html_url stringlengths 48 51 | title stringlengths 1 290 | comments listlengths 0 30 | body stringlengths 0 228k ⌀ | number int64 2 7.08k |
|---|---|---|---|---|
https://github.com/huggingface/datasets/issues/2615 | Jsonlines export error | [
"Thanks for reporting @TevenLeScao! I'm having a look...",
"(not sure what just happened on the assignations sorry)",
"For some reason this happens (both `datasets` version are on master) only on Python 3.6 and not Python 3.8.",
"@TevenLeScao we are using `pandas` to serialize the dataset to JSON Lines. So it... | ## Describe the bug
When exporting large datasets in jsonlines (c4 in my case) the created file has an error every 9999 lines: the 9999th and 10000th are concatenated, thus breaking the jsonlines format. This sounds like it is related to batching, which is by 10000 by default
## Steps to reproduce the bug
This what I'm running:
in python:
```
from datasets import load_dataset
ptb = load_dataset("ptb_text_only")
ptb["train"].to_json("ptb.jsonl")
```
then out of python:
```
head -10000 ptb.jsonl
```
## Expected results
Properly separated lines
## Actual results
The last line is a concatenation of two lines
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.9.1.dev0
- Platform: Linux-5.4.0-1046-gcp-x86_64-with-Ubuntu-18.04-bionic
- Python version: 3.6.9
- PyArrow version: 4.0.1 | 2,615 |
https://github.com/huggingface/datasets/issues/2607 | Streaming local gzip compressed JSON line files is not working | [
"Updating to pyarrow-4.0.1 didn't fix the issue",
"Here is an exemple dataset with 2 of these compressed JSON files: https://huggingface.co/datasets/thomwolf/github-python",
"Hi @thomwolf, thanks for reporting.\r\n\r\nIt seems this might be due to the fact that the JSON Dataset builder uses `pyarrow.json` (`paj... | ## Describe the bug
Using streaming to iterate on local gzip compressed JSON files raise a file not exist error
## Steps to reproduce the bug
```python
from datasets import load_dataset
streamed_dataset = load_dataset('json', split='train', data_files=data_files, streaming=True)
next(iter(streamed_dataset))
```
## Actual results
```
FileNotFoundError Traceback (most recent call last)
<ipython-input-6-27a664e29784> in <module>
----> 1 next(iter(streamed_dataset))
~/Documents/GitHub/datasets/src/datasets/iterable_dataset.py in __iter__(self)
336
337 def __iter__(self):
--> 338 for key, example in self._iter():
339 if self.features:
340 # we encode the example for ClassLabel feature types for example
~/Documents/GitHub/datasets/src/datasets/iterable_dataset.py in _iter(self)
333 else:
334 ex_iterable = self._ex_iterable
--> 335 yield from ex_iterable
336
337 def __iter__(self):
~/Documents/GitHub/datasets/src/datasets/iterable_dataset.py in __iter__(self)
76
77 def __iter__(self):
---> 78 for key, example in self.generate_examples_fn(**self.kwargs):
79 yield key, example
80
~/Documents/GitHub/datasets/src/datasets/iterable_dataset.py in wrapper(**kwargs)
282 def wrapper(**kwargs):
283 python_formatter = PythonFormatter()
--> 284 for key, table in generate_tables_fn(**kwargs):
285 batch = python_formatter.format_batch(table)
286 for i, example in enumerate(_batch_to_examples(batch)):
~/Documents/GitHub/datasets/src/datasets/packaged_modules/json/json.py in _generate_tables(self, files, original_files)
85 file,
86 read_options=self.config.pa_read_options,
---> 87 parse_options=self.config.pa_parse_options,
88 )
89 except pa.ArrowInvalid as err:
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/_json.pyx in pyarrow._json.read_json()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/_json.pyx in pyarrow._json._get_reader()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/io.pxi in pyarrow.lib.get_input_stream()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/io.pxi in pyarrow.lib.get_native_file()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/io.pxi in pyarrow.lib.OSFile.__cinit__()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/io.pxi in pyarrow.lib.OSFile._open_readable()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/error.pxi in pyarrow.lib.pyarrow_internal_check_status()
~/miniconda2/envs/datasets/lib/python3.7/site-packages/pyarrow/error.pxi in pyarrow.lib.check_status()
FileNotFoundError: [Errno 2] Failed to open local file 'gzip://file-000000000000.json::/Users/thomwolf/github-dataset/file-000000000000.json.gz'. Detail: [errno 2] No such file or directory
```
## Environment info
- `datasets` version: 1.9.1.dev0
- Platform: Darwin-19.6.0-x86_64-i386-64bit
- Python version: 3.7.7
- PyArrow version: 1.0.0 | 2,607 |
https://github.com/huggingface/datasets/issues/2606 | [Metrics] addition of wiki_split metrics | [
"#take"
] | **Is your feature request related to a problem? Please describe.**
While training the model on sentence split the task in English we require to evaluate the trained model on `Exact Match`, `SARI` and `BLEU` score
like this
While training we require metrics which can give all the output
Currently, we don't have an exact match for text normalized data
**Describe the solution you'd like**
A custom metrics for wiki_split that can calculate these three values and provide it in the form of a single dictionary
For exact match, we can refer to [this](https://github.com/huggingface/transformers/blob/master/src/transformers/data/metrics/squad_metrics.py)
**Describe alternatives you've considered**
Two metrics are already present one more can be added for an exact match then we can run all three metrics in training script
#self-assign | 2,606 |
https://github.com/huggingface/datasets/issues/2604 | Add option to delete temporary files (e.g. extracted files) when loading dataset | [
"Hi !\r\nIf we want something more general, we could either\r\n1. delete the extracted files after the arrow data generation automatically, or \r\n2. delete each extracted file during the arrow generation right after it has been closed.\r\n\r\nSolution 2 is better to save disk space during the arrow generation. Is ... | I'm loading a dataset constituted of 44 GB of compressed JSON files.
When loading the dataset with the JSON script, extracting the files create about 200 GB of uncompressed files before creating the 180GB of arrow cache tables
Having a simple way to delete the extracted files after usage (or even better, to stream extraction/delete) would be nice to avoid disk cluter.
I can maybe tackle this one in the JSON script unless you want a more general solution. | 2,604 |
https://github.com/huggingface/datasets/issues/2600 | Crash when using multiprocessing (`num_proc` > 1) on `filter` and all samples are discarded | [] | ## Describe the bug
If `filter` is applied to a dataset using multiprocessing (`num_proc` > 1) and all sharded datasets are empty afterwards (due to all samples being discarded), the program crashes.
## Steps to reproduce the bug
```python
from datasets import Dataset
data = Dataset.from_dict({'id': [0,1]})
data.filter(lambda x: False, num_proc=2)
```
## Expected results
An empty table should be returned without crashing.
## Actual results
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/user/venv/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 185, in wrapper
out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
File "/home/user/venv/lib/python3.8/site-packages/datasets/fingerprint.py", line 397, in wrapper
out = func(self, *args, **kwargs)
File "/home/user/venv/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 2143, in filter
return self.map(
File "/home/user/venv/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 1738, in map
result = concatenate_datasets(transformed_shards)
File "/home/user/venv/lib/python3.8/site-packages/datasets/arrow_dataset.py", line 3267, in concatenate_datasets
table = concat_tables(tables_to_concat, axis=axis)
File "/home/user/venv/lib/python3.8/site-packages/datasets/table.py", line 853, in concat_tables
return ConcatenationTable.from_tables(tables, axis=axis)
File "/home/user/venv/lib/python3.8/site-packages/datasets/table.py", line 713, in from_tables
blocks = to_blocks(tables[0])
IndexError: list index out of range
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.9.0
- Platform: Linux-5.12.11-300.fc34.x86_64-x86_64-with-glibc2.2.5
- Python version: 3.8.10
- PyArrow version: 3.0.0
| 2,600 |
https://github.com/huggingface/datasets/issues/2598 | Unable to download omp dataset | [
"Hi @erikadistefano , thanks for reporting the issue.\r\n\r\nI have created a Pull Request that should fix it. \r\n\r\nOnce merged into master, feel free to update your installed `datasets` library (either by installing it from our GitHub master branch or waiting until our next release) to be able to load omp datas... | ## Describe the bug
The omp dataset cannot be downloaded because of a DuplicatedKeysError
## Steps to reproduce the bug
from datasets import load_dataset
omp = load_dataset('omp', 'posts_labeled')
print(omp)
## Expected results
This code should download the omp dataset and print the dictionary
## Actual results
Downloading and preparing dataset omp/posts_labeled (download: 1.27 MiB, generated: 13.31 MiB, post-processed: Unknown size, total: 14.58 MiB) to /home/erika_distefano/.cache/huggingface/datasets/omp/posts_labeled/1.1.0/2fe5b067be3bff1d4588d5b0cbb9b5b22ae1b9d5b026a8ff572cd389f862735b...
0 examples [00:00, ? examples/s]2021-07-06 09:43:55.868815: I tensorflow/stream_executor/platform/default/dso_loader.cc:48] Successfully opened dynamic library libcudart.so.11.0
Traceback (most recent call last):
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/builder.py", line 990, in _prepare_split
writer.write(example, key)
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/arrow_writer.py", line 338, in write
self.check_duplicate_keys()
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/arrow_writer.py", line 349, in check_duplicate_keys
raise DuplicatedKeysError(key)
datasets.keyhash.DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 3326
Keys should be unique and deterministic in nature
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "hf_datasets.py", line 32, in <module>
omp = load_dataset('omp', 'posts_labeled')
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/load.py", line 748, in load_dataset
use_auth_token=use_auth_token,
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/builder.py", line 575, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/builder.py", line 652, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/builder.py", line 992, in _prepare_split
num_examples, num_bytes = writer.finalize()
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/arrow_writer.py", line 409, in finalize
self.check_duplicate_keys()
File "/home/erika_distefano/.local/lib/python3.6/site-packages/datasets/arrow_writer.py", line 349, in check_duplicate_keys
raise DuplicatedKeysError(key)
datasets.keyhash.DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 3326
Keys should be unique and deterministic in nature
## Environment info
- `datasets` version: 1.8.0
- Platform: Ubuntu 18.04.4 LTS
- Python version: 3.6.9
- PyArrow version: 3.0.0
| 2,598 |
https://github.com/huggingface/datasets/issues/2596 | Transformer Class on dataset | [
"Hi ! Do you have an example in mind that shows how this could be useful ?",
"Example:\n\nMerge 2 datasets into one datasets\n\nLabel extraction from dataset\n\ndataset(text, label)\n —> dataset(text, newlabel)\n\nTextCleaning.\n\n\nFor image dataset, \nTransformation are easier (ie linear algebra).\n\n\n\n\n\n... | Just wondering if you have intenttion to create
TransformerClass :
dataset --> dataset
and make determnistic transformation (ie not fit).
| 2,596 |
https://github.com/huggingface/datasets/issues/2595 | ModuleNotFoundError: No module named 'datasets.tasks' while importing common voice datasets | [
"Hi @profsatwinder.\r\n\r\nIt looks like you are using an old version of `datasets`. Please update it with `pip install -U datasets` and indicate if the problem persists.",
"@albertvillanova Thanks for the information. I updated it to 1.9.0 and the issue is resolved. Thanks again. "
] | Error traceback:
---------------------------------------------------------------------------
ModuleNotFoundError Traceback (most recent call last)
<ipython-input-8-a7b592d3bca0> in <module>()
1 from datasets import load_dataset, load_metric
2
----> 3 common_voice_train = load_dataset("common_voice", "pa-IN", split="train+validation")
4 common_voice_test = load_dataset("common_voice", "pa-IN", split="test")
9 frames
/root/.cache/huggingface/modules/datasets_modules/datasets/common_voice/078d412587e9efeb0ae2e574da99c31e18844c496008d53dc5c60f4159ed639b/common_voice.py in <module>()
19
20 import datasets
---> 21 from datasets.tasks import AutomaticSpeechRecognition
22
23
ModuleNotFoundError: No module named 'datasets.tasks' | 2,595 |
https://github.com/huggingface/datasets/issues/2591 | Cached dataset overflowing disk space | [
"Hi! I'm transferring this issue over to `datasets`",
"I'm using the datasets concatenate dataset to combine the datasets and then train.\r\ntrain_dataset = concatenate_datasets([dataset1, dataset2, common_voice_train])\r\n\r\n",
"Hi @BirgerMoell.\r\n\r\nYou have several options:\r\n- to set caching to be store... | I'm training a Swedish Wav2vec2 model on a Linux GPU and having issues that the huggingface cached dataset folder is completely filling up my disk space (I'm training on a dataset of around 500 gb).
The cache folder is 500gb (and now my disk space is full).
Is there a way to toggle caching or set the caching to be stored on a different device (I have another drive with 4 tb that could hold the caching files).
This might not technically be a bug, but I was unsure and I felt that the bug was the closest one.
Traceback (most recent call last):
File "/home/birger/miniconda3/envs/wav2vec2/lib/python3.7/site-packages/multiprocess/pool.py", line 121, in worker
result = (True, func(*args, **kwds))
File "/home/birger/miniconda3/envs/wav2vec2/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 186, in wrapper
out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
File "/home/birger/miniconda3/envs/wav2vec2/lib/python3.7/site-packages/datasets/fingerprint.py", line 397, in wrapper
out = func(self, *args, **kwargs)
File "/home/birger/miniconda3/envs/wav2vec2/lib/python3.7/site-packages/datasets/arrow_dataset.py", line 1983, in _map_single
writer.finalize()
File "/home/birger/miniconda3/envs/wav2vec2/lib/python3.7/site-packages/datasets/arrow_writer.py", line 418, in finalize
self.pa_writer.close()
File "pyarrow/ipc.pxi", line 402, in pyarrow.lib._CRecordBatchWriter.close
File "pyarrow/error.pxi", line 97, in pyarrow.lib.check_status
OSError: [Errno 28] Error writing bytes to file. Detail: [errno 28] No space left on device
"""
The above exception was the direct cause of the following exception:
| 2,591 |
https://github.com/huggingface/datasets/issues/2585 | sqaud_v2 dataset contains misalignment between the answer text and the context value at the answer index | [
"Hi @mmajurski, thanks for reporting this issue.\r\n\r\nIndeed this misalignment arises because the source dataset context field contains leading blank spaces (and these are counted within the answer_start), while our datasets loading script removes these leading blank spaces.\r\n\r\nI'm going to fix our script so ... | ## Describe the bug
The built in huggingface squad_v2 dataset that you can access via datasets.load_dataset contains mis-alignment between the answers['text'] and the characters in the context at the location specified by answers['answer_start'].
For example:
id = '56d1f453e7d4791d009025bd'
answers = {'text': ['Pure Land'], 'answer_start': [146]}
However the actual text in context at location 146 is 'ure Land,'
Which is an off-by-one error from the correct answer.
## Steps to reproduce the bug
```python
import datasets
def check_context_answer_alignment(example):
for a_idx in range(len(example['answers']['text'])):
# check raw dataset for answer consistency between context and answer
answer_text = example['answers']['text'][a_idx]
a_st_idx = example['answers']['answer_start'][a_idx]
a_end_idx = a_st_idx + len(example['answers']['text'][a_idx])
answer_text_from_context = example['context'][a_st_idx:a_end_idx]
if answer_text != answer_text_from_context:
#print(example['id'])
return False
return True
dataset = datasets.load_dataset('squad_v2', split='train', keep_in_memory=True)
start_len = len(dataset)
dataset = dataset.filter(check_context_answer_alignment,
num_proc=1,
keep_in_memory=True)
end_len = len(dataset)
print('{} instances contain mis-alignment between the answer text and answer index.'.format(start_len - end_len))
```
## Expected results
This code should result in 0 rows being filtered out from the dataset.
## Actual results
This filter command results in 258 rows being flagged as containing a discrepancy between the text contained within answers['text'] and the text in example['context'] at the answers['answer_start'] location.
This code will reproduce the problem and produce the following count:
"258 instances contain mis-alignment between the answer text and answer index."
## Environment info
Steps to rebuilt the Conda environment:
```
# create a virtual environment to stuff all these packages into
conda create -n round8 python=3.8 -y
# activate the virtual environment
conda activate round8
# install pytorch (best done through conda to handle cuda dependencies)
conda install pytorch torchvision torchtext cudatoolkit=11.1 -c pytorch-lts -c nvidia
pip install jsonpickle transformers datasets matplotlib
```
OS: Ubuntu 20.04
Python 3.8
Result of `conda env export`:
```
name: round8
channels:
- pytorch-lts
- nvidia
- defaults
dependencies:
- _libgcc_mutex=0.1=main
- _openmp_mutex=4.5=1_gnu
- blas=1.0=mkl
- brotlipy=0.7.0=py38h27cfd23_1003
- bzip2=1.0.8=h7b6447c_0
- ca-certificates=2021.5.25=h06a4308_1
- certifi=2021.5.30=py38h06a4308_0
- cffi=1.14.5=py38h261ae71_0
- chardet=4.0.0=py38h06a4308_1003
- cryptography=3.4.7=py38hd23ed53_0
- cudatoolkit=11.1.74=h6bb024c_0
- ffmpeg=4.2.2=h20bf706_0
- freetype=2.10.4=h5ab3b9f_0
- gmp=6.2.1=h2531618_2
- gnutls=3.6.15=he1e5248_0
- idna=2.10=pyhd3eb1b0_0
- intel-openmp=2021.2.0=h06a4308_610
- jpeg=9b=h024ee3a_2
- lame=3.100=h7b6447c_0
- lcms2=2.12=h3be6417_0
- ld_impl_linux-64=2.35.1=h7274673_9
- libffi=3.3=he6710b0_2
- libgcc-ng=9.3.0=h5101ec6_17
- libgomp=9.3.0=h5101ec6_17
- libidn2=2.3.1=h27cfd23_0
- libopus=1.3.1=h7b6447c_0
- libpng=1.6.37=hbc83047_0
- libstdcxx-ng=9.3.0=hd4cf53a_17
- libtasn1=4.16.0=h27cfd23_0
- libtiff=4.2.0=h85742a9_0
- libunistring=0.9.10=h27cfd23_0
- libuv=1.40.0=h7b6447c_0
- libvpx=1.7.0=h439df22_0
- libwebp-base=1.2.0=h27cfd23_0
- lz4-c=1.9.3=h2531618_0
- mkl=2021.2.0=h06a4308_296
- mkl-service=2.3.0=py38h27cfd23_1
- mkl_fft=1.3.0=py38h42c9631_2
- mkl_random=1.2.1=py38ha9443f7_2
- ncurses=6.2=he6710b0_1
- nettle=3.7.3=hbbd107a_1
- ninja=1.10.2=hff7bd54_1
- numpy=1.20.2=py38h2d18471_0
- numpy-base=1.20.2=py38hfae3a4d_0
- olefile=0.46=py_0
- openh264=2.1.0=hd408876_0
- openssl=1.1.1k=h27cfd23_0
- pillow=8.2.0=py38he98fc37_0
- pip=21.1.2=py38h06a4308_0
- pycparser=2.20=py_2
- pyopenssl=20.0.1=pyhd3eb1b0_1
- pysocks=1.7.1=py38h06a4308_0
- python=3.8.10=h12debd9_8
- pytorch=1.8.1=py3.8_cuda11.1_cudnn8.0.5_0
- readline=8.1=h27cfd23_0
- requests=2.25.1=pyhd3eb1b0_0
- setuptools=52.0.0=py38h06a4308_0
- six=1.16.0=pyhd3eb1b0_0
- sqlite=3.35.4=hdfb4753_0
- tk=8.6.10=hbc83047_0
- torchtext=0.9.1=py38
- torchvision=0.9.1=py38_cu111
- typing_extensions=3.7.4.3=pyha847dfd_0
- urllib3=1.26.4=pyhd3eb1b0_0
- wheel=0.36.2=pyhd3eb1b0_0
- x264=1!157.20191217=h7b6447c_0
- xz=5.2.5=h7b6447c_0
- zlib=1.2.11=h7b6447c_3
- zstd=1.4.9=haebb681_0
- pip:
- click==8.0.1
- cycler==0.10.0
- datasets==1.8.0
- dill==0.3.4
- filelock==3.0.12
- fsspec==2021.6.0
- huggingface-hub==0.0.8
- joblib==1.0.1
- jsonpickle==2.0.0
- kiwisolver==1.3.1
- matplotlib==3.4.2
- multiprocess==0.70.12.2
- packaging==20.9
- pandas==1.2.4
- pyarrow==3.0.0
- pyparsing==2.4.7
- python-dateutil==2.8.1
- pytz==2021.1
- regex==2021.4.4
- sacremoses==0.0.45
- tokenizers==0.10.3
- tqdm==4.49.0
- transformers==4.6.1
- xxhash==2.0.2
prefix: /home/mmajurski/anaconda3/envs/round8
```
| 2,585 |
https://github.com/huggingface/datasets/issues/2583 | Error iteration over IterableDataset using Torch DataLoader | [
"Hi ! This is because you first need to format the dataset for pytorch:\r\n\r\n```python\r\n>>> import torch\r\n>>> from datasets import load_dataset\r\n>>> dataset = load_dataset('oscar', \"unshuffled_deduplicated_en\", split='train', streaming=True)\r\n>>> torch_iterable_dataset = dataset.with_format(\"torch\")\r... | ## Describe the bug
I have an IterableDataset (created using streaming=True) and I am trying to create batches using Torch DataLoader class by passing this IterableDataset to it. This throws error which is pasted below. I can do the same by using Torch IterableDataset. One thing I noticed is that in the former case when I look at the dataloader.sampler class I get torch.utils.data.sampler.SequentialSampler while the latter one gives torch.utils.data.dataloader._InfiniteConstantSampler.
I am not sure if this is how it is meant to be used, but that's what seemed reasonable to me.
## Steps to reproduce the bug
1. Does not work.
```python
>>> from datasets import load_dataset
>>> dataset = load_dataset('oscar', "unshuffled_deduplicated_en", split='train', streaming=True)
>>> dataloader = torch.utils.data.DataLoader(dataset, batch_size=4)
>>> dataloader.sampler
<torch.utils.data.sampler.SequentialSampler object at 0x7f245a510208>
>>> for batch in dataloader:
... print(batch)
```
2. Works.
```python
import torch
from torch.utils.data import Dataset, IterableDataset, DataLoader
class CustomIterableDataset(IterableDataset):
'Characterizes a dataset for PyTorch'
def __init__(self, data):
'Initialization'
self.data = data
def __iter__(self):
return iter(self.data)
data = list(range(12))
dataset = CustomIterableDataset(data)
dataloader = DataLoader(dataset, batch_size=4)
print("dataloader: ", dataloader.sampler)
for batch in dataloader:
print(batch)
```
## Expected results
To get batches of data with the batch size as 4. Output from the latter one (2) though Datasource is different here so actual data is different.
dataloader: <torch.utils.data.dataloader._InfiniteConstantSampler object at 0x7f1cc29e2c50>
tensor([0, 1, 2, 3])
tensor([4, 5, 6, 7])
tensor([ 8, 9, 10, 11])
## Actual results
<torch.utils.data.sampler.SequentialSampler object at 0x7f245a510208>
...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/data/leshekha/lib/HFDatasets/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 435, in __next__
data = self._next_data()
File "/data/leshekha/lib/HFDatasets/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 474, in _next_data
index = self._next_index() # may raise StopIteration
File "/data/leshekha/lib/HFDatasets/lib/python3.6/site-packages/torch/utils/data/dataloader.py", line 427, in _next_index
return next(self._sampler_iter) # may raise StopIteration
File "/data/leshekha/lib/HFDatasets/lib/python3.6/site-packages/torch/utils/data/sampler.py", line 227, in __iter__
for idx in self.sampler:
File "/data/leshekha/lib/HFDatasets/lib/python3.6/site-packages/torch/utils/data/sampler.py", line 67, in __iter__
return iter(range(len(self.data_source)))
TypeError: object of type 'IterableDataset' has no len()
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: '1.8.1.dev0'
- Platform: Linux
- Python version: Python 3.6.8
- PyArrow version: '3.0.0'
| 2,583 |
https://github.com/huggingface/datasets/issues/2573 | Finding right block-size with JSON loading difficult for user | [
"This was actually a second error arising from a too small block-size in the json reader.\r\n\r\nFinding the right block size is difficult for the layman user"
] | As reported by @thomwolf, while loading a JSON Lines file with "json" loading script, he gets
> json.decoder.JSONDecodeError: Extra data: line 2 column 1 (char 383)
| 2,573 |
https://github.com/huggingface/datasets/issues/2572 | Support Zstandard compressed files | [
"I am trying to load a dataset using Hugging Face Datasets load_dataset method. I am getting the value error as show below. Can someone help with this? I am using Windows laptop and Google Colab notebook.\r\n\r\n```\r\n!pip install zstandard\r\nfrom datasets import load_dataset\r\n\r\nlds = load_dataset(\r\n \"j... | Add support for Zstandard compressed files: https://facebook.github.io/zstd/ | 2,572 |
https://github.com/huggingface/datasets/issues/2569 | Weights of model checkpoint not initialized for RobertaModel for Bertscore | [
"Hi @suzyahyah, thanks for reporting.\r\n\r\nThe message you get is indeed not an error message, but a warning coming from Hugging Face `transformers`. The complete warning message is:\r\n```\r\nSome weights of the model checkpoint at roberta-large were not used when initializing RobertaModel: ['lm_head.decoder.wei... | When applying bertscore out of the box,
```Some weights of the model checkpoint at roberta-large were not used when initializing RobertaModel: ['lm_head.decoder.weight', 'lm_head.bias', 'lm_head.dense.bias', 'lm_head.layer_norm.bias', 'lm_head.dense.weight', 'lm_head.layer_norm.weight']```
Following the typical usage from https://huggingface.co/docs/datasets/loading_metrics.html
```
from datasets import load_metric
metric = load_metric('bertscore')
# Example of typical usage
for batch in dataset:
inputs, references = batch
predictions = model(inputs)
metric.add_batch(predictions=predictions, references=references)
score = metric.compute(lang="en")
#score = metric.compute(model_type="roberta-large") # gives the same error
```
I am concerned about this because my usage shouldn't require any further fine-tuning and most people would expect to use BertScore out of the box? I realised the huggingface code is a wrapper around https://github.com/Tiiiger/bert_score, but I think this repo is anyway relying on the model code and weights from huggingface repo....
## Environment info
- `datasets` version: 1.7.0
- Platform: Linux-5.4.0-1041-aws-x86_64-with-glibc2.27
- Python version: 3.9.5
- PyArrow version: 3.0.0
| 2,569 |
https://github.com/huggingface/datasets/issues/2564 | concatenate_datasets for iterable datasets | [
"It is probably worth noting here that the [documentation](https://huggingface.co/docs/datasets/process#concatenate) is misleading (indicating that it does work for IterableDatasets):\r\n\r\n> You can also mix several datasets together by taking alternating examples from each one to create a new dataset. This is kn... | Currently `concatenate_datasets` only works for map-style `Dataset`.
It would be nice to have it work for `IterableDataset` objects as well.
It would simply chain the iterables of the iterable datasets. | 2,564 |
https://github.com/huggingface/datasets/issues/2563 | interleave_datasets for map-style datasets | [] | Currently the `interleave_datasets` functions only works for `IterableDataset`.
Let's make it work for map-style `Dataset` objects as well.
It would work the same way: either alternate between the datasets in order or randomly given probabilities specified by the user. | 2,563 |
https://github.com/huggingface/datasets/issues/2561 | Existing cache for local dataset builder file updates is ignored with `ignore_verifications=True` | [
"Hi ! I just tried to reproduce what you said:\r\n- create a local builder class\r\n- use `load_dataset`\r\n- update the builder class code\r\n- use `load_dataset` again (with or without `ignore_verifications=True`)\r\nAnd it creates a new cache, as expected.\r\n\r\nWhat modifications did you do to your builder's c... | ## Describe the bug
If i have local file defining a dataset builder class and I load it using `load_dataset` functionality, the existing cache is ignored whenever the file is update even with `ignore_verifications=True`. This slows down debugging and cache generator for very large datasets.
## Steps to reproduce the bug
- Create a local dataset builder class
- load the local builder class file using `load_dataset` and let the cache build
- update the file's content
- The cache should rebuilt.
## Expected results
With `ignore_verifications=True`, `load_dataset` should pick up existing cache.
## Actual results
Creates new cache.
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.8.0
- Platform: Linux-5.4.0-52-generic-x86_64-with-debian-bullseye-sid
- Python version: 3.7.7
- PyArrow version: 3.0.0
| 2,561 |
https://github.com/huggingface/datasets/issues/2559 | Memory usage consistently increases when processing a dataset with `.map` | [
"Hi ! Can you share the function you pass to `map` ?\r\nI know you mentioned it would be hard to share some code but this would really help to understand what happened",
"This is the same behavior as in #4883, so I'm closing this issue as a duplicate. "
] | ## Describe the bug
I have a HF dataset with image paths stored in it and I am trying to load those image paths using `.map` with `num_proc=80`. I am noticing that the memory usage consistently keeps on increasing with time. I tried using `DEFAULT_WRITER_BATCH_SIZE=10` in the builder to decrease arrow writer's batch size but that doesn't seem to help.
## Steps to reproduce the bug
Providing code as it is would be hard. I can provide a MVP if that helps.
## Expected results
Memory usage should become consistent after some time following the launch of processing.
## Actual results
Memory usage keeps on increasing.
## Environment info
- `datasets` version: 1.8.0
- Platform: Linux-5.4.0-52-generic-x86_64-with-debian-bullseye-sid
- Python version: 3.7.7
- PyArrow version: 3.0.0 | 2,559 |
https://github.com/huggingface/datasets/issues/2556 | Better DuplicateKeysError error to help the user debug the issue | [
"excuse me, my `datasets` version is `2.2.2`, but I also just see the error info like \r\n```\r\nDuplicatedKeysError: FAILURE TO GENERATE DATASET !\r\nFound duplicate Key: 0\r\nKeys should be unique and deterministic in nature\r\n```",
"Hi ! for which dataset do you have this error ?\r\n\r\nAlso note that this is... | As mentioned in https://github.com/huggingface/datasets/issues/2552 it would be nice to improve the error message when a dataset fails to build because there are duplicate example keys.
The current one is
```python
datasets.keyhash.DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 48
Keys should be unique and deterministic in nature
```
and we could have something that guides the user to debugging the issue:
```python
DuplicateKeysError: both 42th and 1337th examples have the same keys `48`.
Please fix the dataset script at <path/to/the/dataset/script>
``` | 2,556 |
https://github.com/huggingface/datasets/issues/2554 | Multilabel metrics not supported | [
"Hi @GuillemGSubies, thanks for reporting.\r\n\r\nI have made a PR to fix this issue and allow metrics to be computed also for multilabel classification problems.",
"Looks nice, thank you very much! 🚀 ",
"Sorry for reopening but I just noticed that the `_compute` method for the F1 metric is still not good enou... | When I try to use a metric like F1 macro I get the following error:
```
TypeError: int() argument must be a string, a bytes-like object or a number, not 'list'
```
There is an explicit casting here:
https://github.com/huggingface/datasets/blob/fc79f61cbbcfa0e8c68b28c0a8257f17e768a075/src/datasets/features.py#L274
And looks like this is because here
https://github.com/huggingface/datasets/blob/fc79f61cbbcfa0e8c68b28c0a8257f17e768a075/metrics/f1/f1.py#L88
the features can only be integers, so we cannot use that F1 for multilabel. Instead, if I create the following F1 (ints replaced with sequence of ints), it will work:
```python
class F1(datasets.Metric):
def _info(self):
return datasets.MetricInfo(
description=_DESCRIPTION,
citation=_CITATION,
inputs_description=_KWARGS_DESCRIPTION,
features=datasets.Features(
{
"predictions": datasets.Sequence(datasets.Value("int32")),
"references": datasets.Sequence(datasets.Value("int32")),
}
),
reference_urls=["https://scikit-learn.org/stable/modules/generated/sklearn.metrics.f1_score.html"],
)
def _compute(self, predictions, references, labels=None, pos_label=1, average="binary", sample_weight=None):
return {
"f1": f1_score(
references,
predictions,
labels=labels,
pos_label=pos_label,
average=average,
sample_weight=sample_weight,
),
}
```
| 2,554 |
https://github.com/huggingface/datasets/issues/2553 | load_dataset("web_nlg") NonMatchingChecksumError | [
"Hi ! Thanks for reporting. This is due to the WebNLG repository that got updated today.\r\nI just pushed a fix at #2558 - this shouldn't happen anymore in the future.",
"This is fixed on `master` now :)\r\nWe'll do a new release soon !"
] | Hi! It seems the WebNLG dataset gives a NonMatchingChecksumError.
## Steps to reproduce the bug
```python
from datasets import load_dataset
dataset = load_dataset('web_nlg', name="release_v3.0_en", split="dev")
```
Gives
```
NonMatchingChecksumError: Checksums didn't match for dataset source files:
['https://gitlab.com/shimorina/webnlg-dataset/-/archive/master/webnlg-dataset-master.zip']
```
## Environment info
- `datasets` version: 1.8.0
- Platform: macOS-11.3.1-x86_64-i386-64bit
- Python version: 3.9.4
- PyArrow version: 3.0.0
Also tested on Linux, with python 3.6.8 | 2,553 |
https://github.com/huggingface/datasets/issues/2552 | Keys should be unique error on code_search_net | [
"Two questions:\r\n- with `datasets-cli env` we don't have any information on the dataset script version used. Should we give access to this somehow? Either as a note in the Error message or as an argument with the name of the dataset to `datasets-cli env`?\r\n- I don't really understand why the id is duplicated in... | ## Describe the bug
Loading `code_search_net` seems not possible at the moment.
## Steps to reproduce the bug
```python
>>> load_dataset('code_search_net')
Downloading: 8.50kB [00:00, 3.09MB/s]
Downloading: 19.1kB [00:00, 10.1MB/s]
No config specified, defaulting to: code_search_net/all
Downloading and preparing dataset code_search_net/all (download: 4.77 GiB, generated: 5.99 GiB, post-processed: Unknown size, total: 10.76 GiB) to /Users/thomwolf/.cache/huggingface/datasets/code_search_net/all/1.0.0/b3e8278faf5d67da1d06981efbeac3b76a2900693bd2239bbca7a4a3b0d6e52a...
Traceback (most recent call last):
File "/Users/thomwolf/Documents/GitHub/datasets/src/datasets/builder.py", line 1067, in _prepare_split
writer.write(example, key)
File "/Users/thomwolf/Documents/GitHub/datasets/src/datasets/arrow_writer.py", line 343, in write
self.check_duplicate_keys()
File "/Users/thomwolf/Documents/GitHub/datasets/src/datasets/arrow_writer.py", line 354, in check_duplicate_keys
raise DuplicatedKeysError(key)
datasets.keyhash.DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 48
Keys should be unique and deterministic in nature
```
## Environment info
- `datasets` version: 1.8.1.dev0
- Platform: macOS-10.15.7-x86_64-i386-64bit
- Python version: 3.8.5
- PyArrow version: 2.0.0
| 2,552 |
https://github.com/huggingface/datasets/issues/2550 | Allow for incremental cumulative metric updates in a distributed setup | [] | Currently, using a metric allows for one of the following:
- Per example/batch metrics
- Cumulative metrics over the whole data
What I'd like is to have an efficient way to get cumulative metrics over the examples/batches added so far, in order to display it as part of the progress bar during training/evaluation.
Since most metrics are just an average of per-example metrics (which aren't?), an efficient calculation can be done as follows:
`((score_cumulative * n_cumulative) + (score_new * n_new)) / (n_cumulative+ n_new)`
where `n` and `score` refer to number of examples and metric score, `cumulative` refers to the cumulative metric and `new` refers to the addition of new examples.
If you don't want to add this capability in the library, a simple solution exists so users can do it themselves:
It is easy to implement for a single process setup, but in a distributed one there is no way to get the correct `n_new`.
The solution for this is to return the number of examples that was used to compute the metrics in `.compute()` by adding the following line here:
https://github.com/huggingface/datasets/blob/5a3221785311d0ce86c2785b765e86bd6997d516/src/datasets/metric.py#L402-L403
```
output["number_of_examples"] = len(predictions)
```
and also remove the log message here so it won't spam:
https://github.com/huggingface/datasets/blob/3db67f5ff6cbf807b129d2b4d1107af27623b608/src/datasets/metric.py#L411
If this change is ok with you, I'll open a pull request.
| 2,550 |
https://github.com/huggingface/datasets/issues/2549 | Handling unlabeled datasets | [
"Hi @nelson-liu,\r\n\r\nYou can pass the parameter `features` to `load_dataset`: https://huggingface.co/docs/datasets/_modules/datasets/load.html#load_dataset\r\n\r\nIf you look at the code of the MNLI script you referred in your question (https://github.com/huggingface/datasets/blob/master/datasets/multi_nli/multi... | Hi!
Is there a way for datasets to produce unlabeled instances (e.g., the `ClassLabel` can be nullable).
For example, I want to use the MNLI dataset reader ( https://github.com/huggingface/datasets/blob/master/datasets/multi_nli/multi_nli.py ) on a file that doesn't have the `gold_label` field. I tried setting `"label": data.get("gold_label")`, but got the following error:
```
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/load.py", line 748, in load_dataset
use_auth_token=use_auth_token,
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/builder.py", line 575, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/builder.py", line 652, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/builder.py", line 989, in _prepare_split
example = self.info.features.encode_example(record)
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/features.py", line 953, in encode_example
return encode_nested_example(self, example)
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/features.py", line 848, in encode_nested_example
k: encode_nested_example(sub_schema, sub_obj) for k, (sub_schema, sub_obj) in utils.zip_dict(schema, obj)
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/features.py", line 848, in <dictcomp>
k: encode_nested_example(sub_schema, sub_obj) for k, (sub_schema, sub_obj) in utils.zip_dict(schema, obj)
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/features.py", line 875, in encode_nested_example
return schema.encode_example(obj)
File "/home/nfliu/miniconda3/envs/debias/lib/python3.7/site-packages/datasets/features.py", line 653, in encode_example
if not -1 <= example_data < self.num_classes:
TypeError: '<=' not supported between instances of 'int' and 'NoneType'
```
What's the proper way to handle reading unlabeled datasets, especially for downstream usage with Transformers? | 2,549 |
https://github.com/huggingface/datasets/issues/2548 | Field order issue in loading json | [
"Hi @luyug, thanks for reporting.\r\n\r\nThe good news is that we fixed this issue only 9 days ago: #2507.\r\n\r\nThe patch is already in the master branch of our repository and it will be included in our next `datasets` release version 1.9.0.\r\n\r\nFeel free to reopen the issue if the problem persists."
] | ## Describe the bug
The `load_dataset` function expects columns in alphabetical order when loading json files.
Similar bug was previously reported for csv in #623 and fixed in #684.
## Steps to reproduce the bug
For a json file `j.json`,
```
{"c":321, "a": 1, "b": 2}
```
Running the following,
```
f= datasets.Features({'a': Value('int32'), 'b': Value('int32'), 'c': Value('int32')})
json_data = datasets.load_dataset('json', data_files='j.json', features=f)
```
## Expected results
A successful load.
## Actual results
```
File "pyarrow/table.pxi", line 1409, in pyarrow.lib.Table.cast
ValueError: Target schema's field names are not matching the table's field names: ['c', 'a', 'b'], ['a', 'b', 'c']
```
## Environment info
- `datasets` version: 1.8.0
- Platform: Linux-3.10.0-957.1.3.el7.x86_64-x86_64-with-glibc2.10
- Python version: 3.8.8
- PyArrow version: 3.0.0
| 2,548 |
https://github.com/huggingface/datasets/issues/2547 | Dataset load_from_disk is too slow | [
"Hi ! It looks like an issue with the virtual disk you are using.\r\n\r\nWe load datasets using memory mapping. In general it makes it possible to load very big files instantaneously since it doesn't have to read the file (it just assigns virtual memory to the file on disk).\r\nHowever there happens to be issues wi... | @lhoestq
## Describe the bug
It's not normal that I have to wait 7-8 hours for a dataset to be loaded from disk, as there are no preprocessing steps, it's only loading it with load_from_disk. I have 96 cpus, however only 1 is used for this, which is inefficient. Moreover, its usage is at 1%... This is happening in the context of a language model training, therefore I'm wasting 100$ each time I have to load the dataset from disk again (because the spot instance was stopped by aws and I need to relaunch it for example).
## Steps to reproduce the bug
Just get the oscar in spanish (around 150GGB) and try to first save in disk and then load the processed dataset. It's not dependent on the task you're doing, it just depends on the size of the text dataset.
## Expected results
I expect the dataset to be loaded in a normal time, by using the whole machine for loading it, I mean if you store the dataset in multiple files (.arrow) and then load it from multiple files, you can use multiprocessing for that and therefore don't waste so much time.
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.8.0
- Platform: Ubuntu 18
- Python version: 3.8
I've seen you're planning to include a streaming mode for load_dataset, but that only saves the downloading and processing time, that's not being a problem for me, you cannot save the pure loading from disk time, therefore that's not a solution for my use case or for anyone who wants to use your library for training a language model. | 2,547 |
https://github.com/huggingface/datasets/issues/2543 | switching some low-level log.info's to log.debug? | [
"Hi @stas00, thanks for pointing out this issue with logging.\r\n\r\nI agree that `datasets` can sometimes be too verbose... I can create a PR and we could discuss there the choice of the log levels for different parts of the code."
] | In https://github.com/huggingface/transformers/pull/12276 we are now changing the examples to have `datasets` on the same log level as `transformers`, so that one setting can do a consistent logging across all involved components.
The trouble is that now we get a ton of these:
```
06/23/2021 12:15:31 - INFO - datasets.utils.filelock - Lock 139627640431136 acquired on /home/stas/.cache/huggingface/metrics/sacrebleu/default/default_experiment-1-0.arrow.lock
06/23/2021 12:15:31 - INFO - datasets.arrow_writer - Done writing 50 examples in 12280 bytes /home/stas/.cache/huggingface/metrics/sacrebleu/default/default_experiment-1-0.arrow.
06/23/2021 12:15:31 - INFO - datasets.arrow_dataset - Set __getitem__(key) output type to python objects for no columns (when key is int or slice) and don't output other (un-formatted) columns.
06/23/2021 12:15:31 - INFO - datasets.utils.filelock - Lock 139627640431136 released on /home/stas/.cache/huggingface/metrics/sacrebleu/default/default_experiment-1-0.arrow.lock
```
May I suggest that these can be `log.debug` as it's no informative to the user.
More examples: these are not informative - too much information:
```
06/23/2021 12:14:26 - INFO - datasets.load - Checking /home/stas/.cache/huggingface/datasets/downloads/459933f1fe47711fad2f6ff8110014ff189120b45ad159ef5b8e90ea43a174fa.e23e7d1259a8c6274a82a42a8936dd1b87225302c6dc9b7261beb3bc2daac640.py for additional imports.
06/23/2021 12:14:27 - INFO - datasets.builder - Constructing Dataset for split train, validation, test, from /home/stas/.cache/huggingface/datasets/wmt16/ro-en/1.0.0/0d9fb3e814712c785176ad8cdb9f465fbe6479000ee6546725db30ad8a8b5f8a
```
While these are:
```
06/23/2021 12:14:27 - INFO - datasets.info - Loading Dataset Infos from /home/stas/.cache/huggingface/modules/datasets_modules/datasets/wmt16/0d9fb3e814712c785176ad8cdb9f465fbe6479000ee6546725db30ad8a8b5f8a
06/23/2021 12:14:27 - WARNING - datasets.builder - Reusing dataset wmt16 (/home/stas/.cache/huggingface/datasets/wmt16/ro-en/1.0.0/0d9fb3e814712c785176ad8cdb9f465fbe6479000ee6546725db30ad8a8b5f8a)
```
I also realize that `transformers` examples don't have do use `info` for `datasets` to let the default `warning` keep logging to less noisy.
But I think currently the log levels are slightly misused and skewed by 1 level. Many `warnings` will better be `info`s and most `info`s be `debug`.
e.g.:
```
06/23/2021 12:14:27 - WARNING - datasets.builder - Reusing dataset wmt16 (/home/stas/.cache/huggingface/datasets/wmt16/ro-en/1.0.0/0d9fb3e814712c785176ad8cdb9f465fbe6479000ee6546725db30ad8a8b5f8a)
```
why is this a warning? it is informing me that the cache is used, there is nothing to be worried about. I'd have it as `info`.
Warnings are typically something that's bordering error or the first thing to check when things don't work as expected.
infrequent info is there to inform of the different stages or important events.
Everything else is debug.
At least the way I understand things.
| 2,543 |
https://github.com/huggingface/datasets/issues/2542 | `datasets.keyhash.DuplicatedKeysError` for `drop` and `adversarial_qa/adversarialQA` | [
"very much related: https://github.com/huggingface/datasets/pull/2333",
"Hi @VictorSanh, thank you for reporting this issue with duplicated keys.\r\n\r\n- The issue with \"adversarial_qa\" was fixed 23 days ago: #2433. Current version of `datasets` (1.8.0) includes the patch.\r\n- I am investigating the issue wit... | ## Describe the bug
Failure to generate the datasets (`drop` and subset `adversarialQA` from `adversarial_qa`) because of duplicate keys.
## Steps to reproduce the bug
```python
from datasets import load_dataset
load_dataset("drop")
load_dataset("adversarial_qa", "adversarialQA")
```
## Expected results
The examples keys should be unique.
## Actual results
```bash
>>> load_dataset("drop")
Using custom data configuration default
Downloading and preparing dataset drop/default (download: 7.92 MiB, generated: 111.88 MiB, post-processed: Unknown size, total: 119.80 MiB) to /home/hf/.cache/huggingface/datasets/drop/default/0.1.0/7a94f1e2bb26c4b5c75f89857c06982967d7416e5af935a9374b9bccf5068026...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/load.py", line 751, in load_dataset
use_auth_token=use_auth_token,
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/builder.py", line 575, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/builder.py", line 652, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/builder.py", line 992, in _prepare_split
num_examples, num_bytes = writer.finalize()
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/arrow_writer.py", line 409, in finalize
self.check_duplicate_keys()
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/arrow_writer.py", line 349, in check_duplicate_keys
raise DuplicatedKeysError(key)
datasets.keyhash.DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 28553293-d719-441b-8f00-ce3dc6df5398
Keys should be unique and deterministic in nature
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.7.0
- Platform: Linux-5.4.0-1044-gcp-x86_64-with-Ubuntu-18.04-bionic
- Python version: 3.7.10
- PyArrow version: 3.0.0
| 2,542 |
https://github.com/huggingface/datasets/issues/2538 | Loading partial dataset when debugging | [
"Hi ! `load_dataset` downloads the full dataset once and caches it, so that subsequent calls to `load_dataset` just reloads the dataset from your disk.\r\nThen when you specify a `split` in `load_dataset`, it will just load the requested split from the disk. If your specified split is a sliced split (e.g. `\"train[... | I am using PyTorch Lightning along with datasets (thanks for so many datasets already prepared and the great splits).
Every time I execute load_dataset for the imdb dataset it takes some time even if I specify a split involving very few samples. I guess this due to hashing as per the other issues.
Is there a way to only load part of the dataset on load_dataset? This would really speed up my workflow.
Something like a debug mode would really help. Thanks! | 2,538 |
https://github.com/huggingface/datasets/issues/2536 | Use `Audio` features for `AutomaticSpeechRecognition` task template | [
"I'm just retaking and working on #2324. 😉 ",
"Resolved via https://github.com/huggingface/datasets/pull/4006."
] | In #2533 we added a task template for speech recognition that relies on the file paths to the audio files. As pointed out by @SBrandeis this is brittle as it doesn't port easily across different OS'.
The solution is to use dedicated `Audio` features when casting the dataset. These features are not yet available in `datasets`, but should be included in the `AutomaticSpeechRecognition` template once they are. | 2,536 |
https://github.com/huggingface/datasets/issues/2532 | Tokenizer's normalization preprocessor cause misalignment in return_offsets_mapping for tokenizer classification task | [
"Hi @jerryIsHere, thanks for reporting the issue. But are you sure this is a bug in HuggingFace **Datasets**?",
"> Hi @jerryIsHere, thanks for reporting the issue. But are you sure this is a bug in HuggingFace **Datasets**?\r\n\r\nOh, I am sorry\r\nI would reopen the post on huggingface/transformers"
] | [This colab notebook](https://colab.research.google.com/drive/151gKyo0YIwnlznrOHst23oYH_a3mAe3Z?usp=sharing) implements a token classification input pipeline extending the logic from [this hugging example](https://huggingface.co/transformers/custom_datasets.html#tok-ner).
The pipeline works fine with most instance in different languages, but unfortunately, [the Japanese Kana ligature (a form of abbreviation? I don't know Japanese well)](https://en.wikipedia.org/wiki/Kana_ligature) break the alignment of `return_offsets_mapping`:
Without the try catch block, it riase `ValueError: NumPy boolean array indexing assignment cannot assign 88 input values to the 87 output values where the mask is true`, example shown here [(another colab notebook)](https://colab.research.google.com/drive/1MmOqf3ppzzdKKyMWkn0bJy6DqzOO0SSm?usp=sharing)
It is clear that the normalizer is the process that break the alignment, as it is observed that `tokenizer._tokenizer.normalizer.normalize_str('ヿ')` return 'コト'.
One workaround is to include `tokenizer._tokenizer.normalizer.normalize_str` before the tokenizer preprocessing pipeline, which is also provided in the [first colab notebook](https://colab.research.google.com/drive/151gKyo0YIwnlznrOHst23oYH_a3mAe3Z?usp=sharing) with the name `udposTestDatasetWorkaround`.
I guess similar logics should be included inside the tokenizer and the offsets_mapping generation process such that user don't need to include them in their code. But I don't understand the code of tokenizer well that I think I am not able to do this.
p.s.
**I am using my own dataset building script in the provided example, but the script should be equivalent to the changes made by this [update](https://github.com/huggingface/datasets/pull/2466)**
`get_dataset `is just a simple wrapping for `load_dataset`
and the `tokenizer` is just `XLMRobertaTokenizerFast.from_pretrained("xlm-roberta-large")` | 2,532 |
https://github.com/huggingface/datasets/issues/2528 | Logging cannot be set to NOTSET similar to transformers | [
"Hi @joshzwiebel, thanks for reporting. We are going to align with `transformers`."
] | ## Describe the bug
In the transformers library you can set the verbosity level to logging.NOTSET to work around the usage of tqdm and IPywidgets, however in Datasets this is no longer possible. This is because transformers set the verbosity level of tqdm with [this](https://github.com/huggingface/transformers/blob/b53bc55ba9bb10d5ee279eab51a2f0acc5af2a6b/src/transformers/file_utils.py#L1449)
`disable=bool(logging.get_verbosity() == logging.NOTSET)`
and datasets accomplishes this like [so](https://github.com/huggingface/datasets/blob/83554e410e1ab8c6f705cfbb2df7953638ad3ac1/src/datasets/utils/file_utils.py#L493)
`not_verbose = bool(logger.getEffectiveLevel() > WARNING)`
## Steps to reproduce the bug
```python
import datasets
import logging
datasets.logging.get_verbosity = lambda : logging.NOTSET
datasets.load_dataset("patrickvonplaten/librispeech_asr_dummy")
```
## Expected results
The code should download and load the dataset as normal without displaying progress bars
## Actual results
```ImportError Traceback (most recent call last)
<ipython-input-4-aec65c0509c6> in <module>
----> 1 datasets.load_dataset("patrickvonplaten/librispeech_asr_dummy")
~/venv/lib/python3.7/site-packages/datasets/load.py in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, ignore_verifications, keep_in_memory, save_infos, script_version, use_auth_token, task, **config_kwargs)
713 dataset=True,
714 return_resolved_file_path=True,
--> 715 use_auth_token=use_auth_token,
716 )
717 # Set the base path for downloads as the parent of the script location
~/venv/lib/python3.7/site-packages/datasets/load.py in prepare_module(path, script_version, download_config, download_mode, dataset, force_local_path, dynamic_modules_path, return_resolved_file_path, **download_kwargs)
350 file_path = hf_bucket_url(path, filename=name, dataset=False)
351 try:
--> 352 local_path = cached_path(file_path, download_config=download_config)
353 except FileNotFoundError:
354 raise FileNotFoundError(
~/venv/lib/python3.7/site-packages/datasets/utils/file_utils.py in cached_path(url_or_filename, download_config, **download_kwargs)
289 use_etag=download_config.use_etag,
290 max_retries=download_config.max_retries,
--> 291 use_auth_token=download_config.use_auth_token,
292 )
293 elif os.path.exists(url_or_filename):
~/venv/lib/python3.7/site-packages/datasets/utils/file_utils.py in get_from_cache(url, cache_dir, force_download, proxies, etag_timeout, resume_download, user_agent, local_files_only, use_etag, max_retries, use_auth_token)
668 headers=headers,
669 cookies=cookies,
--> 670 max_retries=max_retries,
671 )
672
~/venv/lib/python3.7/site-packages/datasets/utils/file_utils.py in http_get(url, temp_file, proxies, resume_size, headers, cookies, timeout, max_retries)
493 initial=resume_size,
494 desc="Downloading",
--> 495 disable=not_verbose,
496 )
497 for chunk in response.iter_content(chunk_size=1024):
~/venv/lib/python3.7/site-packages/tqdm/notebook.py in __init__(self, *args, **kwargs)
217 total = self.total * unit_scale if self.total else self.total
218 self.container = self.status_printer(
--> 219 self.fp, total, self.desc, self.ncols)
220 self.sp = self.display
221
~/venv/lib/python3.7/site-packages/tqdm/notebook.py in status_printer(_, total, desc, ncols)
95 if IProgress is None: # #187 #451 #558 #872
96 raise ImportError(
---> 97 "IProgress not found. Please update jupyter and ipywidgets."
98 " See https://ipywidgets.readthedocs.io/en/stable"
99 "/user_install.html")
ImportError: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.8.0
- Platform: Linux-5.4.95-42.163.amzn2.x86_64-x86_64-with-debian-10.8
- Python version: 3.7.10
- PyArrow version: 3.0.0
I am running this code on Deepnote and which important to this issue **does not** support IPywidgets
| 2,528 |
https://github.com/huggingface/datasets/issues/2526 | Add COCO datasets | [
"I'm currently adding it, the entire dataset is quite big around 30 GB so I add splits separately. You can take a look here https://huggingface.co/datasets/merve/coco",
"I talked to @lhoestq and it's best if I download this dataset through TensorFlow datasets instead, so I'll be implementing that one really soon.... | ## Adding a Dataset
- **Name:** COCO
- **Description:** COCO is a large-scale object detection, segmentation, and captioning dataset.
- **Paper + website:** https://cocodataset.org/#home
- **Data:** https://cocodataset.org/#download
- **Motivation:** It would be great to have COCO available in HuggingFace datasets, as we are moving beyond just text. COCO includes multi-modalities (images + text), as well as a huge amount of images annotated with objects, segmentation masks, keypoints etc., on which models like DETR (which I recently added to HuggingFace Transformers) are trained. Currently, one needs to download everything from the website and place it in a local folder, but it would be much easier if we can directly access it through the datasets API.
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
| 2,526 |
https://github.com/huggingface/datasets/issues/2523 | Fr | [] | __Originally posted by @lewtun in https://github.com/huggingface/datasets/pull/2469__ | 2,523 |
https://github.com/huggingface/datasets/issues/2522 | Documentation Mistakes in Dataset: emotion | [
"Hi,\r\n\r\nthis issue has been already reported in the dataset repo (https://github.com/dair-ai/emotion_dataset/issues/2), so this is a bug on their side.",
"The documentation has another bug in the dataset card [here](https://huggingface.co/datasets/emotion). \r\n\r\nIn the dataset summary **six** emotions are ... | As per documentation,
Dataset: emotion
Homepage: https://github.com/dair-ai/emotion_dataset
Dataset: https://github.com/huggingface/datasets/blob/master/datasets/emotion/emotion.py
Permalink: https://huggingface.co/datasets/viewer/?dataset=emotion
Emotion is a dataset of English Twitter messages with eight basic emotions: anger, anticipation, disgust, fear, joy, sadness, surprise, and trust. For more detailed information please refer to the paper.
But when we view the data, there are only 6 emotions, anger, fear, joy, sadness, surprise, and trust. | 2,522 |
https://github.com/huggingface/datasets/issues/2520 | Datasets with tricky task templates | [
"The `task_templates` API is deprecated in favor of the `train-eval-index` YAML field, so I'm closing this issue."
] | I'm collecting a list of datasets here that don't follow the "standard" taxonomy and require further investigation to implement task templates for.
## Text classification
* [hatexplain](https://huggingface.co/datasets/hatexplain): ostensibly a form of text classification, but not in the standard `(text, target)` format and each sample appears to be tokenized.
* [muchocine](https://huggingface.co/datasets/muchocine): contains two candidate text columns (long-form and summary) which in principle requires two `TextClassification` templates which is not currently supported | 2,520 |
https://github.com/huggingface/datasets/issues/2516 | datasets.map pickle issue resulting in invalid mapping function | [
"Hi ! `map` calls `__getstate__` using `dill` to hash your map function. This is used by the caching mechanism to recover previously computed results. That's why you don't see any `__setstate__` call.\r\n\r\nWhy do you change an attribute of your tokenizer when `__getstate__` is called ?",
"@lhoestq because if I ... | I trained my own tokenizer, and I needed to use a python custom class. Because of this I have to detach the custom step before saving and reattach after restore. I did this using the standard pickle `__get_state__` / `__set_state__` mechanism. I think it's correct but it fails when I use it inside a function which is mapped to a dataset, i.e. in the manner of run_mlm.py and other huggingface scripts.
The following reproduces the issue - most likely I'm missing something
A simulated tokeniser which can be pickled
```
class CustomTokenizer:
def __init__(self):
self.state = "init"
def __getstate__(self):
print("__getstate__ called")
out = self.__dict__.copy()
self.state = "pickled"
return out
def __setstate__(self, d):
print("__setstate__ called")
self.__dict__ = d
self.state = "restored"
tokenizer = CustomTokenizer()
```
Test that it actually works - prints "__getstate__ called" and "__setstate__ called"
```
import pickle
serialized = pickle.dumps(tokenizer)
restored = pickle.loads(serialized)
assert restored.state == "restored"
```
Simulate a function that tokenises examples, when dataset.map is called, this function
```
def tokenize_function(examples):
assert tokenizer.state == "restored" # this shouldn't fail but it does
output = tokenizer(examples) # this will fail as tokenizer isn't really a tokenizer
return output
```
Use map to simulate tokenization
```
import glob
from datasets import load_dataset
assert tokenizer.state == "restored"
train_files = glob.glob('train*.csv')
validation_files = glob.glob('validation*.csv')
datasets = load_dataset("csv", data_files=dict(train=train_files, validation=validation_files))
tokenized_datasets = datasets.map(
tokenize_function,
batched=True,
)
```
What's happening is I can see that __getstate__ is called but not __setstate__, so the state of `tokenize_function` is invalid at the point that it's actually executed. This doesn't matter as far as I can see for the standard tokenizers as they don't use __getstate__ / __setstate__. I'm not sure if there's another hook I'm supposed to implement as well?
---------------------------------------------------------------------------
AssertionError Traceback (most recent call last)
<ipython-input-22-a2aef4f74aaa> in <module>
8 tokenized_datasets = datasets.map(
9 tokenize_function,
---> 10 batched=True,
11 )
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/dataset_dict.py in map(self, function, with_indices, input_columns, batched, batch_size, remove_columns, keep_in_memory, load_from_cache_file, cache_file_names, writer_batch_size, features, disable_nullable, fn_kwargs, num_proc, desc)
487 desc=desc,
488 )
--> 489 for k, dataset in self.items()
490 }
491 )
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/dataset_dict.py in <dictcomp>(.0)
487 desc=desc,
488 )
--> 489 for k, dataset in self.items()
490 }
491 )
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in map(self, function, with_indices, input_columns, batched, batch_size, drop_last_batch, remove_columns, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, features, disable_nullable, fn_kwargs, num_proc, suffix_template, new_fingerprint, desc)
1633 fn_kwargs=fn_kwargs,
1634 new_fingerprint=new_fingerprint,
-> 1635 desc=desc,
1636 )
1637 else:
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in wrapper(*args, **kwargs)
184 }
185 # apply actual function
--> 186 out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
187 datasets: List["Dataset"] = list(out.values()) if isinstance(out, dict) else [out]
188 # re-apply format to the output
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/fingerprint.py in wrapper(*args, **kwargs)
395 # Call actual function
396
--> 397 out = func(self, *args, **kwargs)
398
399 # Update fingerprint of in-place transforms + update in-place history of transforms
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in _map_single(self, function, with_indices, input_columns, batched, batch_size, drop_last_batch, remove_columns, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, features, disable_nullable, fn_kwargs, new_fingerprint, rank, offset, desc)
1961 indices,
1962 check_same_num_examples=len(input_dataset.list_indexes()) > 0,
-> 1963 offset=offset,
1964 )
1965 except NumExamplesMismatch:
~/.pyenv/versions/3.7.6/envs/xxx/lib/python3.7/site-packages/datasets/arrow_dataset.py in apply_function_on_filtered_inputs(inputs, indices, check_same_num_examples, offset)
1853 effective_indices = [i + offset for i in indices] if isinstance(indices, list) else indices + offset
1854 processed_inputs = (
-> 1855 function(*fn_args, effective_indices, **fn_kwargs) if with_indices else function(*fn_args, **fn_kwargs)
1856 )
1857 if update_data is None:
<ipython-input-21-8ee4a8ba5b1b> in tokenize_function(examples)
1 def tokenize_function(examples):
----> 2 assert tokenizer.state == "restored"
3 tokenizer(examples)
4 return examples
| 2,516 |
https://github.com/huggingface/datasets/issues/2514 | Can datasets remove duplicated rows? | [
"Hi ! For now this is probably the best option.\r\nWe might add a feature like this in the feature as well.\r\n\r\nDo you know any deduplication method that works on arbitrary big datasets without filling up RAM ?\r\nOtherwise we can have do the deduplication in memory like pandas but I feel like this is going to b... | **Is your feature request related to a problem? Please describe.**
i find myself more and more relying on datasets just to do all the preprocessing. One thing however, for removing duplicated rows, I couldn't find out how and am always converting datasets to pandas to do that..
**Describe the solution you'd like**
have a functionality of " remove duplicated rows"
**Describe alternatives you've considered**
convert dataset to pandas, remove duplicate, and convert back...
**Additional context**
no | 2,514 |
https://github.com/huggingface/datasets/issues/2513 | Corelation should be Correlation | [
"Hi @colbym-MM, thanks for reporting. We are fixing it."
] | https://github.com/huggingface/datasets/blob/0e87e1d053220e8ecddfa679bcd89a4c7bc5af62/metrics/matthews_correlation/matthews_correlation.py#L66 | 2,513 |
https://github.com/huggingface/datasets/issues/2512 | seqeval metric does not work with a recent version of sklearn: classification_report() got an unexpected keyword argument 'output_dict' | [
"Sorry, I was using an old version of sequeval"
] | ## Describe the bug
A clear and concise description of what the bug is.
## Steps to reproduce the bug
```python
from datasets import load_dataset, load_metric
seqeval = load_metric("seqeval")
seqeval.compute(predictions=[['A']], references=[['A']])
```
## Expected results
The function computes a dict with metrics
## Actual results
```
---------------------------------------------------------------------------
TypeError Traceback (most recent call last)
<ipython-input-39-69a57f5cf06f> in <module>
1 from datasets import load_dataset, load_metric
2 seqeval = load_metric("seqeval")
----> 3 seqeval.compute(predictions=[['A']], references=[['A']])
~/p3/lib/python3.7/site-packages/datasets/metric.py in compute(self, *args, **kwargs)
396 references = self.data["references"]
397 with temp_seed(self.seed):
--> 398 output = self._compute(predictions=predictions, references=references, **kwargs)
399
400 if self.buf_writer is not None:
~/.cache/huggingface/modules/datasets_modules/metrics/seqeval/81eda1ff004361d4fa48754a446ec69bb7aa9cf4d14c7215f407d1475941c5ff/seqeval.py in _compute(self, predictions, references, suffix)
95
96 def _compute(self, predictions, references, suffix=False):
---> 97 report = classification_report(y_true=references, y_pred=predictions, suffix=suffix, output_dict=True)
98 report.pop("macro avg")
99 report.pop("weighted avg")
TypeError: classification_report() got an unexpected keyword argument 'output_dict'
```
## Environment info
sklearn=0.24
datasets=1.1.3
| 2,512 |
https://github.com/huggingface/datasets/issues/2511 | Add C4 | [
"Update on this: I'm computing the checksums of the data files. It will be available soon",
"Added in #2575 :)"
] | ## Adding a Dataset
- **Name:** *C4*
- **Description:** *https://github.com/allenai/allennlp/discussions/5056*
- **Paper:** *https://arxiv.org/abs/1910.10683*
- **Data:** *https://huggingface.co/datasets/allenai/c4*
- **Motivation:** *Used a lot for pretraining*
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
Should fix https://github.com/huggingface/datasets/issues/1710 | 2,511 |
https://github.com/huggingface/datasets/issues/2508 | Load Image Classification Dataset from Local | [
"Hi ! Is this folder structure a standard, a bit like imagenet ?\r\nIn this case maybe we can consider having a dataset loader for cifar-like, imagenet-like, squad-like, conll-like etc. datasets ?\r\n```python\r\nfrom datasets import load_dataset\r\n\r\nmy_custom_cifar = load_dataset(\"cifar_like\", data_dir=\"path... | **Is your feature request related to a problem? Please describe.**
Yes - we would like to load an image classification dataset with datasets without having to write a custom data loader.
**Describe the solution you'd like**
Given a folder structure with images of each class in each folder, the ability to load these folders into a HuggingFace dataset like "cifar10".
**Describe alternatives you've considered**
Implement ViT training outside of the HuggingFace Trainer and without datasets (we did this but prefer to stay on the main path)
Write custom data loader logic
**Additional context**
We're training ViT on custom dataset
| 2,508 |
https://github.com/huggingface/datasets/issues/2503 | SubjQA wrong boolean values in entries | [
"Hi @arnaudstiegler, thanks for reporting. I'm investigating it.",
"@arnaudstiegler I have just checked that these mismatches are already present in the original dataset: https://github.com/megagonlabs/SubjQA\r\n\r\nWe are going to contact the dataset owners to report this.",
"I have:\r\n- opened an issue in th... | ## Describe the bug
SubjQA seems to have a boolean that's consistently wrong.
It defines:
- question_subj_level: The subjectiviy level of the question (on a 1 to 5 scale with 1 being the most subjective).
- is_ques_subjective: A boolean subjectivity label derived from question_subj_level (i.e., scores below 4 are considered as subjective)
However, `is_ques_subjective` seems to have wrong values in the entire dataset.
For instance, in the example in the dataset card, we have:
- "question_subj_level": 2
- "is_ques_subjective": false
However, according to the description, the question should be subjective since the `question_subj_level` is below 4
| 2,503 |
https://github.com/huggingface/datasets/issues/2499 | Python Programming Puzzles | [
"👀 @TalSchuster",
"Thanks @VictorSanh!\r\nThere's also a [notebook](https://aka.ms/python_puzzles) and [demo](https://aka.ms/python_puzzles_study) available now to try out some of the puzzles"
] | ## Adding a Dataset
- **Name:** Python Programming Puzzles
- **Description:** Programming challenge called programming puzzles, as an objective and comprehensive evaluation of program synthesis
- **Paper:** https://arxiv.org/pdf/2106.05784.pdf
- **Data:** https://github.com/microsoft/PythonProgrammingPuzzles ([Scrolling through the data](https://github.com/microsoft/PythonProgrammingPuzzles/blob/main/problems/README.md))
- **Motivation:** Spans a large range of difficulty, problems, and domains. A useful resource for evaluation as we don't have a clear understanding of the abilities and skills of extremely large LMs.
Note: it's a growing dataset (contributions are welcome), so we'll need careful versioning for this dataset.
Instructions to add a new dataset can be found [here](https://github.com/huggingface/datasets/blob/master/ADD_NEW_DATASET.md).
| 2,499 |
https://github.com/huggingface/datasets/issues/2498 | Improve torch formatting performance | [
"That’s interesting thanks, let’s see what we can do. Can you detail your last sentence? I’m not sure I understand it well.",
"Hi ! I just re-ran a quick benchmark and using `to_numpy()` seems to be faster now:\r\n\r\n```python\r\nimport pyarrow as pa # I used pyarrow 3.0.0\r\nimport numpy as np\r\n\r\nn, max_le... | **Is your feature request related to a problem? Please describe.**
It would be great, if possible, to further improve read performance of raw encoded datasets and their subsequent conversion to torch tensors.
A bit more background. I am working on LM pre-training using HF ecosystem. We use encoded HF Wikipedia and BookCorpus datasets. The training machines are similar to DGX-1 workstations. We use HF trainer torch.distributed training approach on a single machine with 8 GPUs.
The current performance is about 30% slower than NVidia optimized BERT [examples](https://github.com/NVIDIA/DeepLearningExamples/tree/master/PyTorch/LanguageModeling) baseline. Quite a bit of customized code and training loop tricks were used to achieve the baseline performance. It would be great to achieve the same performance while using nothing more than off the shelf HF ecosystem. Perhaps, in the future, with @stas00 work on deepspeed integration, it could even be exceeded.
**Describe the solution you'd like**
Using profiling tools we've observed that appx. 25% of cumulative run time is spent on data loader next call.
As you can observe most of the data loader next call is spent in HF datasets torch_formatter.py format_batch call.
Digging a bit deeper into format_batch we can see the following profiler data:
Once again, a lot of time is spent in pyarrow table conversion to pandas which seems like an intermediary step. Offline @lhoestq told me that this approach was, for some unknown reason, faster than direct to numpy conversion.
**Describe alternatives you've considered**
I am not familiar with pyarrow and have not yet considered the alternatives to the current approach.
Most of the online advice around data loader performance improvements revolve around increasing number of workers, using pin memory for copying tensors from host device to gpus but we've already tried these avenues without much performance improvement. Weights & Biases dashboard for the pre-training task reports CPU utilization of ~ 10%, GPUs are completely saturated (GPU utilization is above 95% on all GPUs), while disk utilization is above 90%.
| 2,498 |
https://github.com/huggingface/datasets/issues/2496 | Dataset fingerprint changes after moving the cache directory, which prevent cache reload when using `map` | [] | `Dataset.map` uses the dataset fingerprint (a hash) for caching.
However the fingerprint seems to change when someone moves the cache directory of the dataset.
This is because it uses the default fingerprint generation:
1. the dataset path is used to get the fingerprint
2. the modification times of the arrow file is also used to get the fingerprint
To fix that we could set the fingerprint of the dataset to be a hash of (<dataset_name>, <config_name>, <version>, <script_hash>), i.e. a hash of the the cache path relative to the cache directory. | 2,496 |
https://github.com/huggingface/datasets/issues/2495 | JAX formatting | [] | We already support pytorch, tensorflow, numpy, pandas and arrow dataset formatting. Let's add jax as well | 2,495 |
https://github.com/huggingface/datasets/issues/2494 | Improve docs on Enhancing performance | [
"Hi @albertvillanova, I hope you are doing well.\r\n\r\nI am interested in this issue, is this still unresolved and open ?\r\n\r\nThe link you have provided in the above message directs to a webpage that does not exist.\r\n\r\nThanks and Regards"
] | In the ["Enhancing performance"](https://huggingface.co/docs/datasets/loading_datasets.html#enhancing-performance) section of docs, add specific use cases:
- How to make datasets the fastest
- How to make datasets take the less RAM
- How to make datasets take the less hard drive mem
cc: @thomwolf
| 2,494 |
https://github.com/huggingface/datasets/issues/2489 | Allow latest pyarrow version once segfault bug is fixed | [] | As pointed out by @symeneses (see https://github.com/huggingface/datasets/pull/2268#issuecomment-860048613), pyarrow has fixed the segfault bug present in version 4.0.0 (see https://issues.apache.org/jira/browse/ARROW-12568):
- it was fixed on 3 May 2021
- version 4.0.1 was released on 19 May 2021 with the bug fix | 2,489 |
https://github.com/huggingface/datasets/issues/2485 | Implement layered building | [] | As discussed with @stas00 and @lhoestq (see also here https://github.com/huggingface/datasets/issues/2481#issuecomment-859712190):
> My suggestion for this would be to have this enabled by default.
>
> Plus I don't know if there should be a dedicated issue to that is another functionality. But I propose layered building rather than all at once. That is:
>
> 1. uncompress a handful of files via a generator enough to generate one arrow file
> 2. process arrow file 1
> 3. delete all the files that went in and aren't needed anymore.
>
> rinse and repeat.
>
> 1. This way much less disc space will be required - e.g. on JZ we won't be running into inode limitation, also it'd help with the collaborative hub training project
> 2. The user doesn't need to go and manually clean up all the huge files that were left after pre-processing
> 3. It would already include deleting temp files this issue is talking about
>
> I wonder if the new streaming API would be of help, except here the streaming would be into arrow files as the destination, rather than dataloaders. | 2,485 |
https://github.com/huggingface/datasets/issues/2484 | Implement loading a dataset builder | [
"#self-assign"
] | As discussed with @stas00 and @lhoestq, this would allow things like:
```python
from datasets import load_dataset_builder
dataset_name = "openwebtext"
builder = load_dataset_builder(dataset_name)
print(builder.cache_dir)
``` | 2,484 |
https://github.com/huggingface/datasets/issues/2481 | Delete extracted files to save disk space | [
"My suggestion for this would be to have this enabled by default.\r\n\r\nPlus I don't know if there should be a dedicated issue to that is another functionality. But I propose layered building rather than all at once. That is:\r\n\r\n1. uncompress a handful of files via a generator enough to generate one arrow file... | As discussed with @stas00 and @lhoestq, allowing the deletion of extracted files would save a great amount of disk space to typical user. | 2,481 |
https://github.com/huggingface/datasets/issues/2480 | Set download/extracted paths configurable | [
"For example to be able to send uncompressed and temp build files to another volume/partition, so that the user gets the minimal disk usage on their primary setup - and ends up with just the downloaded compressed data + arrow files, but outsourcing the huge files and building to another partition. e.g. on JZ there... | As discussed with @stas00 and @lhoestq, setting these paths configurable may allow to overcome disk space limitation on different partitions/drives.
TODO:
- [x] Set configurable extracted datasets path: #2487
- [x] Set configurable downloaded datasets path: #2488
- [ ] Set configurable "incomplete" datasets path? | 2,480 |
https://github.com/huggingface/datasets/issues/2478 | Create release script | [
"I've aligned the release script with Transformers in #6004, so I think this issue can be closed."
] | Create a script so that releases can be done automatically (as done in `transformers`). | 2,478 |
https://github.com/huggingface/datasets/issues/2475 | Issue in timit_asr database | [
"This bug was fixed in #1995. Upgrading datasets to version 1.6 fixes the issue!",
"Indeed was a fixed bug.\r\nWorks on version 1.8\r\nThanks "
] | ## Describe the bug
I am trying to load the timit_asr dataset however only the first record is shown (duplicated over all the rows).
I am using the next code line
dataset = load_dataset(“timit_asr”, split=“test”).shuffle().select(range(10))
The above code result with the same sentence duplicated ten times.
It also happens when I use the dataset viewer at Streamlit .
## Steps to reproduce the bug
from datasets import load_dataset
dataset = load_dataset(“timit_asr”, split=“test”).shuffle().select(range(10))
data = dataset.to_pandas()
# Sample code to reproduce the bug
```
## Expected results
table with different row information
## Actual results
Specify the actual results or traceback.
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.4.1 (also occur in the latest version)
- Platform: Linux-4.15.0-143-generic-x86_64-with-Ubuntu-18.04-bionic
- Python version: 3.6.9
- PyTorch version (GPU?): 1.8.1+cu102 (False)
- Tensorflow version (GPU?): 1.15.3 (False)
- Using GPU in script?: No
- Using distributed or parallel set-up in script?: No
- `datasets` version:
- Platform:
- Python version:
- PyArrow version:
| 2,475 |
https://github.com/huggingface/datasets/issues/2474 | cache_dir parameter for load_from_disk ? | [
"Hi ! `load_from_disk` doesn't move the data. If you specify a local path to your mounted drive, then the dataset is going to be loaded directly from the arrow file in this directory. The cache files that result from `map` operations are also stored in the same directory by default.\r\n\r\nHowever note than writing... | **Is your feature request related to a problem? Please describe.**
When using Google Colab big datasets can be an issue, as they won't fit on the VM's disk. Therefore mounting google drive could be a possible solution. Unfortunatly when loading my own dataset by using the _load_from_disk_ function, the data gets cached to the VM's disk:
`
from datasets import load_from_disk
myPreprocessedData = load_from_disk("/content/gdrive/MyDrive/ASR_data/myPreprocessedData")
`
I know that chaching on google drive could slow down learning. But at least it would run.
**Describe the solution you'd like**
Add cache_Dir parameter to the load_from_disk function.
**Describe alternatives you've considered**
It looks like you could write a custom loading script for the load_dataset function. But this seems to be much too complex for my use case. Is there perhaps a template here that uses the load_from_disk function?
| 2,474 |
https://github.com/huggingface/datasets/issues/2472 | Fix automatic generation of Zenodo DOI | [
"I have received a reply from Zenodo support:\r\n> We are currently investigating and fixing this issue related to GitHub releases. As soon as we have solved it we will reach back to you.",
"Other repo maintainers had the same problem with Zenodo. \r\n\r\nThere is an open issue on their GitHub repo: zenodo/zenodo... | After the last release of Datasets (1.8.0), the automatic generation of the Zenodo DOI failed: it appears in yellow as "Received", instead of in green as "Published".
I have contacted Zenodo support to fix this issue.
TODO:
- [x] Check with Zenodo to fix the issue
- [x] Check BibTeX entry is right | 2,472 |
https://github.com/huggingface/datasets/issues/2471 | Fix PermissionError on Windows when using tqdm >=4.50.0 | [] | See: https://app.circleci.com/pipelines/github/huggingface/datasets/235/workflows/cfb6a39f-68eb-4802-8b17-2cd5e8ea7369/jobs/1111
```
PermissionError: [WinError 32] The process cannot access the file because it is being used by another process
``` | 2,471 |
https://github.com/huggingface/datasets/issues/2470 | Crash when `num_proc` > dataset length for `map()` on a `datasets.Dataset`. | [
"Hi ! It looks like the issue comes from pyarrow. What version of pyarrow are you using ? How did you install it ?",
"Thank you for the quick reply! I have `pyarrow==4.0.0`, and I am installing with `pip`. It's not one of my explicit dependencies, so I assume it came along with something else.",
"Could you tryi... | ## Describe the bug
Crash if when using `num_proc` > 1 (I used 16) for `map()` on a `datasets.Dataset`.
I believe I've had cases where `num_proc` > 1 works before, but now it seems either inconsistent, or depends on my data. I'm not sure whether the issue is on my end, because it's difficult for me to debug! Any tips greatly appreciated, I'm happy to provide more info if it would helps us diagnose.
## Steps to reproduce the bug
```python
# this function will be applied with map()
def tokenize_function(examples):
return tokenizer(
examples["text"],
padding=PaddingStrategy.DO_NOT_PAD,
truncation=True,
)
# data_files is a Dict[str, str] mapping name -> path
datasets = load_dataset("text", data_files={...})
# this is where the error happens if num_proc = 16,
# but is fine if num_proc = 1
tokenized_datasets = datasets.map(
tokenize_function,
batched=True,
num_proc=num_workers,
)
```
## Expected results
The `map()` function succeeds with `num_proc` > 1.
## Actual results
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.6.2
- Platform: Linux-5.4.0-73-generic-x86_64-with-glibc2.31
- Python version: 3.9.5
- PyTorch version (GPU?): 1.8.1+cu111 (True)
- Tensorflow version (GPU?): not installed (NA)
- Using GPU in script?: Yes, but I think N/A for this issue
- Using distributed or parallel set-up in script?: Multi-GPU on one machine, but I think also N/A for this issue
| 2,470 |
https://github.com/huggingface/datasets/issues/2462 | Merge DatasetDict and Dataset | [
"Any update on this? @lhoestq ",
"Unless there is high demande I don't think we will end up implementing this. This is a lot of work with very few advantages"
] | As discussed in #2424 and #2437 (please see there for detailed conversation):
- It would be desirable to improve UX with respect the confusion between DatasetDict and Dataset.
- The difference between Dataset and DatasetDict is an additional abstraction complexity that confuses "typical" end users.
- A user expects a "Dataset" (whatever it contains multiple or a single split) and maybe it could be interesting to try to simplify the user-facing API as much as possible to hide this complexity from the end user.
Here is a proposal for discussion and refined (and potential abandon if it's not good enough):
- let's consider that a DatasetDict is also a Dataset with the various split concatenated one after the other
- let's disallow the use of integers in split names (probably not a very big breaking change)
- when you index with integers you access the examples progressively in split after the other is finished (in a deterministic order)
- when you index with strings/split name you have the same behavior as now (full backward compat)
- let's then also have all the methods of a Dataset on the DatasetDict
The end goal would be to merge both Dataset and DatasetDict object in a single object that would be (pretty much totally) backward compatible with both.
There are a few things that we could discuss if we want to merge Dataset and DatasetDict:
1. what happens if you index by a string ? Does it return the column or the split ? We could disallow conflicts between column names and split names to avoid ambiguities. It can be surprising to be able to get a column or a split using the same indexing feature
```
from datasets import load_dataset
dataset = load_dataset(...)
dataset["train"]
dataset["input_ids"]
```
2. what happens when you iterate over the object ? I guess it should iterate over the examples as a Dataset object, but a DatasetDict used to iterate over the splits as they are the dictionary keys. This is a breaking change that we can discuss.
Moreover regarding your points:
- integers are not allowed as split names already
- it's definitely doable to have all the methods. Maybe some of them like train_test_split that is currently only available for Dataset can be tweaked to work for a split dataset
cc: @thomwolf @lhoestq | 2,462 |
https://github.com/huggingface/datasets/issues/2459 | `Proto_qa` hosting seems to be broken | [
"@VictorSanh , I think @mariosasko is already working on it. "
] | ## Describe the bug
The hosting (on Github) of the `proto_qa` dataset seems broken. I haven't investigated more yet, just flagging it for now.
@zaidalyafeai if you want to dive into it, I think it's just a matter of changing the links in `proto_qa.py`
## Steps to reproduce the bug
```python
from datasets import load_dataset
dataset = load_dataset("proto_qa")
```
## Actual results
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/load.py", line 751, in load_dataset
use_auth_token=use_auth_token,
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/builder.py", line 575, in download_and_prepare
dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/builder.py", line 630, in _download_and_prepare
split_generators = self._split_generators(dl_manager, **split_generators_kwargs)
File "/home/hf/.cache/huggingface/modules/datasets_modules/datasets/proto_qa/445346efaad5c5f200ecda4aa7f0fb50ff1b55edde3003be424a2112c3e8102e/proto_qa.py", line 131, in _split_generators
train_fpath = dl_manager.download(_URLs[self.config.name]["train"])
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/utils/download_manager.py", line 199, in download
num_proc=download_config.num_proc,
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/utils/py_utils.py", line 195, in map_nested
return function(data_struct)
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/utils/download_manager.py", line 218, in _download
return cached_path(url_or_filename, download_config=download_config)
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/utils/file_utils.py", line 291, in cached_path
use_auth_token=download_config.use_auth_token,
File "/home/hf/dev/promptsource/.venv/lib/python3.7/site-packages/datasets/utils/file_utils.py", line 621, in get_from_cache
raise FileNotFoundError("Couldn't find file at {}".format(url))
FileNotFoundError: Couldn't find file at https://raw.githubusercontent.com/iesl/protoqa-data/master/data/train/protoqa_train.jsonl
``` | 2,459 |
https://github.com/huggingface/datasets/issues/2458 | Revert default in-memory for small datasets | [
"cc: @krandiash (pinged in reverted PR)."
] | Users are reporting issues and confusion about setting default in-memory to True for small datasets.
We see 2 clear use cases of Datasets:
- the "canonical" way, where you can work with very large datasets, as they are memory-mapped and cached (after every transformation)
- some edge cases (speed benchmarks, interactive/exploratory analysis,...), where default in-memory can explicitly be enabled, and no caching will be done
After discussing with @lhoestq we have agreed to:
- revert this feature (implemented in #2182)
- explain in the docs how to optimize speed/performance by setting default in-memory
cc: @stas00 https://github.com/huggingface/datasets/pull/2409#issuecomment-856210552 | 2,458 |
https://github.com/huggingface/datasets/issues/2452 | MRPC test set differences between torch and tensorflow datasets | [
"Realized that `tensorflow_datasets` is not provided by Huggingface and should therefore raise the issue there."
] | ## Describe the bug
When using `load_dataset("glue", "mrpc")` to load the MRPC dataset, the test set includes the labels. When using `tensorflow_datasets.load('glue/{}'.format('mrpc'))` to load the dataset the test set does not contain the labels. There should be consistency between torch and tensorflow ways of importing the GLUE datasets.
## Steps to reproduce the bug
Minimal working code
```python
from datasets import load_dataset
import tensorflow as tf
import tensorflow_datasets
# torch
dataset = load_dataset("glue", "mrpc")
# tf
data = tensorflow_datasets.load('glue/{}'.format('mrpc'))
data = list(data['test'].as_numpy_iterator())
for i in range(40,50):
tf_sentence1 = data[i]['sentence1'].decode("utf-8")
tf_sentence2 = data[i]['sentence2'].decode("utf-8")
tf_label = data[i]['label']
index = data[i]['idx']
print('Index {}'.format(index))
torch_sentence1 = dataset['test']['sentence1'][index]
torch_sentence2 = dataset['test']['sentence2'][index]
torch_label = dataset['test']['label'][index]
print('Tensorflow: \n\tSentence1 {}\n\tSentence2 {}\n\tLabel {}'.format(tf_sentence1, tf_sentence2, tf_label))
print('Torch: \n\tSentence1 {}\n\tSentence2 {}\n\tLabel {}'.format(torch_sentence1, torch_sentence2, torch_label))
```
Sample output
```
Index 954
Tensorflow:
Sentence1 Sabri Yakou , an Iraqi native who is a legal U.S. resident , appeared before a federal magistrate yesterday on charges of violating U.S. arms-control laws .
Sentence2 The elder Yakou , an Iraqi native who is a legal U.S. resident , appeared before a federal magistrate Wednesday on charges of violating U.S. arms control laws .
Label -1
Torch:
Sentence1 Sabri Yakou , an Iraqi native who is a legal U.S. resident , appeared before a federal magistrate yesterday on charges of violating U.S. arms-control laws .
Sentence2 The elder Yakou , an Iraqi native who is a legal U.S. resident , appeared before a federal magistrate Wednesday on charges of violating U.S. arms control laws .
Label 1
Index 711
Tensorflow:
Sentence1 Others keep records sealed for as little as five years or as much as 30 .
Sentence2 Some states make them available immediately ; others keep them sealed for as much as 30 years .
Label -1
Torch:
Sentence1 Others keep records sealed for as little as five years or as much as 30 .
Sentence2 Some states make them available immediately ; others keep them sealed for as much as 30 years .
Label 0
```
## Expected results
I would expect the datasets to be independent of whether I am working with torch or tensorflow.
## Actual results
Test set labels are provided in the `datasets.load_datasets()` for MRPC. However MRPC is the only task where the test set labels are not -1.
## Environment info
- `datasets` version: 1.7.0
- Platform: Linux-5.4.109+-x86_64-with-Ubuntu-18.04-bionic
- Python version: 3.7.10
- PyArrow version: 3.0.0
| 2,452 |
https://github.com/huggingface/datasets/issues/2450 | BLUE file not found | [
"Hi ! The `blue` metric doesn't exist, but the `bleu` metric does.\r\nYou can get the full list of metrics [here](https://github.com/huggingface/datasets/tree/master/metrics) or by running\r\n```python\r\nfrom datasets import list_metrics\r\n\r\nprint(list_metrics())\r\n```",
"Ah, my mistake. Thanks for correctin... | Hi, I'm having the following issue when I try to load the `blue` metric.
```shell
import datasets
metric = datasets.load_metric('blue')
Traceback (most recent call last):
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/load.py", line 320, in prepare_module
local_path = cached_path(file_path, download_config=download_config)
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/utils/file_utils.py", line 291, in cached_path
use_auth_token=download_config.use_auth_token,
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/utils/file_utils.py", line 621, in get_from_cache
raise FileNotFoundError("Couldn't find file at {}".format(url))
FileNotFoundError: Couldn't find file at https://raw.githubusercontent.com/huggingface/datasets/1.7.0/metrics/blue/blue.py
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/load.py", line 332, in prepare_module
local_path = cached_path(file_path, download_config=download_config)
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/utils/file_utils.py", line 291, in cached_path
use_auth_token=download_config.use_auth_token,
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/utils/file_utils.py", line 621, in get_from_cache
raise FileNotFoundError("Couldn't find file at {}".format(url))
FileNotFoundError: Couldn't find file at https://raw.githubusercontent.com/huggingface/datasets/master/metrics/blue/blue.py
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "<input>", line 1, in <module>
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/load.py", line 605, in load_metric
dataset=False,
File "/home/irfan/environments/Perplexity_Transformers/lib/python3.6/site-packages/datasets/load.py", line 343, in prepare_module
combined_path, github_file_path
FileNotFoundError: Couldn't find file locally at blue/blue.py, or remotely at https://raw.githubusercontent.com/huggingface/datasets/1.7.0/metrics/blue/blue.py.
The file is also not present on the master branch on github.
```
Here is dataset installed version info
```shell
pip freeze | grep datasets
datasets==1.7.0
```
| 2,450 |
https://github.com/huggingface/datasets/issues/2447 | dataset adversarial_qa has no answers in the "test" set | [
"Hi ! I'm pretty sure that the answers are not made available for the test set on purpose because it is part of the DynaBench benchmark, for which you can submit your predictions on the website.\r\nIn any case we should mention this in the dataset card of this dataset.",
"Makes sense, but not intuitive for someon... | ## Describe the bug
When loading the adversarial_qa dataset the 'test' portion has no answers. Only the 'train' and 'validation' portions do. This occurs with all four of the configs ('adversarialQA', 'dbidaf', 'dbert', 'droberta')
## Steps to reproduce the bug
```
from datasets import load_dataset
examples = load_dataset('adversarial_qa', 'adversarialQA', script_version="master")['test']
print('Loaded {:,} examples'.format(len(examples)))
has_answers = 0
for e in examples:
if e['answers']['text']:
has_answers += 1
print('{:,} have answers'.format(has_answers))
>>> Loaded 3,000 examples
>>> 0 have answers
examples = load_dataset('adversarial_qa', 'adversarialQA', script_version="master")['validation']
<...code above...>
>>> Loaded 3,000 examples
>>> 3,000 have answers
```
## Expected results
If 'test' is a valid dataset, it should have answers. Also note that all of the 'train' and 'validation' sets have answers, there are no "no answer" questions with this set (not sure if this is correct or not).
## Environment info
- `datasets` version: 1.7.0
- Platform: Linux-5.8.0-53-generic-x86_64-with-glibc2.29
- Python version: 3.8.5
- PyArrow version: 1.0.0
| 2,447 |
https://github.com/huggingface/datasets/issues/2446 | `yelp_polarity` is broken | [
"```\r\nFile \"/home/sasha/.local/share/virtualenvs/lib-ogGKnCK_/lib/python3.7/site-packages/streamlit/script_runner.py\", line 332, in _run_script\r\n exec(code, module.__dict__)\r\nFile \"/home/sasha/nlp-viewer/run.py\", line 233, in <module>\r\n configs = get_confs(option)\r\nFile \"/home/sasha/.local/shar... | 
| 2,446 |
https://github.com/huggingface/datasets/issues/2444 | Sentence Boundaries missing in Dataset: xtreme / udpos | [
"Hi,\r\n\r\nThis is a known issue. More info on this issue can be found in #2061. If you are looking for an open-source contribution, there are step-by-step instructions in the linked issue that you can follow to fix it.",
"Closed by #2466."
] | I was browsing through annotation guidelines, as suggested by the datasets introduction.
The guidlines saids "There must be exactly one blank line after every sentence, including the last sentence in the file. Empty sentences are not allowed." in the [Sentence Boundaries and Comments section](https://universaldependencies.org/format.html#sentence-boundaries-and-comments)
But the sentence boundaries seems not to be represented by huggingface datasets features well. I found out that multiple sentence are concatenated together as a 1D array, without any delimiter.
PAN-x, which is another token classification subset from xtreme do represent the sentence boundary using a 2D array.
You may compare in PAN-x.en and udpos.English in the explorer:
https://huggingface.co/datasets/viewer/?dataset=xtreme | 2,444 |
https://github.com/huggingface/datasets/issues/2443 | Some tests hang on Windows | [
"Hi ! That would be nice indeed to at least have a warning, since we don't handle the max path length limit.\r\nAlso if we could have an error instead of an infinite loop I'm sure windows users will appreciate that",
"Unfortunately, I know this problem very well... 😅 \r\n\r\nI remember having proposed to throw a... | Currently, several tests hang on Windows if the max path limit of 260 characters is not disabled. This happens due to the changes introduced by #2223 that cause an infinite loop in `WindowsFileLock` described in #2220. This can be very tricky to debug, so I think now is a good time to address these issues/PRs. IMO throwing an error is too harsh, but maybe we can emit a warning in the top-level `__init__.py ` on startup if long paths are not enabled.
| 2,443 |
https://github.com/huggingface/datasets/issues/2441 | DuplicatedKeysError on personal dataset | [
"Hi ! In your dataset script you must be yielding examples like\r\n```python\r\nfor line in file:\r\n ...\r\n yield key, {...}\r\n```\r\n\r\nSince `datasets` 1.7.0 we enforce the keys to be unique.\r\nHowever it looks like your examples generator creates duplicate keys: at least two examples have key 0.\r\n\r... | ## Describe the bug
Ever since today, I have been getting a DuplicatedKeysError while trying to load my dataset from my own script.
Error returned when running this line: `dataset = load_dataset('/content/drive/MyDrive/Thesis/Datasets/book_preprocessing/goodreads_maharjan_trimmed_and_nered/goodreadsnered.py')`
Note that my script was working fine with earlier versions of the Datasets library. Cannot say with 100% certainty if I have been doing something wrong with my dataset script this whole time or if this is simply a bug with the new version of datasets.
## Steps to reproduce the bug
I cannot provide code to reproduce the error as I am working with my own dataset. I can however provide my script if requested.
## Expected results
For my data to be loaded.
## Actual results
**DuplicatedKeysError** exception is raised
```
Downloading and preparing dataset good_reads_practice_dataset/main_domain (download: Unknown size, generated: Unknown size, post-processed: Unknown size, total: Unknown size) to /root/.cache/huggingface/datasets/good_reads_practice_dataset/main_domain/1.1.0/64ff7c3fee2693afdddea75002eb6887d4fedc3d812ae3622128c8504ab21655...
---------------------------------------------------------------------------
DuplicatedKeysError Traceback (most recent call last)
<ipython-input-6-c342ea0dae9d> in <module>()
----> 1 dataset = load_dataset('/content/drive/MyDrive/Thesis/Datasets/book_preprocessing/goodreads_maharjan_trimmed_and_nered/goodreadsnered.py')
5 frames
/usr/local/lib/python3.7/dist-packages/datasets/load.py in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, ignore_verifications, keep_in_memory, save_infos, script_version, use_auth_token, task, **config_kwargs)
749 try_from_hf_gcs=try_from_hf_gcs,
750 base_path=base_path,
--> 751 use_auth_token=use_auth_token,
752 )
753
/usr/local/lib/python3.7/dist-packages/datasets/builder.py in download_and_prepare(self, download_config, download_mode, ignore_verifications, try_from_hf_gcs, dl_manager, base_path, use_auth_token, **download_and_prepare_kwargs)
573 if not downloaded_from_gcs:
574 self._download_and_prepare(
--> 575 dl_manager=dl_manager, verify_infos=verify_infos, **download_and_prepare_kwargs
576 )
577 # Sync info
/usr/local/lib/python3.7/dist-packages/datasets/builder.py in _download_and_prepare(self, dl_manager, verify_infos, **prepare_split_kwargs)
650 try:
651 # Prepare split will record examples associated to the split
--> 652 self._prepare_split(split_generator, **prepare_split_kwargs)
653 except OSError as e:
654 raise OSError(
/usr/local/lib/python3.7/dist-packages/datasets/builder.py in _prepare_split(self, split_generator)
990 writer.write(example, key)
991 finally:
--> 992 num_examples, num_bytes = writer.finalize()
993
994 split_generator.split_info.num_examples = num_examples
/usr/local/lib/python3.7/dist-packages/datasets/arrow_writer.py in finalize(self, close_stream)
407 # In case current_examples < writer_batch_size, but user uses finalize()
408 if self._check_duplicates:
--> 409 self.check_duplicate_keys()
410 # Re-intializing to empty list for next batch
411 self.hkey_record = []
/usr/local/lib/python3.7/dist-packages/datasets/arrow_writer.py in check_duplicate_keys(self)
347 for hash, key in self.hkey_record:
348 if hash in tmp_record:
--> 349 raise DuplicatedKeysError(key)
350 else:
351 tmp_record.add(hash)
DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 0
Keys should be unique and deterministic in nature
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.7.0
- Platform: Windows-10-10.0.19041-SP0
- Python version: 3.7.9
- PyArrow version: 3.0.0
| 2,441 |
https://github.com/huggingface/datasets/issues/2440 | Remove `extended` field from dataset tagger | [
"The tagger also doesn't insert the value for the `size_categories` field automatically, so this should be fixed too",
"Thanks for reporting. Indeed the `extended` tag doesn't exist. Not sure why we had that in the tagger.\r\nThe repo of the tagger is here if someone wants to give this a try: https://github.com/h... | ## Describe the bug
While working on #2435 I used the [dataset tagger](https://huggingface.co/datasets/tagging/) to generate the missing tags for the YAML metadata of each README.md file. However, it seems that our CI raises an error when the `extended` field is included:
```
dataset_name = 'arcd'
@pytest.mark.parametrize("dataset_name", get_changed_datasets(repo_path))
def test_changed_dataset_card(dataset_name):
card_path = repo_path / "datasets" / dataset_name / "README.md"
assert card_path.exists()
error_messages = []
try:
ReadMe.from_readme(card_path)
except Exception as readme_error:
error_messages.append(f"The following issues have been found in the dataset cards:\nREADME:\n{readme_error}")
try:
DatasetMetadata.from_readme(card_path)
except Exception as metadata_error:
error_messages.append(
f"The following issues have been found in the dataset cards:\nYAML tags:\n{metadata_error}"
)
if error_messages:
> raise ValueError("\n".join(error_messages))
E ValueError: The following issues have been found in the dataset cards:
E YAML tags:
E __init__() got an unexpected keyword argument 'extended'
tests/test_dataset_cards.py:70: ValueError
```
Consider either removing this tag from the tagger or including it as part of the validation step in the CI.
cc @yjernite | 2,440 |
https://github.com/huggingface/datasets/issues/2434 | Extend QuestionAnsweringExtractive template to handle nested columns | [
"this is also the case for the following datasets and configurations:\r\n\r\n* `mlqa` with config `mlqa-translate-train.ar`\r\n\r\n",
"The current task API is somewhat deprecated (we plan to align it with `train eval index` at some point), so I think we can close this issue."
] | Currently the `QuestionAnsweringExtractive` task template and `preprare_for_task` only support "flat" features. We should extend the functionality to cover QA datasets like:
* `iapp_wiki_qa_squad`
* `parsinlu_reading_comprehension`
where the nested features differ with those from `squad` and trigger an `ArrowNotImplementedError`:
```
---------------------------------------------------------------------------
ArrowNotImplementedError Traceback (most recent call last)
<ipython-input-12-50e5b8f69c20> in <module>
----> 1 ds.prepare_for_task("question-answering-extractive")[0]
~/git/datasets/src/datasets/arrow_dataset.py in prepare_for_task(self, task)
1436 # We found a template so now flush `DatasetInfo` to skip the template update in `DatasetInfo.__post_init__`
1437 dataset.info.task_templates = None
-> 1438 dataset = dataset.cast(features=template.features)
1439 return dataset
1440
~/git/datasets/src/datasets/arrow_dataset.py in cast(self, features, batch_size, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, num_proc)
977 format = self.format
978 dataset = self.with_format("arrow")
--> 979 dataset = dataset.map(
980 lambda t: t.cast(schema),
981 batched=True,
~/git/datasets/src/datasets/arrow_dataset.py in map(self, function, with_indices, input_columns, batched, batch_size, drop_last_batch, remove_columns, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, features, disable_nullable, fn_kwargs, num_proc, suffix_template, new_fingerprint, desc)
1600
1601 if num_proc is None or num_proc == 1:
-> 1602 return self._map_single(
1603 function=function,
1604 with_indices=with_indices,
~/git/datasets/src/datasets/arrow_dataset.py in wrapper(*args, **kwargs)
176 }
177 # apply actual function
--> 178 out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs)
179 datasets: List["Dataset"] = list(out.values()) if isinstance(out, dict) else [out]
180 # re-apply format to the output
~/git/datasets/src/datasets/fingerprint.py in wrapper(*args, **kwargs)
395 # Call actual function
396
--> 397 out = func(self, *args, **kwargs)
398
399 # Update fingerprint of in-place transforms + update in-place history of transforms
~/git/datasets/src/datasets/arrow_dataset.py in _map_single(self, function, with_indices, input_columns, batched, batch_size, drop_last_batch, remove_columns, keep_in_memory, load_from_cache_file, cache_file_name, writer_batch_size, features, disable_nullable, fn_kwargs, new_fingerprint, rank, offset, desc)
1940 ) # Something simpler?
1941 try:
-> 1942 batch = apply_function_on_filtered_inputs(
1943 batch,
1944 indices,
~/git/datasets/src/datasets/arrow_dataset.py in apply_function_on_filtered_inputs(inputs, indices, check_same_num_examples, offset)
1836 effective_indices = [i + offset for i in indices] if isinstance(indices, list) else indices + offset
1837 processed_inputs = (
-> 1838 function(*fn_args, effective_indices, **fn_kwargs) if with_indices else function(*fn_args, **fn_kwargs)
1839 )
1840 if update_data is None:
~/git/datasets/src/datasets/arrow_dataset.py in <lambda>(t)
978 dataset = self.with_format("arrow")
979 dataset = dataset.map(
--> 980 lambda t: t.cast(schema),
981 batched=True,
982 batch_size=batch_size,
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/table.pxi in pyarrow.lib.Table.cast()
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/table.pxi in pyarrow.lib.ChunkedArray.cast()
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/compute.py in cast(arr, target_type, safe)
241 else:
242 options = CastOptions.unsafe(target_type)
--> 243 return call_function("cast", [arr], options)
244
245
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/_compute.pyx in pyarrow._compute.call_function()
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/_compute.pyx in pyarrow._compute.Function.call()
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/error.pxi in pyarrow.lib.pyarrow_internal_check_status()
~/miniconda3/envs/datasets/lib/python3.8/site-packages/pyarrow/error.pxi in pyarrow.lib.check_status()
ArrowNotImplementedError: Unsupported cast from struct<answer_end: list<item: int32>, answer_start: list<item: int32>, text: list<item: string>> to struct using function cast_struct
``` | 2,434 |
https://github.com/huggingface/datasets/issues/2431 | DuplicatedKeysError when trying to load adversarial_qa | [
"Thanks for reporting !\r\n#2433 fixed the issue, thanks @mariosasko :)\r\n\r\nWe'll do a patch release soon of the library.\r\nIn the meantime, you can use the fixed version of adversarial_qa by adding `script_version=\"master\"` in `load_dataset`"
] | ## Describe the bug
A clear and concise description of what the bug is.
## Steps to reproduce the bug
```python
dataset = load_dataset('adversarial_qa', 'adversarialQA')
```
## Expected results
The dataset should be loaded into memory
## Actual results
>DuplicatedKeysError: FAILURE TO GENERATE DATASET !
>Found duplicate Key: 4d3cb5677211ee32895ca9c66dad04d7152254d4
>Keys should be unique and deterministic in nature
>
>
>During handling of the above exception, another exception occurred:
>
>DuplicatedKeysError Traceback (most recent call last)
>
>/usr/local/lib/python3.7/dist-packages/datasets/arrow_writer.py in check_duplicate_keys(self)
> 347 for hash, key in self.hkey_record:
> 348 if hash in tmp_record:
>--> 349 raise DuplicatedKeysError(key)
> 350 else:
> 351 tmp_record.add(hash)
>
>DuplicatedKeysError: FAILURE TO GENERATE DATASET !
>Found duplicate Key: 4d3cb5677211ee32895ca9c66dad04d7152254d4
>Keys should be unique and deterministic in nature
## Environment info
- `datasets` version: 1.7.0
- Platform: Linux-5.4.109+-x86_64-with-Ubuntu-18.04-bionic
- Python version: 3.7.10
- PyArrow version: 3.0.0
| 2,431 |
https://github.com/huggingface/datasets/issues/2426 | Saving Graph/Structured Data in Datasets | [
"It should probably work out of the box to save structured data. If you want to show an example we can help you.",
"An example of a toy dataset is like:\r\n```json\r\n[\r\n {\r\n \"name\": \"mike\",\r\n \"friends\": [\r\n \"tom\",\r\n \"lily\"\r\n ],\r\n \"arti... | Thanks for this amazing library! And my question is I have structured data that is organized with a graph. For example, a dataset with users' friendship relations and user's articles. When I try to save a python dict in the dataset, an error occurred ``did not recognize Python value type when inferring an Arrow data type''.
Although I also know that storing a python dict in pyarrow datasets is not the best practice, but I have no idea about how to save structured data in the Datasets.
Thank you very much for your help. | 2,426 |
https://github.com/huggingface/datasets/issues/2424 | load_from_disk and save_to_disk are not compatible with each other | [
"Hi,\r\n\r\n`load_dataset` returns an instance of `DatasetDict` if `split` is not specified, so instead of `Dataset.load_from_disk`, use `DatasetDict.load_from_disk` to load the dataset from disk.",
"Thanks it worked!",
"Though I see a stream of issues open by people lost between datasets and datasets dicts so ... | ## Describe the bug
load_from_disk and save_to_disk are not compatible. When I use save_to_disk to save a dataset to disk it works perfectly but given the same directory load_from_disk throws an error that it can't find state.json. looks like the load_from_disk only works on one split
## Steps to reproduce the bug
```python
from datasets import load_dataset
dataset = load_dataset("art")
dataset.save_to_disk("mydir")
d = Dataset.load_from_disk("mydir")
```
## Expected results
It is expected that these two functions be the reverse of each other without more manipulation
## Actual results
FileNotFoundError: [Errno 2] No such file or directory: 'mydir/art/state.json'
## Environment info
- `datasets` version: 1.6.2
- Platform: Linux-5.4.0-73-generic-x86_64-with-Ubuntu-18.04-bionic
- Python version: 3.7.10
- PyTorch version (GPU?): 1.8.1+cu102 (True)
- Tensorflow version (GPU?): not installed (NA)
- Using GPU in script?: <fill in>
- Using distributed or parallel set-up in script?: <fill in>
| 2,424 |
https://github.com/huggingface/datasets/issues/2415 | Cached dataset not loaded | [
"It actually seems to happen all the time in above configuration:\r\n* the function `filter_by_duration` correctly loads cached processed dataset\r\n* the function `prepare_dataset` is always reexecuted\r\n\r\nI end up solving the issue by saving to disk my dataset at the end but I'm still wondering if it's a bug o... | ## Describe the bug
I have a large dataset (common_voice, english) where I use several map and filter functions.
Sometimes my cached datasets after specific functions are not loaded.
I always use the same arguments, same functions, no seed…
## Steps to reproduce the bug
```python
def filter_by_duration(batch):
return (
batch["duration"] <= 10
and batch["duration"] >= 1
and len(batch["target_text"]) > 5
)
def prepare_dataset(batch):
batch["input_values"] = processor(
batch["speech"], sampling_rate=batch["sampling_rate"][0]
).input_values
with processor.as_target_processor():
batch["labels"] = processor(batch["target_text"]).input_ids
return batch
train_dataset = train_dataset.filter(
filter_by_duration,
remove_columns=["duration"],
num_proc=data_args.preprocessing_num_workers,
)
# PROBLEM HERE -> below function is reexecuted and cache is not loaded
train_dataset = train_dataset.map(
prepare_dataset,
remove_columns=train_dataset.column_names,
batch_size=training_args.per_device_train_batch_size,
batched=True,
num_proc=data_args.preprocessing_num_workers,
)
# Later in script
set_caching_enabled(False)
# apply map on trained model to eval/test sets
```
## Expected results
The cached dataset should always be reloaded.
## Actual results
The function is reexecuted.
I have access to cached files `cache-xxxxx.arrow`.
Is there a way I can somehow load manually 2 versions and see how the hash was created for debug purposes (to know if it's an issue with dataset or function)?
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.6.2
- Platform: Linux-5.8.0-45-generic-x86_64-with-glibc2.29
- Python version: 3.8.5
- PyTorch version (GPU?): 1.8.1+cu102 (True)
- Tensorflow version (GPU?): not installed (NA)
- Using GPU in script?: Yes
- Using distributed or parallel set-up in script?: No | 2,415 |
https://github.com/huggingface/datasets/issues/2413 | AttributeError: 'DatasetInfo' object has no attribute 'task_templates' | [
"Hi ! Can you try using a more up-to-date version ? We added the task_templates in `datasets` 1.7.0.\r\n\r\nIdeally when you're working on new datasets, you should install and use the local version of your fork of `datasets`. Here I think you tried to run the 1.7.0 tests with the 1.6.2 code"
] | ## Describe the bug
Hello,
I'm trying to add dataset and contribute, but test keep fail with below cli.
` RUN_SLOW=1 pytest tests/test_dataset_common.py::LocalDatasetTest::test_load_dataset_all_configs_<my_dataset>`
## Steps to reproduce the bug
It seems like a bug when I see an error with the existing dataset, not the dataset I'm trying to add.
` RUN_SLOW=1 pytest tests/test_dataset_common.py::LocalDatasetTest::test_load_dataset_all_configs_<any_dataset>`
## Expected results
All test passed
## Actual results
```
# check that dataset is not empty
self.parent.assertListEqual(sorted(dataset_builder.info.splits.keys()), sorted(dataset))
for split in dataset_builder.info.splits.keys():
# check that loaded datset is not empty
self.parent.assertTrue(len(dataset[split]) > 0)
# check that we can cast features for each task template
> task_templates = dataset_builder.info.task_templates
E AttributeError: 'DatasetInfo' object has no attribute 'task_templates'
tests/test_dataset_common.py:175: AttributeError
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.6.2
- Platform: Darwin-20.4.0-x86_64-i386-64bit
- Python version: 3.7.7
- PyTorch version (GPU?): 1.7.0 (False)
- Tensorflow version (GPU?): 2.3.0 (False)
- Using GPU in script?: No
- Using distributed or parallel set-up in script?: No
| 2,413 |
https://github.com/huggingface/datasets/issues/2412 | Docstring mistake: dataset vs. metric | [
"> I can provide a PR l8er...\r\n\r\nSee #2425 "
] | This:
https://github.com/huggingface/datasets/blob/d95b95f8cf3cb0cff5f77a675139b584dcfcf719/src/datasets/load.py#L582
Should better be something like:
`a metric identifier on HuggingFace AWS bucket (list all available metrics and ids with ``datasets.list_metrics()``)`
I can provide a PR l8er... | 2,412 |
https://github.com/huggingface/datasets/issues/2407 | .map() function got an unexpected keyword argument 'cache_file_name' | [
"Hi @cindyxinyiwang,\r\nDid you try adding `.arrow` after `cache_file_name` argument? Here I think they're expecting something like that only for a cache file:\r\nhttps://github.com/huggingface/datasets/blob/e08362256fb157c0b3038437fc0d7a0bbb50de5c/src/datasets/arrow_dataset.py#L1556-L1558",
"Hi ! `cache_file_nam... | ## Describe the bug
I'm trying to save the result of datasets.map() to a specific file, so that I can easily share it among multiple computers without reprocessing the dataset. However, when I try to pass an argument 'cache_file_name' to the .map() function, it throws an error that ".map() function got an unexpected keyword argument 'cache_file_name'".
I believe I'm using the latest dataset 1.6.2. Also seems like the document and the actual code indicates there is an argument 'cache_file_name' for the .map() function.
Here is the code I use
## Steps to reproduce the bug
```datasets = load_from_disk(dataset_path=my_path)
[...]
def tokenize_function(examples):
return tokenizer(examples[text_column_name])
logger.info("Mapping dataset to tokenized dataset.")
tokenized_datasets = datasets.map(
tokenize_function,
batched=True,
num_proc=preprocessing_num_workers,
remove_columns=column_names,
load_from_cache_file=True,
cache_file_name="my_tokenized_file"
)
```
## Actual results
tokenized_datasets = datasets.map(
TypeError: map() got an unexpected keyword argument 'cache_file_name'
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version:1.6.2
- Platform:Linux-4.18.0-193.28.1.el8_2.x86_64-x86_64-with-glibc2.10
- Python version:3.8.5
- PyArrow version:3.0.0
| 2,407 |
https://github.com/huggingface/datasets/issues/2406 | Add guide on using task templates to documentation | [] | Once we have a stable API on the text classification and question answering task templates, add a guide on how to use them in the documentation.
| 2,406 |
https://github.com/huggingface/datasets/issues/2402 | PermissionError on Windows when using temp dir for caching | [] | Currently, the following code raises a PermissionError on master if working on Windows:
```python
# run as a script or call exit() in REPL to initiate the temp dir cleanup
from datasets import *
d = load_dataset("sst", split="train", keep_in_memory=False)
set_caching_enabled(False)
d.map(lambda ex: ex)
```
Error stack trace:
```
Traceback (most recent call last):
File "C:\Users\Mario\Anaconda3\envs\hf-datasets\lib\weakref.py", line 624, in _exitfunc
f()
File "C:\Users\Mario\Anaconda3\envs\hf-datasets\lib\weakref.py", line 548, in __call__
return info.func(*info.args, **(info.kwargs or {}))
File "C:\Users\Mario\Anaconda3\envs\hf-datasets\lib\tempfile.py", line 799, in _cleanup
_shutil.rmtree(name)
File "C:\Users\Mario\Anaconda3\envs\hf-datasets\lib\shutil.py", line 500, in rmtree
return _rmtree_unsafe(path, onerror)
File "C:\Users\Mario\Anaconda3\envs\hf-datasets\lib\shutil.py", line 395, in _rmtree_unsafe
onerror(os.unlink, fullname, sys.exc_info())
File "C:\Users\Mario\Anaconda3\envs\hf-datasets\lib\shutil.py", line 393, in _rmtree_unsafe
os.unlink(fullname)
PermissionError: [WinError 5] Access is denied: 'C:\\Users\\Mario\\AppData\\Local\\Temp\\tmp20epyhmq\\cache-87a87ffb5a956e68.arrow'
``` | 2,402 |
https://github.com/huggingface/datasets/issues/2401 | load_dataset('natural_questions') fails with "ValueError: External features info don't match the dataset" | [
"I faced the similar problem. Downgrading datasets to 1.5.0 fixed it.",
"Thanks for reporting, I'm looking into it",
"I just opened #2438 to fix this :)",
"Hi ! This has been fixed in the 1.8.0 release of `datasets`"
] | ## Describe the bug
load_dataset('natural_questions') throws ValueError
## Steps to reproduce the bug
```python
from datasets import load_dataset
datasets = load_dataset('natural_questions', split='validation[:10]')
```
## Expected results
Call to load_dataset returns data.
## Actual results
```
Using custom data configuration default
Reusing dataset natural_questions (/mnt/d/huggingface/datasets/natural_questions/default/0.0.2/19bc04755018a3ad02ee74f7045cde4ba9b4162cb64450a87030ab786b123b76)
---------------------------------------------------------------------------
ValueError Traceback (most recent call last)
<ipython-input-2-d55ab8a8cc1c> in <module>
----> 1 datasets = load_dataset('natural_questions', split='validation[:10]', cache_dir='/mnt/d/huggingface/datasets')
~/miniconda3/lib/python3.8/site-packages/datasets/load.py in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, ignore_verifications, keep_in_memory, save_infos, script_version, use_auth_token, **config_kwargs)
756 keep_in_memory if keep_in_memory is not None else is_small_dataset(builder_instance.info.dataset_size)
757 )
--> 758 ds = builder_instance.as_dataset(split=split, ignore_verifications=ignore_verifications, in_memory=keep_in_memory)
759 if save_infos:
760 builder_instance._save_infos()
~/miniconda3/lib/python3.8/site-packages/datasets/builder.py in as_dataset(self, split, run_post_process, ignore_verifications, in_memory)
735
736 # Create a dataset for each of the given splits
--> 737 datasets = utils.map_nested(
738 partial(
739 self._build_single_dataset,
~/miniconda3/lib/python3.8/site-packages/datasets/utils/py_utils.py in map_nested(function, data_struct, dict_only, map_list, map_tuple, map_numpy, num_proc, types)
193 # Singleton
194 if not isinstance(data_struct, dict) and not isinstance(data_struct, types):
--> 195 return function(data_struct)
196
197 disable_tqdm = bool(logger.getEffectiveLevel() > INFO)
~/miniconda3/lib/python3.8/site-packages/datasets/builder.py in _build_single_dataset(self, split, run_post_process, ignore_verifications, in_memory)
762
763 # Build base dataset
--> 764 ds = self._as_dataset(
765 split=split,
766 in_memory=in_memory,
~/miniconda3/lib/python3.8/site-packages/datasets/builder.py in _as_dataset(self, split, in_memory)
838 in_memory=in_memory,
839 )
--> 840 return Dataset(**dataset_kwargs)
841
842 def _post_process(self, dataset: Dataset, resources_paths: Dict[str, str]) -> Optional[Dataset]:
~/miniconda3/lib/python3.8/site-packages/datasets/arrow_dataset.py in __init__(self, arrow_table, info, split, indices_table, fingerprint)
271 assert self._fingerprint is not None, "Fingerprint can't be None in a Dataset object"
272 if self.info.features.type != inferred_features.type:
--> 273 raise ValueError(
274 "External features info don't match the dataset:\nGot\n{}\nwith type\n{}\n\nbut expected something like\n{}\nwith type\n{}".format(
275 self.info.features, self.info.features.type, inferred_features, inferred_features.type
ValueError: External features info don't match the dataset:
Got
{'id': Value(dtype='string', id=None), 'document': {'title': Value(dtype='string', id=None), 'url': Value(dtype='string', id=None), 'html': Value(dtype='string', id=None), 'tokens': Sequence(feature={'token': Value(dtype='string', id=None), 'is_html': Value(dtype='bool', id=None)}, length=-1, id=None)}, 'question': {'text': Value(dtype='string', id=None), 'tokens': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}, 'annotations': Sequence(feature={'id': Value(dtype='string', id=None), 'long_answer': {'start_token': Value(dtype='int64', id=None), 'end_token': Value(dtype='int64', id=None), 'start_byte': Value(dtype='int64', id=None), 'end_byte': Value(dtype='int64', id=None)}, 'short_answers': Sequence(feature={'start_token': Value(dtype='int64', id=None), 'end_token': Value(dtype='int64', id=None), 'start_byte': Value(dtype='int64', id=None), 'end_byte': Value(dtype='int64', id=None), 'text': Value(dtype='string', id=None)}, length=-1, id=None), 'yes_no_answer': ClassLabel(num_classes=2, names=['NO', 'YES'], names_file=None, id=None)}, length=-1, id=None)}
with type
struct<annotations: struct<id: list<item: string>, long_answer: list<item: struct<start_token: int64, end_token: int64, start_byte: int64, end_byte: int64>>, short_answers: list<item: struct<end_byte: list<item: int64>, end_token: list<item: int64>, start_byte: list<item: int64>, start_token: list<item: int64>, text: list<item: string>>>, yes_no_answer: list<item: int64>>, document: struct<title: string, url: string, html: string, tokens: struct<is_html: list<item: bool>, token: list<item: string>>>, id: string, question: struct<text: string, tokens: list<item: string>>>
but expected something like
{'id': Value(dtype='string', id=None), 'document': {'html': Value(dtype='string', id=None), 'title': Value(dtype='string', id=None), 'tokens': {'is_html': Sequence(feature=Value(dtype='bool', id=None), length=-1, id=None), 'token': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}, 'url': Value(dtype='string', id=None)}, 'question': {'text': Value(dtype='string', id=None), 'tokens': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}, 'annotations': {'id': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'long_answer': [{'end_byte': Value(dtype='int64', id=None), 'end_token': Value(dtype='int64', id=None), 'start_byte': Value(dtype='int64', id=None), 'start_token': Value(dtype='int64', id=None)}], 'short_answers': [{'end_byte': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'end_token': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'start_byte': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'start_token': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'text': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}], 'yes_no_answer': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None)}}
with type
struct<annotations: struct<id: list<item: string>, long_answer: list<item: struct<end_byte: int64, end_token: int64, start_byte: int64, start_token: int64>>, short_answers: list<item: struct<end_byte: list<item: int64>, end_token: list<item: int64>, start_byte: list<item: int64>, start_token: list<item: int64>, text: list<item: string>>>, yes_no_answer: list<item: int64>>, document: struct<html: string, title: string, tokens: struct<is_html: list<item: bool>, token: list<item: string>>, url: string>, id: string, question: struct<text: string, tokens: list<item: string>>>
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.6.2
- Platform: Linux-5.4.72-microsoft-standard-WSL2-x86_64-with-glibc2.10
- Python version: 3.8.3
- PyTorch version (GPU?): 1.6.0 (False)
- Tensorflow version (GPU?): not installed (NA)
- Using GPU in script?: No
- Using distributed or parallel set-up in script?: No
| 2,401 |
https://github.com/huggingface/datasets/issues/2400 | Concatenate several datasets with removed columns is not working. | [
"Hi,\r\n\r\ndid you fill out the env info section manually or by copy-pasting the output of the `datasets-cli env` command?\r\n\r\nThis code should work without issues on 1.6.2 version (I'm working on master (1.6.2.dev0 version) and can't reproduce this error).",
"@mariosasko you are right I was still on `1.5.0`.... | ## Describe the bug
You can't concatenate datasets when you removed columns before.
## Steps to reproduce the bug
```python
from datasets import load_dataset, concatenate_datasets
wikiann= load_dataset("wikiann","en")
wikiann["train"] = wikiann["train"].remove_columns(["langs","spans"])
wikiann["test"] = wikiann["test"].remove_columns(["langs","spans"])
assert wikiann["train"].features.type == wikiann["test"].features.type
concate = concatenate_datasets([wikiann["train"],wikiann["test"]])
```
## Expected results
Merged dataset
## Actual results
```python
ValueError: External features info don't match the dataset:
Got
{'tokens': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'ner_tags': Sequence(feature=ClassLabel(num_classes=7, names=['O', 'B-PER', 'I-PER', 'B-ORG', 'I-ORG', 'B-LOC', 'I-LOC'], names_file=None, id=None), length=-1, id=None), 'langs': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'spans': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}
with type
struct<langs: list<item: string>, ner_tags: list<item: int64>, spans: list<item: string>, tokens: list<item: string>>
but expected something like
{'ner_tags': Sequence(feature=Value(dtype='int64', id=None), length=-1, id=None), 'tokens': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}
with type
struct<ner_tags: list<item: int64>, tokens: list<item: string>>
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: ~1.6.2~ 1.5.0
- Platform: macos
- Python version: 3.8.5
- PyArrow version: 3.0.0
| 2,400 |
https://github.com/huggingface/datasets/issues/2398 | News_commentary Dataset Translation Pairs are of Incorrect Language Specified Pairs | [
"These ranges seem to be valid English. Closing."
] | I used load_dataset to load the news_commentary dataset for "ar-en" translation pairs but found translations from Arabic to Hindi.
```
train_ds = load_dataset("news_commentary", "ar-en", split='train[:98%]')
val_ds = load_dataset("news_commentary", "ar-en", split='train[98%:]')
# filtering out examples that are not ar-en translations but ar-hi
val_ds = val_ds.filter(lambda example, indice: indice not in chain(range(1312,1327) ,range(1384,1399), range(1030,1042)), with_indices=True)
```
* I'm fairly new to using datasets so I might be doing something wrong | 2,398 |
https://github.com/huggingface/datasets/issues/2396 | strange datasets from OSCAR corpus | [
"Hi ! Thanks for reporting\r\ncc @pjox is this an issue from the data ?\r\n\r\nAnyway we should at least mention that OSCAR could contain such contents in the dataset card, you're totally right @jerryIsHere ",
"Hi @jerryIsHere , sorry for the late response! Sadly this is normal, the problem comes form fasttext's ... | 
From the [official site ](https://oscar-corpus.com/), the Yue Chinese dataset should have 2.2KB data.
7 training instances is obviously not a right number.
As I can read Yue Chinese, I call tell the last instance is definitely not something that would appear on Common Crawl.
And even if you don't read Yue Chinese, you can tell the first six instance are problematic.
(It is embarrassing, as the 7 training instances look exactly like something from a pornographic novel or flitting messages in a chat of a dating app)
It might not be the problem of the huggingface/datasets implementation, because when I tried to download the dataset from the official site, I found out that the zip file is corrupted.
I will try to inform the host of OSCAR corpus later.
Awy a remake about this dataset in huggingface/datasets is needed, perhaps after the host of the dataset fixes the issue.
> Hi @jerryIsHere , sorry for the late response! Sadly this is normal, the problem comes form fasttext's classifier which we used to create the original corpus. In general the classifier is not really capable of properly recognizing Yue Chineese so the file ends un being just noise from Common Crawl. Some of these problems with OSCAR were already discussed [here](https://arxiv.org/pdf/2103.12028.pdf) but we are working on explicitly documenting the problems by language on our website. In fact, could please you open an issue on [our repo](https://github.com/oscar-corpus/oscar-website/issues) as well so that we can track it?
Thanks a lot, the new post is here:
https://github.com/oscar-corpus/oscar-website/issues/11 | 2,396 |
https://github.com/huggingface/datasets/issues/2391 | Missing original answers in kilt-TriviaQA | [
"That could be useful indeed! Feel free to open a PR on the dataset card if you already have some code that runs, otherwise we'll take care of it soon :) ",
"I can open a PR but there is 2 details to fix:\r\n- the name for the corresponding key (e.g. `original_answer`)\r\n- how to implement it: I’m not sure what ... | I previously opened an issue at https://github.com/facebookresearch/KILT/issues/42 but from the answer of @fabiopetroni it seems that the problem comes from HF-datasets
## Describe the bug
The `answer` field in kilt-TriviaQA, e.g. `kilt_tasks['train_triviaqa'][0]['output']['answer']` contains a list of alternative answer which are accepted for the question.
However it'd be nice to know the original answer to the question (the only fields in `output` are `'answer', 'meta', 'provenance'`)
## How to fix
It can be fixed by retrieving the original answer from the original TriviaQA (e.g. `trivia_qa['train'][0]['answer']['value']`), perhaps at the same place as here where one retrieves the questions https://github.com/huggingface/datasets/blob/master/datasets/kilt_tasks/README.md#loading-the-kilt-knowledge-source-and-task-data
cc @yjernite who previously answered to an issue about KILT and TriviaQA :)
| 2,391 |
https://github.com/huggingface/datasets/issues/2388 | Incorrect URLs for some datasets | [] | ## Describe the bug
It seems that the URLs for the following datasets are invalid:
- [ ] `bn_hate_speech` has been renamed: https://github.com/rezacsedu/Bengali-Hate-Speech-Dataset/commit/c67ecfc4184911e12814f6b36901f9828df8a63a
- [ ] `covid_tweets_japanese` has been renamed: http://www.db.info.gifu-u.ac.jp/covid-19-twitter-dataset/
As a result we can no longer load these datasets using `load_dataset`. The simple fix is to rename the URL in the dataset script - will do this asap.
## Steps to reproduce the bug
```python
from datasets import load_dataset
# pick one of the datasets from the list above
ds = load_dataset("bn_hate_speech")
```
## Expected results
Dataset loads without error.
## Actual results
```
Downloading: 3.36kB [00:00, 1.07MB/s]
Downloading: 2.03kB [00:00, 678kB/s]
Using custom data configuration default
Downloading and preparing dataset bn_hate_speech/default (download: 951.48 KiB, generated: 949.84 KiB, post-processed: Unknown size, total: 1.86 MiB) to /Users/lewtun/.cache/huggingface/datasets/bn_hate_speech/default/0.0.0/a2dc726e511a2177523301bcad196af05d4d8a2cff30d2769ba8aacc1f5fdb5c...
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/load.py", line 744, in load_dataset
builder_instance.download_and_prepare(
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/builder.py", line 574, in download_and_prepare
self._download_and_prepare(
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/builder.py", line 630, in _download_and_prepare
split_generators = self._split_generators(dl_manager, **split_generators_kwargs)
File "/Users/lewtun/.cache/huggingface/modules/datasets_modules/datasets/bn_hate_speech/a2dc726e511a2177523301bcad196af05d4d8a2cff30d2769ba8aacc1f5fdb5c/bn_hate_speech.py", line 76, in _split_generators
train_path = dl_manager.download_and_extract(_URL)
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/utils/download_manager.py", line 287, in download_and_extract
return self.extract(self.download(url_or_urls))
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/utils/download_manager.py", line 195, in download
downloaded_path_or_paths = map_nested(
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/utils/py_utils.py", line 195, in map_nested
return function(data_struct)
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/utils/download_manager.py", line 218, in _download
return cached_path(url_or_filename, download_config=download_config)
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/utils/file_utils.py", line 281, in cached_path
output_path = get_from_cache(
File "/Users/lewtun/miniconda3/envs/hf-hub_eval/lib/python3.8/site-packages/datasets/utils/file_utils.py", line 621, in get_from_cache
raise FileNotFoundError("Couldn't find file at {}".format(url))
FileNotFoundError: Couldn't find file at https://raw.githubusercontent.com/rezacsedu/Bengali-Hate-Speech-Dataset/main/Bengali_%20Hate_Speech_Dataset_Subset.csv
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: 1.6.2.dev0
- Platform: macOS-10.16-x86_64-i386-64bit
- Python version: 3.8.8
- PyArrow version: 3.0.0
| 2,388 |
https://github.com/huggingface/datasets/issues/2387 | datasets 1.6 ignores cache | [
"Looks like there are multiple issues regarding this (#2386, #2322) and it's a WIP #2329. Currently these datasets are being loaded in-memory which is causing this issue. Quoting @mariosasko here for a quick fix:\r\n\r\n> set `keep_in_memory` to `False` when loading a dataset (`sst = load_dataset(\"sst\", keep_in_m... | Moving from https://github.com/huggingface/transformers/issues/11801#issuecomment-845546612
Quoting @VictorSanh:
>
> I downgraded datasets to `1.5.0` and printed `tokenized_datasets.cache_files` (L335):
>
> > `{'train': [{'filename': '/home/victor/.cache/huggingface/datasets/openwebtext10k/plain_text/1.0.0/3a8df094c671b4cb63ed0b41f40fb3bd855e9ce2e3765e5df50abcdfb5ec144b/cache-c6aefe81ca4e5152.arrow'}], 'validation': [{'filename': '/home/victor/.cache/huggingface/datasets/openwebtext10k/plain_text/1.0.0/3a8df094c671b4cb63ed0b41f40fb3bd855e9ce2e3765e5df50abcdfb5ec144b/cache-97cf4c813e6469c6.arrow'}]}`
>
> while the same command with the latest version of datasets (actually starting at `1.6.0`) gives:
> > `{'train': [], 'validation': []}`
>
I also confirm that downgrading to `datasets==1.5.0` makes things fast again - i.e. cache is used.
to reproduce:
```
USE_TF=0 python examples/pytorch/language-modeling/run_clm.py \
--model_name_or_path gpt2 \
--dataset_name "stas/openwebtext-10k" \
--output_dir output_dir \
--overwrite_output_dir \
--do_train \
--do_eval \
--max_train_samples 1000 \
--max_eval_samples 200 \
--per_device_train_batch_size 4 \
--per_device_eval_batch_size 4 \
--num_train_epochs 1 \
--warmup_steps 8 \
--block_size 64 \
--fp16 \
--report_to none
```
the first time the startup is slow and some 5 tqdm bars. It shouldn't do it on consequent runs. but with `datasets>1.5.0` it rebuilds on every run.
@lhoestq
| 2,387 |
https://github.com/huggingface/datasets/issues/2386 | Accessing Arrow dataset cache_files | [
"Thanks @bhavitvyamalik for referencing the workaround. Setting `keep_in_memory=False` is working."
] | ## Describe the bug
In datasets 1.5.0 the following code snippet would have printed the cache_files:
```
train_data = load_dataset('conll2003', split='train', cache_dir='data')
print(train_data.cache_files[0]['filename'])
```
However, in the newest release (1.6.1), it prints an empty list.
I also tried loading the dataset with `keep_in_memory=True` argument but still `cache_files` is empty.
Was wondering if this is a bug or I need to pass additional arguments so I can access the cache_files.
| 2,386 |
https://github.com/huggingface/datasets/issues/2382 | DuplicatedKeysError: FAILURE TO GENERATE DATASET ! load_dataset('head_qa', 'en') | [] | Hello everyone,
I try to use head_qa dataset in [https://huggingface.co/datasets/viewer/?dataset=head_qa&config=en](url)
```
!pip install datasets
from datasets import load_dataset
dataset = load_dataset(
'head_qa', 'en')
```
When I write above load_dataset(.), it throws the following:
```
DuplicatedKeysError Traceback (most recent call last)
<ipython-input-6-ea87002d32f0> in <module>()
2 from datasets import load_dataset
3 dataset = load_dataset(
----> 4 'head_qa', 'en')
5 frames
/usr/local/lib/python3.7/dist-packages/datasets/arrow_writer.py in check_duplicate_keys(self)
347 for hash, key in self.hkey_record:
348 if hash in tmp_record:
--> 349 raise DuplicatedKeysError(key)
350 else:
351 tmp_record.add(hash)
DuplicatedKeysError: FAILURE TO GENERATE DATASET !
Found duplicate Key: 1
Keys should be unique and deterministic in nature
```
How can I fix the error? Thanks
| 2,382 |
https://github.com/huggingface/datasets/issues/2378 | Add missing dataset_infos.json files | [] | Some of the datasets in `datasets` are missing a `dataset_infos.json` file, e.g.
```
[PosixPath('datasets/chr_en/chr_en.py'), PosixPath('datasets/chr_en/README.md')]
[PosixPath('datasets/telugu_books/README.md'), PosixPath('datasets/telugu_books/telugu_books.py')]
[PosixPath('datasets/reclor/README.md'), PosixPath('datasets/reclor/reclor.py')]
[PosixPath('datasets/json/README.md')]
[PosixPath('datasets/csv/README.md')]
[PosixPath('datasets/wikihow/wikihow.py'), PosixPath('datasets/wikihow/README.md')]
[PosixPath('datasets/c4/c4.py'), PosixPath('datasets/c4/README.md')]
[PosixPath('datasets/text/README.md')]
[PosixPath('datasets/lm1b/README.md'), PosixPath('datasets/lm1b/lm1b.py')]
[PosixPath('datasets/pandas/README.md')]
```
For `json`, `text`, csv`, and `pandas` this is expected, but not for the others which should be fixed
| 2,378 |
https://github.com/huggingface/datasets/issues/2377 | ArrowDataset.save_to_disk produces files that cannot be read using pyarrow.feather | [
"Hi ! This is because we are actually using the arrow streaming format. We plan to switch to the arrow IPC format.\r\nMore info at #1933 ",
"Not sure if this was resolved, but I am getting a similar error when trying to load a dataset.arrow file directly: `ArrowInvalid: Not an Arrow file`",
"Since we're using t... | ## Describe the bug
A clear and concise description of what the bug is.
## Steps to reproduce the bug
```python
from datasets import load_dataset
from pyarrow import feather
dataset = load_dataset('imdb', split='train')
dataset.save_to_disk('dataset_dir')
table = feather.read_table('dataset_dir/dataset.arrow')
```
## Expected results
I expect that the saved dataset can be read by the official Apache Arrow methods.
## Actual results
```
File "/usr/local/lib/python3.7/site-packages/pyarrow/feather.py", line 236, in read_table
reader.open(source, use_memory_map=memory_map)
File "pyarrow/feather.pxi", line 67, in pyarrow.lib.FeatherReader.open
File "pyarrow/error.pxi", line 123, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 85, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: Not a Feather V1 or Arrow IPC file
```
## Environment info
<!-- You can run the command `datasets-cli env` and copy-and-paste its output below. -->
- `datasets` version: datasets-1.6.2
- Platform: Linux
- Python version: 3.7
- PyArrow version: 0.17.1, also 2.0.0
| 2,377 |
https://github.com/huggingface/datasets/issues/2373 | Loading dataset from local path | [
"Version below works, checked again in the docs, and data_files should be a path.\r\n```\r\nds = datasets.load_dataset('my_script.py', \r\n data_files='/data/dir/corpus.txt', \r\n cache_dir='.')\r\n```"
] | I'm trying to load a local dataset with the code below
```
ds = datasets.load_dataset('my_script.py',
data_files='corpus.txt',
data_dir='/data/dir',
cache_dir='.')
```
But internally a BuilderConfig is created, which tries to use getmtime on the data_files string, without using data_dir. Is this a bug or am I not using the load_dataset correctly?
https://github.com/huggingface/datasets/blob/bc61954083f74e6460688202e9f77dde2475319c/src/datasets/builder.py#L153 | 2,373 |
https://github.com/huggingface/datasets/issues/2371 | Align question answering tasks with sub-domains | [
"Closing this issue as the `task_templates` API has been deprecated."
] | As pointed out by @thomwolf in #2255 we should consider breaking with the pipeline taxonomy of `transformers` to account for the various types of question-answering domains:
> `question-answering` exists in two forms: abstractive and extractive question answering.
>
> we can keep a generic `question-answering` but then it will probably mean diferrent schema of input/output for both (abstractive will have text for both while extractive can use spans indication as well as text).
>
> Or we can also propose to use `abstractive-question-answering` and `extractive-question-answering` for instance.
> Maybe we could have `question-answering-abstractive` and `question-answering-extractive` if somehow we can use a for a completion or search in the future (detail).
> Actually I see that people are more organizing in terms of general and sub-tasks, for instance on paperwithcode: https://paperswithcode.com/area/natural-language-processing and on nlpprogress: https://github.com/sebastianruder/NLP-progress/blob/master/english/question_answering.md#squad
>
> Probably the best is to align with one of these in terms of denomination, PaperWithCode is probably the most active and maintained and we work with them as well.
> Maybe you want to check with a few QA datasets that this schema make sense. Typically NaturalQuestions, TriviaQA and can be good second datasets to compare to and be sure of the generality of the schema.
>
> A good recent list of QA datasets to compare the schemas among, is for instance in the UnitedQA paper: https://arxiv.org/abs/2101.00178
Investigate which grouping of QA is best suited for `datasets` and adapt / extend the QA task template accordingly. | 2,371 |
https://github.com/huggingface/datasets/issues/2366 | Json loader fails if user-specified features don't match the json data fields order | [] | If you do
```python
dataset = load_dataset("json", data_files=data_files, features=features)
```
Then depending on the order of the features in the json data field it fails:
```python
[...]
~/Desktop/hf/datasets/src/datasets/packaged_modules/json/json.py in _generate_tables(self, files)
94 if self.config.schema:
95 # Cast allows str <-> int/float, while parse_option explicit_schema does NOT
---> 96 pa_table = pa_table.cast(self.config.schema)
97 yield i, pa_table
[...]
ValueError: Target schema's field names are not matching the table's field names: ['tokens', 'ner_tags'], ['ner_tags', 'tokens']
```
This is because one must first re-order the columns of the table to match the `self.config.schema` before calling cast.
One way to fix the `cast` would be to replace it with:
```python
# reorder the arrays if necessary + cast to schema
# we can't simply use .cast here because we may need to change the order of the columns
pa_table = pa.Table.from_arrays([pa_table[name] for name in schema.names], schema=schema)
``` | 2,366 |
https://github.com/huggingface/datasets/issues/2365 | Missing ClassLabel encoding in Json loader | [] | Currently if you want to load a json dataset this way
```python
dataset = load_dataset("json", data_files=data_files, features=features)
```
Then if your features has ClassLabel types and if your json data needs class label encoding (i.e. if the labels in the json files are strings and not integers), then it would fail:
```python
[...]
~/Desktop/hf/datasets/src/datasets/packaged_modules/json/json.py in _generate_tables(self, files)
94 if self.config.schema:
95 # Cast allows str <-> int/float, while parse_option explicit_schema does NOT
---> 96 pa_table = pa_table.cast(self.config.schema)
97 yield i, pa_table
[...]
ArrowInvalid: Failed to parse string: 'O' as a scalar of type int64
```
This is because it just tries to cast the string data to integers, without applying the mapping str->int first
The current workaround is to do instead
```python
dataset = load_dataset("json", data_files=data_files)
dataset = dataset.map(features.encode_example, features=features)
``` | 2,365 |
https://github.com/huggingface/datasets/issues/2360 | Automatically detect datasets with compatible task schemas | [] | See description of #2255 for details.
| 2,360 |
https://github.com/huggingface/datasets/issues/2359 | Allow model labels to be passed during task preparation | [
"We now have the `align_labels_with_mapping` method in the API for this purpose."
] | Models have a config with label2id. And we have the same for datasets with the ClassLabel feature type. At one point either the model or the dataset must sync with the other. It would be great to do that on the dataset side.
For example for sentiment classification on amazon reviews with you could have these labels:
- "1 star", "2 stars", "3 stars", "4 stars", "5 stars"
- "1", "2", "3", "4", "5"
Some models may use the first set, while other models use the second set.
Here in the `TextClassification` class, the user can only specify one set of labels, while many models could actually be compatible but have different sets of labels. Should we allow users to pass a list of compatible labels sets ?
Then in terms of API, users could use `dataset.prepare_for_task("text-classification", labels=model.labels)` or something like that.
The label set could also be the same but not in the same order. For NLI for example, some models use `["neutral", "entailment", "contradiction"]` and some others use `["neutral", "contradiction", "entailment"]`, so we should take care of updating the order of the labels in the dataset to match the labels order of the model.
Let me know what you think ! This can be done in a future PR
_Originally posted by @lhoestq in https://github.com/huggingface/datasets/pull/2255#discussion_r632412792_ | 2,359 |
https://github.com/huggingface/datasets/issues/2354 | Document DatasetInfo attributes | [] | **Is your feature request related to a problem? Please describe.**
As noted in PR #2255, the attributes of `DatasetInfo` are not documented in the [docs](https://huggingface.co/docs/datasets/package_reference/main_classes.html?highlight=datasetinfo#datasetinfo). It would be nice to do so :)
| 2,354 |
https://github.com/huggingface/datasets/issues/2350 | `FaissIndex.save` throws error on GPU | [
"Just in case, this is a workaround that I use in my code and it seems to do the job.\r\n\r\n```python\r\nif use_gpu_index:\r\n data[\"train\"]._indexes[\"text_emb\"].faiss_index = faiss.index_gpu_to_cpu(data[\"train\"]._indexes[\"text_emb\"].faiss_index)\r\n```"
] | ## Describe the bug
After training an index with a factory string `OPQ16_128,IVF512,PQ32` on GPU, `.save_faiss_index` throws this error.
```
File "index_wikipedia.py", line 119, in <module>
data["train"].save_faiss_index("text_emb", index_save_path)
File "/home/vlialin/miniconda3/envs/cat/lib/python3.8/site-packages/datasets/search.py", line 470, in save_faiss_index
index.save(file)
File "/home/vlialin/miniconda3/envs/cat/lib/python3.8/site-packages/datasets/search.py", line 334, in save
faiss.write_index(index, str(file))
File "/home/vlialin/miniconda3/envs/cat/lib/python3.8/site-packages/faiss/swigfaiss_avx2.py", line 5654, in write_index
return _swigfaiss.write_index(*args)
RuntimeError: Error in void faiss::write_index(const faiss::Index*, faiss::IOWriter*) at /root/miniconda3/conda-bld/faiss-pkg_1613235005464/work/faiss/impl/index_write.cpp:453: don't know how to serialize this type of index
```
## Steps to reproduce the bug
Any dataset will do, I just selected a familiar one.
```python
import numpy as np
import datasets
INDEX_STR = "OPQ16_128,IVF512,PQ32"
INDEX_SAVE_PATH = "will_not_save.faiss"
data = datasets.load_dataset("Fraser/news-category-dataset", split=f"train[:10000]")
def encode(item):
return {"text_emb": np.random.randn(768).astype(np.float32)}
data = data.map(encode)
data.add_faiss_index(column="text_emb", string_factory=INDEX_STR, train_size=10_000, device=0)
data.save_faiss_index("text_emb", INDEX_SAVE_PATH)
```
## Expected results
Saving the index
## Actual results
Error in void faiss::write_index(const faiss::Index*, faiss::IOWriter*) ... don't know how to serialize this type of index
## Environment info
- `datasets` version: 1.6.2
- Platform: Linux-4.15.0-142-generic-x86_64-with-glibc2.10
- Python version: 3.8.8
- PyTorch version (GPU?): 1.8.1+cu111 (True)
- Tensorflow version (GPU?): 2.2.0 (False)
- Using GPU in script?: Yes
- Using distributed or parallel set-up in script?: No
I will be proposing a fix in a couple of minutes | 2,350 |
https://github.com/huggingface/datasets/issues/2347 | Add an API to access the language and pretty name of a dataset | [
"Hi ! With @bhavitvyamalik we discussed about having something like\r\n```python\r\nfrom datasets import load_dataset_card\r\n\r\ndataset_card = load_dataset_card(\"squad\")\r\nprint(dataset_card.metadata.pretty_name)\r\n# Stanford Question Answering Dataset (SQuAD)\r\nprint(dataset_card.metadata.languages)\r\n# [\... | It would be super nice to have an API to get some metadata of the dataset from the name and args passed to `load_dataset`. This way we could programmatically infer the language and the name of a dataset when creating model cards automatically in the Transformers examples scripts. | 2,347 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.