com/repos/huggingface/datasets/issues/4730/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4730/comments

https://api.github.com/repos/huggingface/datasets/issues/4730/events

https://github.com/huggingface/datasets/issues/4730

I_kwDODunzps5OSTfP

Loading imagenet-1k validation split takes much more RAM than expected

closed

CONTRIBUTOR

## Describe the bug Loading into memory the validation split of imagenet-1k takes much more RAM than expected. Assuming ImageNet-1k is 150 GB, split is 50000 validation images and 1,281,167 train images, I would expect only about 6 GB loaded in RAM. ## Steps to reproduce the bug ```python from datasets import load_dataset dataset = load_dataset("imagenet-1k", split="validation") print(dataset) """prints Dataset({ features: ['image', 'label'], num_rows: 50000 }) """ pipe_inputs = dataset["image"] # and wait :-) ``` ## Expected results Use only < 10 GB RAM when loading the images. ## Actual results  ``` Using custom data configuration default Reusing dataset imagenet-1k (/home/fxmarty/.cache/huggingface/datasets/imagenet-1k/default/1.0.0/a1e9bfc56c3a7350165007d1176b15e9128fcaf9ab972147840529aed3ae52bc) Killed ``` ## Environment info - `datasets` version: 2.3.3.dev0 - Platform: Linux-5.15.0-41-generic-x86_64-with-glibc2.35 - Python version: 3.9.12 - PyArrow version: 7.0.0 - Pandas version: 1.3.5 - datasets commit: 4e4222f1b6362c2788aec0dd2cd8cede6dd17b80

https://api.github.com/repos/huggingface/datasets/issues/4730/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4729

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

https://api.github.com/repos/huggingface/datasets/issues/4729/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4729/comments

https://api.github.com/repos/huggingface/datasets/issues/4729/events

https://github.com/huggingface/datasets/pull/4729

PR_kwDODunzps473GmR

Refactor Hub tests

closed

MEMBER

This PR refactors `test_upstream_hub` by removing unittests and using the following pytest Hub fixtures: - `ci_hub_config` - `set_ci_hub_access_token`: to replace setUp/tearDown - `temporary_repo` context manager: to replace `try... finally` - `cleanup_repo`: to delete repo accidentally created if one of the tests fails This is a preliminary work done to manage unit/integration tests separately.

https://api.github.com/repos/huggingface/datasets/issues/4729/timeline

https://api.github.com/repos/huggingface/datasets/issues/4728

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

https://api.github.com/repos/huggingface/datasets/issues/4728/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4728/comments

https://api.github.com/repos/huggingface/datasets/issues/4728/events

https://github.com/huggingface/datasets/issues/4728

I_kwDODunzps5OQTmu

load_dataset gives "403" error when using Financial Phrasebank

open

NONE

I tried both codes below to download the financial phrasebank dataset (https://huggingface.co/datasets/financial_phrasebank) with the sentences_allagree subset. However, the code gives a 403 error when executed from multiple machines locally or on the cloud. ``` from datasets import load_dataset, DownloadMode load_dataset(path='financial_phrasebank',name='sentences_allagree',download_mode=DownloadMode.FORCE_REDOWNLOAD) ``` ``` from datasets import load_dataset, DownloadMode load_dataset(path='financial_phrasebank',name='sentences_allagree') ``` **Error** ConnectionError: Couldn't reach https://www.researchgate.net/profile/Pekka_Malo/publication/251231364_FinancialPhraseBank-v10/data/0c96051eee4fb1d56e000000/FinancialPhraseBank-v10.zip (error 403)

https://api.github.com/repos/huggingface/datasets/issues/4728/timeline

https://api.github.com/repos/huggingface/datasets/issues/4727

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

https://api.github.com/repos/huggingface/datasets/issues/4727/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4727/comments

https://api.github.com/repos/huggingface/datasets/issues/4727/events

https://github.com/huggingface/datasets/issues/4727

I_kwDODunzps5OPWEP

Dataset Viewer issue for TheNoob3131/mosquito-data

closed

NONE

### Link https://huggingface.co/datasets/TheNoob3131/mosquito-data/viewer/TheNoob3131--mosquito-data/test ### Description Dataset preview not showing with large files. Says 'split cache is empty' even though there are train and test splits. ### Owner _No response_

https://api.github.com/repos/huggingface/datasets/issues/4727/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4726

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

https://api.github.com/repos/huggingface/datasets/issues/4726/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4726/comments

https://api.github.com/repos/huggingface/datasets/issues/4726/events

https://github.com/huggingface/datasets/pull/4726

PR_kwDODunzps47ykPI

Fix broken link to the Hub

closed

MEMBER

The Markdown link fails to render if it is in the same line as the `<span>`. This PR implements @mishig25's fix by using `<a href=" ">` instead. 

https://api.github.com/repos/huggingface/datasets/issues/4726/timeline

https://api.github.com/repos/huggingface/datasets/issues/4725

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

https://api.github.com/repos/huggingface/datasets/issues/4725/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4725/comments

https://api.github.com/repos/huggingface/datasets/issues/4725/events

https://github.com/huggingface/datasets/issues/4725

I_kwDODunzps5OMh0Y

the_pile datasets URL broken.

closed

NONE

https://github.com/huggingface/datasets/pull/3627 changed the Eleuther AI Pile dataset URL from https://the-eye.eu/ to https://mystic.the-eye.eu/ but the latter is now broken and the former works again. Note that when I git clone the repo and use `pip install -e .` and then edit the URL back the codebase doesn't seem to use this edit so the mystic URL is also cached somewhere else that I can't find?

https://api.github.com/repos/huggingface/datasets/issues/4725/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4724

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

https://api.github.com/repos/huggingface/datasets/issues/4724/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4724/comments

https://api.github.com/repos/huggingface/datasets/issues/4724/events

https://github.com/huggingface/datasets/pull/4724

PR_kwDODunzps47vLrP

Download and prepare as Parquet for cloud storage

open

MEMBER

Download a dataset as Parquet in a cloud storage can be useful for streaming mode and to use with spark/dask/ray. This PR adds support for `fsspec` URIs like `s3://...`, `gcs://...` etc. and ads the `file_format` to save as parquet instead of arrow: ```python from datasets import * cache_dir = "s3://..." builder = load_dataset_builder("crime_and_punish", cache_dir=cache_dir) builder.download_and_prepare(file_format="parquet") ``` credentials to cloud storage can be passed using the `storage_options` argument in `load_dataset_builder` For consistency with the BeamBasedBuilder, I name the parquet files `{builder.name}-{split}-xxxxx-of-xxxxx.parquet`. I think this is fine since we'll need to implement parquet sharding after this PR, so that a dataset can be used efficiently with dask for example. Note that images/audio files are not embedded yet in the parquet files, this will added in a subsequent PR TODO: - [x] docs - [x] tests

https://api.github.com/repos/huggingface/datasets/issues/4724/timeline

https://api.github.com/repos/huggingface/datasets/issues/4723

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

https://api.github.com/repos/huggingface/datasets/issues/4723/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4723/comments

https://api.github.com/repos/huggingface/datasets/issues/4723/events

https://github.com/huggingface/datasets/pull/4723

PR_kwDODunzps47uoSj

Refactor conftest fixtures

closed

MEMBER

Previously, fixture modules `hub_fixtures` and `s3_fixtures`: - were both at the root test directory - were imported using `import *` - as a side effect, the modules `os` and `pytest` were imported from `s3_fixtures` into `conftest` This PR: - puts both fixture modules in a dedicated directory `fixtures` - renames both to: `fixtures.hub` and `fixtures.s3` - imports them into `conftest` as plugins, using the `pytest_plugins`: this avoids the `import *` - additionally creates a new fixture module `fixtures.files` with all file-related fixtures

https://api.github.com/repos/huggingface/datasets/issues/4723/timeline

https://api.github.com/repos/huggingface/datasets/issues/4722

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

https://api.github.com/repos/huggingface/datasets/issues/4722/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4722/comments

https://api.github.com/repos/huggingface/datasets/issues/4722/events

https://github.com/huggingface/datasets/pull/4722

PR_kwDODunzps47t_HJ

Docs: Fix same-page haslinks

closed

CONTRIBUTOR

`href="/docs/datasets/quickstart#audio"` implicitly goes to `href="/docs/datasets/{$LATEST_STABLE_VERSION}/quickstart#audio"`. Therefore, https://huggingface.co/docs/datasets/quickstart#audio #audio hashlink does not work since the new docs were not added to v2.3.2 (LATEST_STABLE_VERSION) to preserve the version, it should be just `href="#audio"`, which will implicilty go to curren_page + #audio element

https://api.github.com/repos/huggingface/datasets/issues/4722/timeline

https://api.github.com/repos/huggingface/datasets/issues/4721

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

https://api.github.com/repos/huggingface/datasets/issues/4721/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4721/comments

https://api.github.com/repos/huggingface/datasets/issues/4721/events

https://github.com/huggingface/datasets/issues/4721

I_kwDODunzps5OGOHw

PyArrow Dataset error when calling `load_dataset`

open

NONE

## Describe the bug I am fine tuning a wav2vec2 model following the script here using my own dataset: https://github.com/huggingface/transformers/blob/main/examples/pytorch/speech-recognition/run_speech_recognition_ctc.py Loading my Audio dataset from the hub which was originally generated from disk results in the following PyArrow error: ```sh File "/home/ubuntu/w2v2/run_speech_recognition_ctc.py", line 227, in main raw_datasets = load_dataset( File "/home/ubuntu/.virtualenvs/meval/lib/python3.8/site-packages/datasets/load.py", line 1679, in load_dataset builder_instance.download_and_prepare( File "/home/ubuntu/.virtualenvs/meval/lib/python3.8/site-packages/datasets/builder.py", line 704, in download_and_prepare self._download_and_prepare( File "/home/ubuntu/.virtualenvs/meval/lib/python3.8/site-packages/datasets/builder.py", line 793, in _download_and_prepare self._prepare_split(split_generator, **prepare_split_kwargs) File "/home/ubuntu/.virtualenvs/meval/lib/python3.8/site-packages/datasets/builder.py", line 1268, in _prepare_split for key, table in logging.tqdm( File "/home/ubuntu/.virtualenvs/meval/lib/python3.8/site-packages/tqdm/std.py", line 1195, in __iter__ for obj in iterable: File "/home/ubuntu/.virtualenvs/meval/lib/python3.8/site-packages/datasets/packaged_modules/parquet/parquet.py", line 68, in _generate_tables for batch_idx, record_batch in enumerate( File "pyarrow/_parquet.pyx", line 1309, in iter_batches File "pyarrow/error.pxi", line 121, in pyarrow.lib.check_status pyarrow.lib.ArrowNotImplementedError: Nested data conversions not implemented for chunked array outputs ``` ## Steps to reproduce the bug I created a dataset from a JSON lines manifest of `audio_filepath`, `text`, and `duration`. When creating the dataset, I do something like this: ```python import json from datasets import Dataset, Audio # manifest_lines is a list of dicts w/ "audio_filepath", "duration", and "text for line in manifest_lines: line = line.strip() if line: line_dict = json.loads(line) manifest_dict["audio"].append(f"{root_path}/{line_dict['audio_filepath']}") manifest_dict["duration"].append(line_dict["duration"]) manifest_dict["transcription"].append(line_dict["text"]) # Create a HF dataset dataset = Dataset.from_dict(manifest_dict).cast_column( "audio", Audio(sampling_rate=16_000), ) # From the docs for saving to disk # https://huggingface.co/docs/datasets/v2.3.2/en/package_reference/main_classes#datasets.Dataset.save_to_disk def read_audio_file(example): with open(example["audio"]["path"], "rb") as f: return {"audio": {"bytes": f.read()}} dataset = dataset.map(read_audio_file, num_proc=70) dataset.save_to_disk(f"/audio-data/hf/{artifact_name}") dataset.push_to_hub(f"{org-name}/{artifact_name}", max_shard_size="5GB", private=True) ``` Then when I call `load_dataset()` in my training script, with the same dataset I generated above, and download from the huggingface hub I get the above stack trace. I am able to load the dataset fine if I use `load_from_disk()`. ## Expected results `load_dataset()` should behave just like `load_from_disk()` and not cause any errors. ## Actual results See above ## Environment info <!-- You can run the command `datasets-cli env` and copy-and-paste its output below. --> I am using the `huggingface/transformers-pytorch-gpu:latest` image - `datasets` version: 2.3.0 - Platform: Docker/Ubuntu 20.04 - Python version: 3.8 - PyArrow version: 8.0.0

https://api.github.com/repos/huggingface/datasets/issues/4721/timeline

https://api.github.com/repos/huggingface/datasets/issues/4720

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

https://api.github.com/repos/huggingface/datasets/issues/4720/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4720/comments

https://api.github.com/repos/huggingface/datasets/issues/4720/events

https://github.com/huggingface/datasets/issues/4720

I_kwDODunzps5OFLYj

Dataset Viewer issue for shamikbose89/lancaster_newsbooks

open

NONE

### Link https://huggingface.co/datasets/shamikbose89/lancaster_newsbooks ### Description Status code: 400 Exception: ValueError Message: Cannot seek streaming HTTP file I am able to use the dataset loading script locally and it also runs when I'm using the one from the hub, but the viewer still doesn't load ### Owner Yes

https://api.github.com/repos/huggingface/datasets/issues/4720/timeline

https://api.github.com/repos/huggingface/datasets/issues/4719

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

https://api.github.com/repos/huggingface/datasets/issues/4719/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4719/comments

https://api.github.com/repos/huggingface/datasets/issues/4719/events

https://github.com/huggingface/datasets/issues/4719

I_kwDODunzps5OEssc

Issue loading TheNoob3131/mosquito-data dataset

closed

NONE

 So my dataset is public in the Huggingface Hub, but when I try to load it using the load_dataset command, it shows that it is downloading the files, but throws a ValueError. When I went to my directory to see if the files were downloaded, the folder was blank. Here is the error below: ValueError Traceback (most recent call last) Input In [8], in <cell line: 3>() 1 from datasets import load_dataset ----> 3 dataset = load_dataset("TheNoob3131/mosquito-data", split="train") File ~\Anaconda3\lib\site-packages\datasets\load.py:1679, in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, ignore_verifications, keep_in_memory, save_infos, revision, use_auth_token, task, streaming, **config_kwargs) 1676 try_from_hf_gcs = path not in _PACKAGED_DATASETS_MODULES 1678 # Download and prepare data -> 1679 builder_instance.download_and_prepare( 1680 download_config=download_config, 1681 download_mode=download_mode, 1682 ignore_verifications=ignore_verifications, 1683 try_from_hf_gcs=try_from_hf_gcs, 1684 use_auth_token=use_auth_token, 1685 ) 1687 # Build dataset for splits 1688 keep_in_memory = ( 1689 keep_in_memory if keep_in_memory is not None else is_small_dataset(builder_instance.info.dataset_size) 1690 ) Is the dataset in the wrong format or is there some security permission that I should enable?

https://api.github.com/repos/huggingface/datasets/issues/4719/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4718

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

https://api.github.com/repos/huggingface/datasets/issues/4718/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4718/comments

https://api.github.com/repos/huggingface/datasets/issues/4718/events

https://github.com/huggingface/datasets/pull/4718

PR_kwDODunzps47prWR

Make Extractor accept Path as input

closed

MEMBER

This PR: - Makes `Extractor` accept instance of `Path` as input - Removes unnecessary castings of `Path` to `str`

https://api.github.com/repos/huggingface/datasets/issues/4718/timeline

https://api.github.com/repos/huggingface/datasets/issues/4717

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

https://api.github.com/repos/huggingface/datasets/issues/4717/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4717/comments

https://api.github.com/repos/huggingface/datasets/issues/4717/events

https://github.com/huggingface/datasets/issues/4717

I_kwDODunzps5ODZMj

Dataset Viewer issue for LawalAfeez/englishreview-ds-mini

closed

NONE

### Link _No response_ ### Description Unable to view the split data ### Owner _No response_

https://api.github.com/repos/huggingface/datasets/issues/4717/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4716

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

https://api.github.com/repos/huggingface/datasets/issues/4716/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4716/comments

https://api.github.com/repos/huggingface/datasets/issues/4716/events

https://github.com/huggingface/datasets/pull/4716

PR_kwDODunzps47pdbh

Support "tags" yaml tag

closed

MEMBER

Added the "tags" YAML tag, so that users can specify data domain/topics keywords for dataset search

https://api.github.com/repos/huggingface/datasets/issues/4716/timeline

https://api.github.com/repos/huggingface/datasets/issues/4715

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

https://api.github.com/repos/huggingface/datasets/issues/4715/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4715/comments

https://api.github.com/repos/huggingface/datasets/issues/4715/events

https://github.com/huggingface/datasets/pull/4715

PR_kwDODunzps47pSui

Fix POS tags

closed

MEMBER

We're now using `part-of-speech` and not `part-of-speech-tagging`, see discussion here: https://github.com/huggingface/datasets/commit/114c09aff2fa1519597b46fbcd5a8e0c0d3ae020#r78794777

https://api.github.com/repos/huggingface/datasets/issues/4715/timeline

https://api.github.com/repos/huggingface/datasets/issues/4714

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

https://api.github.com/repos/huggingface/datasets/issues/4714/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4714/comments

https://api.github.com/repos/huggingface/datasets/issues/4714/events

https://github.com/huggingface/datasets/pull/4714

PR_kwDODunzps47o0YG

Fix named split sorting and remove unnecessary casting

closed

MEMBER

This PR: - makes `NamedSplit` sortable: so that `sorted()` can be called on them - removes unnecessary `sorted()` on `dict.keys()`: `dict_keys` view is already like a `set` - removes unnecessary casting of `NamedSplit` to `str`

https://api.github.com/repos/huggingface/datasets/issues/4714/timeline

https://api.github.com/repos/huggingface/datasets/issues/4713

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

https://api.github.com/repos/huggingface/datasets/issues/4713/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4713/comments

https://api.github.com/repos/huggingface/datasets/issues/4713/events

https://github.com/huggingface/datasets/pull/4713

PR_kwDODunzps47ojC1

Document installation of sox OS dependency for audio

closed

MEMBER

The `sox` OS package needs being installed manually using the distribution package manager. This PR adds this explanation to the docs.

https://api.github.com/repos/huggingface/datasets/issues/4713/timeline

https://api.github.com/repos/huggingface/datasets/issues/4712

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

https://api.github.com/repos/huggingface/datasets/issues/4712/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4712/comments

https://api.github.com/repos/huggingface/datasets/issues/4712/events

https://github.com/huggingface/datasets/pull/4712

PR_kwDODunzps47ohdr

Highlight non-commercial license in amazon_reviews_multi dataset card

closed

CONTRIBUTOR

Highlight that the licence granted by Amazon only covers non-commercial research use.

https://api.github.com/repos/huggingface/datasets/issues/4712/timeline

https://api.github.com/repos/huggingface/datasets/issues/4711

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

https://api.github.com/repos/huggingface/datasets/issues/4711/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4711/comments

https://api.github.com/repos/huggingface/datasets/issues/4711/events

https://github.com/huggingface/datasets/issues/4711

I_kwDODunzps5OB96K

Document how to create a dataset loading script for audio/vision

open

MEMBER

Currently, in our docs for Audio/Vision/Text, we explain how to: - Load data - Process data However we only explain how to *Create a dataset loading script* for text data. I think it would be useful that we add the same for Audio/Vision as these have some specificities different from Text. See, for example: - #4697 CC: @stevhliu

https://api.github.com/repos/huggingface/datasets/issues/4711/timeline

https://api.github.com/repos/huggingface/datasets/issues/4710

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

https://api.github.com/repos/huggingface/datasets/issues/4710/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4710/comments

https://api.github.com/repos/huggingface/datasets/issues/4710/events

https://github.com/huggingface/datasets/pull/4710

PR_kwDODunzps47ny0L

Add object detection processing tutorial

closed

CONTRIBUTOR

The following adds a quick guide on how to process object detection datasets with `albumentations`.

https://api.github.com/repos/huggingface/datasets/issues/4710/timeline

https://api.github.com/repos/huggingface/datasets/issues/4709

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

https://api.github.com/repos/huggingface/datasets/issues/4709/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4709/comments

https://api.github.com/repos/huggingface/datasets/issues/4709/events

https://github.com/huggingface/datasets/issues/4709

I_kwDODunzps5OACgF

WMT21 & WMT22

open

NONE

## Adding a Dataset - **Name:** WMT21 & WMT22 - **Description:** We are going to have three tracks: two small tasks and a large task. The small tracks evaluate translation between fairly related languages and English (all pairs). The large track uses 101 languages. - **Paper:** / - **Data:** https://statmt.org/wmt21/large-scale-multilingual-translation-task.html https://statmt.org/wmt22/large-scale-multilingual-translation-task.html - **Motivation:** Many more languages than previous WMT versions - Could be very high impact Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/main/ADD_NEW_DATASET.md). I could also tackle this. I saw the existing logic for WMT models is a bit complex (datasets are stored on the wmt account & retrieved in separate wmt datasets afaict). How long do you think it would take me? @lhoestq

https://api.github.com/repos/huggingface/datasets/issues/4709/timeline

https://api.github.com/repos/huggingface/datasets/issues/4708

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

https://api.github.com/repos/huggingface/datasets/issues/4708/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4708/comments

https://api.github.com/repos/huggingface/datasets/issues/4708/events

https://github.com/huggingface/datasets/pull/4708

PR_kwDODunzps47lewm

Fix require torchaudio and refactor test requirements

closed

MEMBER

Currently there is a bug in `require_torchaudio` (indeed it is requiring `sox` instead): ```python def require_torchaudio(test_case): if find_spec("sox") is None: ... ``` The bug was introduced by: - #3685 - Commit: https://github.com/huggingface/datasets/pull/3685/commits/b5a3e7122d49c4dcc9333b1d8d18a833fc04b940 which moved ```python require_sndfile = pytest.mark.skipif( # In Windows and OS X, soundfile installs sndfile (sys.platform != "linux" and find_spec("soundfile") is None) # In Linux, soundfile throws RuntimeError if sndfile not installed with distribution package manager or (sys.platform == "linux" and find_library("sndfile") is None), reason="Test requires 'sndfile': `pip install soundfile`; " "Linux requires sndfile installed with distribution package manager, e.g.: `sudo apt-get install libsndfile1`", ) require_sox = pytest.mark.skipif( find_library("sox") is None, reason="Test requires 'sox'; only available in non-Windows, e.g.: `sudo apt-get install sox`", ) require_torchaudio = pytest.mark.skipif(find_spec("torchaudio") is None, reason="Test requires 'torchaudio'") ``` to ```python def require_sndfile(test_case): """ Decorator marking a test that requires soundfile. These tests are skipped when soundfile isn't installed. """ if (sys.platform != "linux" and find_spec("soundfile") is None) or ( sys.platform == "linux" and find_library("sndfile") is None ): test_case = unittest.skip( "test requires 'sndfile': `pip install soundfile`; " "Linux requires sndfile installed with distribution package manager, e.g.: `sudo apt-get install libsndfile1`", )(test_case) return test_case def require_sox(test_case): """ Decorator marking a test that requires sox. These tests are skipped when sox isn't installed. """ if find_library("sox") is None: return unittest.skip("test requires 'sox'; only available in non-Windows, e.g.: `sudo apt-get install sox`")( test_case ) return test_case def require_torchaudio(test_case): """ Decorator marking a test that requires torchaudio. These tests are skipped when torchaudio isn't installed. """ if find_spec("sox") is None: return unittest.skip("test requires 'torchaudio'")(test_case) return test_case ``` This PR; - fixes the bug in `require_torchaudio` - refactors the test requirements back to using `pytest` instead of `unittest` - the text in `pytest.skipif` `reason` can be used if needed in a test body: `require_torchaudio.kwargs["reason"]`

https://api.github.com/repos/huggingface/datasets/issues/4708/timeline

https://api.github.com/repos/huggingface/datasets/issues/4707

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

https://api.github.com/repos/huggingface/datasets/issues/4707/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4707/comments

https://api.github.com/repos/huggingface/datasets/issues/4707/events

https://github.com/huggingface/datasets/issues/4707

I_kwDODunzps5N-lUN

Dataset Viewer issue for TheNoob3131/mosquito-data

closed

NONE

### Link _No response_ ### Description Getting this error when trying to view dataset preview: Message: 401, message='Unauthorized', url=URL('https://huggingface.co/datasets/TheNoob3131/mosquito-data/resolve/8aceebd6c4a359d216d10ef020868bd9e8c986dd/0_Africa_train.csv') ### Owner _No response_

https://api.github.com/repos/huggingface/datasets/issues/4707/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/4706

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

https://api.github.com/repos/huggingface/datasets/issues/4706/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4706/comments

https://api.github.com/repos/huggingface/datasets/issues/4706/events

https://github.com/huggingface/datasets/pull/4706

PR_kwDODunzps47lNBg

Fix empty examples in xtreme dataset for bucc18 config

closed

MEMBER

As reported in https://huggingface.co/muibk, there are empty examples in xtreme/bucc18.de I applied your fix @mustaszewski I also used a dict to make the dataset generation much faster

https://api.github.com/repos/huggingface/datasets/issues/4706/timeline

https://api.github.com/repos/huggingface/datasets/issues/4705

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

https://api.github.com/repos/huggingface/datasets/issues/4705/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4705/comments

https://api.github.com/repos/huggingface/datasets/issues/4705/events

https://github.com/huggingface/datasets/pull/4705

PR_kwDODunzps47lFDo

Fix crd3

closed

MEMBER

As reported in https://huggingface.co/datasets/crd3/discussions/1#62cc377073b2512b81662794, each split of the dataset was containing the same data. This issues comes from a bug in the dataset script I fixed it and also uploaded the data to hf.co to make the dataset work in streaming mode

https://api.github.com/repos/huggingface/datasets/issues/4705/timeline

https://api.github.com/repos/huggingface/datasets/issues/4704

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

https://api.github.com/repos/huggingface/datasets/issues/4704/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4704/comments

https://api.github.com/repos/huggingface/datasets/issues/4704/events

https://github.com/huggingface/datasets/pull/4704

PR_kwDODunzps47lCFt

Skip tests only for lz4/zstd params if not installed

closed

MEMBER

Currently, if `zstandard` or `lz4` are not installed, `test_compression_filesystems` and `test_streaming_dl_manager_extract_all_supported_single_file_compression_types` are skipped for all compression format parameters. This PR fixes these tests, so that if `zstandard` or `lz4` are not installed, the tests are skipped only for the corresponding compression parameters (`zstd` or `lz4`), whereas the tests are not skipped for all the other compression parameters (`gzip`, `xz` and `bz2`). Related to: - #4688

https://api.github.com/repos/huggingface/datasets/issues/4704/timeline

https://api.github.com/repos/huggingface/datasets/issues/4703

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

https://api.github.com/repos/huggingface/datasets/issues/4703/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4703/comments

https://api.github.com/repos/huggingface/datasets/issues/4703/events

https://github.com/huggingface/datasets/pull/4703

PR_kwDODunzps47kABf

Make cast in `from_pandas` more robust

closed

CONTRIBUTOR

Make the cast in `from_pandas` more robust (as it was done for the packaged modules in https://github.com/huggingface/datasets/pull/4364) This should be useful in situations like [this one](https://discuss.huggingface.co/t/loading-custom-audio-dataset-and-fine-tuning-model/8836/4).

https://api.github.com/repos/huggingface/datasets/issues/4703/timeline

https://api.github.com/repos/huggingface/datasets/issues/4702

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

https://api.github.com/repos/huggingface/datasets/issues/4702/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4702/comments

https://api.github.com/repos/huggingface/datasets/issues/4702/events

https://github.com/huggingface/datasets/issues/4702

I_kwDODunzps5N81mT

Domain specific dataset discovery on the Hugging Face hub

open

CONTRIBUTOR

**Is your feature request related to a problem? Please describe.** ## The problem The datasets hub currently has `8,239` datasets. These datasets span a wide range of different modalities and tasks (currently with a bias towards textual data). There are various ways of identifying datasets that may be relevant for a particular use case: - searching - various filters Currently, however, there isn't an easy way to identify datasets belonging to a specific domain. For example, I want to browse machine learning datasets related to 'social science' or 'climate change research'. The ability to identify datasets relating to a specific domain has come up in discussions around the [BigLA](https://github.com/bigscience-workshop/lam/) datasets hackathon https://github.com/bigscience-workshop/lam/discussions/31#discussioncomment-3123610. As part of the hackathon, we're currently collecting datasets related to Libraries, Archives and Museums and making them available via the hub. We currently do this under a Hugging Face organization (https://huggingface.co/biglam). However, going forward, I can see some of these datasets being migrated to sit under an organization that is the custodian of the dataset (for example, a national library the data was originally from). At this point, it becomes more difficult to quickly identify datasets from this domain without relying on search. This is also related to some existing issues on Github related to metadata on the hub: - https://github.com/huggingface/datasets/issues/3625 - https://github.com/huggingface/datasets/issues/3877 **Describe the solution you'd like** ### Some possible solutions that may help with this: #### Enable domain tags (from a controlled vocabulary) - This would add metadata field to the YAML for the domain a dataset relates to - Advantages: - the list is controlled, allowing it to be more easily integrated into the datasets tag app (https://huggingface.co/space/huggingface/datasets-tagging) - the controlled vocabulary could align with an existing controlled vocabulary - this additional metadata can be used to perform filtering by domain - disadvantages - choosing the best controlled vocab may be difficult - there are many datasets that are likely to fit into the 'machine learning' domain (i.e. there is a long tail of datasets that aren't in more 'generic' machine learning domain #### Enable topic tags (user-generated) Enable 'free form' topic tags for datasets and models. This would be closer to GitHub's repository topics which can be chosen from a controlled list (https://github.com/topics/) but can also be more user/org specific. This could potentially be useful for organizations to also manage their own models and datasets as the number they hold in their org grows. For example, they may create 'topic tags' for a specific project, so it's clearer which datasets /models are related to that project. #### Collections This solution would likely be the biggest shift and may require significant changes in the hub fronted. Collections could work in several different ways but would include: Users can curate particular datasets, models, spaces, etc., into a collection. For example, they may create a collection of 'historic newspapers suitable for training language models'. These collections would not be mutually exclusive, i.e. a dataset can belong to zero, one or many collections. Collections can also potentially be nested under other collections. This is fairly common on other data reposotiores for example the following collections: <img width="293" alt="Screenshot 2022-07-18 at 11 50 44" src="https://user-images.githubusercontent.com/8995957/179496445-963ed122-5e26-4574-96e8-41081bce3e2b.png"> all belong under a higher level collection (https://bl.iro.bl.uk/collections/353c908d-b495-4413-b047-87236d2573e3?locale=en). There are different models one could use for how these collections could be created: - only within an org - for any dataset/model - the owner or a dataset/model has to agree to be added to a collection - a collection owner can have people suggest additions to their collection - other models.... These collections could be thematic, related to particular training approaches, curate models with particular inference properties etc. Whilst some of these features may duplicate current/or future tag filters on the hub, they offer the advantage of being flexible and not having to predict what users will want to do upfront. There is also potential for automating the creation of these collections based on existing metadata. For example, one could collect models trained on a collection of datasets so for example, if we had a collection of 'historic newspapers suitable for training language models' that contained 30 datasets, we could create another collection 'historic newspaper language models' that takes any model on the hub whose metadata says it used one or more of those 30 datasets. There is also the option of exploring ML approaches to suggest models/datasets may be relevant to a particular collection. This approach is likely to be quite difficult to implement well and would require significant thought. There is also likely to be a benefit in doing quite a bit of upfront work in curating useful collections to demonstrate the benefits of collections. **Describe alternatives you've considered** A clear and concise description of any alternative solutions or features you've considered. It is possible to collate this information externally, i.e. one could link back to the relevant models/datasets from an external platform. **Additional context** Add any other context about the feature request here. I'm cc'ing others involved in the BigLAM hackathon who may also have thoughts @cakiki @clancyoftheoverflow @albertvillanova

https://api.github.com/repos/huggingface/datasets/issues/4702/timeline

https://api.github.com/repos/huggingface/datasets/issues/4701

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

https://api.github.com/repos/huggingface/datasets/issues/4701/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/4701/comments

https://api.github.com/repos/huggingface/datasets/issues/4701/events

https://github.com/huggingface/datasets/pull/4701

PR_kwDODunzps47jeE9

Added more information in the README about contributors of the Arabic Speech Corpus

closed

CONTRIBUTOR