id
int64 599M
3.26B
| number
int64 1
7.7k
| title
stringlengths 1
290
| body
stringlengths 0
228k
⌀ | state
stringclasses 2
values | html_url
stringlengths 46
51
| created_at
timestamp[s]date 2020-04-14 10:18:02
2025-07-23 08:04:53
| updated_at
timestamp[s]date 2020-04-27 16:04:17
2025-07-23 18:53:44
| closed_at
timestamp[s]date 2020-04-14 12:01:40
2025-07-23 16:44:42
⌀ | user
dict | labels
listlengths 0
4
| is_pull_request
bool 2
classes | comments
listlengths 0
0
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|
2,063,839,916
| 6,554
|
Parquet exports are used even if revision is passed
|
We should not used Parquet exports if `revision` is passed.
I think this is a regression.
|
closed
|
https://github.com/huggingface/datasets/issues/6554
| 2024-01-03T11:32:26
| 2024-02-02T10:35:29
| 2024-02-02T10:35:29
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[
{
"name": "bug",
"color": "d73a4a"
}
] | false
|
[] |
2,063,474,183
| 6,553
|
Cannot import name 'load_dataset' from .... module ‘datasets’
|
### Describe the bug
use python -m pip install datasets to install
### Steps to reproduce the bug
from datasets import load_dataset
### Expected behavior
it doesn't work
### Environment info
datasets version==2.15.0
python == 3.10.12
linux version I don't know??
|
closed
|
https://github.com/huggingface/datasets/issues/6553
| 2024-01-03T08:18:21
| 2024-02-21T00:38:24
| 2024-02-21T00:38:24
|
{
"login": "ciaoyizhen",
"id": 83450192,
"type": "User"
}
|
[] | false
|
[] |
2,063,157,187
| 6,552
|
Loading a dataset from Google Colab hangs at "Resolving data files".
|
### Describe the bug
Hello,
I'm trying to load a dataset from Google Colab but the process hangs at `Resolving data files`:
It is happening when the `_get_origin_metadata` definition is invoked:
```python
def _get_origin_metadata(
data_files: List[str],
max_workers=64,
download_config: Optional[DownloadConfig] = None,
) -> Tuple[str]:
return thread_map(
partial(_get_single_origin_metadata, download_config=download_config),
data_files,
max_workers=max_workers,
tqdm_class=hf_tqdm,
desc="Resolving data files",
disable=len(data_files) <= 16,
```
The thread is then stuck at `waiter.acquire()` in the builtin `threading.py` file.
I can load the dataset just fine on my machine.
Cheers,
Thomas
### Steps to reproduce the bug
In Google Colab:
```python
!pip install datasets
from datasets import load_dataset
dataset = load_dataset("colour-science/color-checker-detection-dataset")
```
### Expected behavior
The dataset should be loaded.
### Environment info
- `datasets` version: 2.16.1
- Platform: Linux-6.1.58+-x86_64-with-glibc2.35
- Python version: 3.10.12
- `huggingface_hub` version: 0.20.1
- PyArrow version: 10.0.1
- Pandas version: 1.5.3
- `fsspec` version: 2023.6.0
|
closed
|
https://github.com/huggingface/datasets/issues/6552
| 2024-01-03T02:18:17
| 2024-01-08T10:09:04
| 2024-01-08T10:09:04
|
{
"login": "KelSolaar",
"id": 99779,
"type": "User"
}
|
[] | false
|
[] |
2,062,768,400
| 6,551
|
Fix parallel downloads for datasets without scripts
|
Enable parallel downloads using multiprocessing when `num_proc` is passed to `load_dataset`.
It was enabled for datasets with scripts already (if they passed lists to `dl_manager.download`) but not for no-script datasets (we pass dicts {split: [list of files]} to `dl_manager.download` for those ones).
I fixed this by parallelising on the lists contained in the data files dicts when possible.
I also added a context manager `stack_multiprocessing_download_progress_bars` in `DownloadManager` to stack the progress bard of the downloads (from `cached_path(...)` calls). Otherwise the progress bars overlap each other with an annoying flickering effect.
|
closed
|
https://github.com/huggingface/datasets/pull/6551
| 2024-01-02T18:06:18
| 2024-01-06T20:14:57
| 2024-01-03T13:19:48
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,062,556,493
| 6,550
|
Multi gpu docs
|
after discussions in https://github.com/huggingface/datasets/pull/6415
|
closed
|
https://github.com/huggingface/datasets/pull/6550
| 2024-01-02T15:11:58
| 2024-01-31T13:45:15
| 2024-01-31T13:38:59
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,062,420,259
| 6,549
|
Loading from hf hub with clearer error message
|
### Feature request
Shouldn't this kinda work ?
```
Dataset.from_json("hf://datasets/HuggingFaceTB/eval_data/resolve/main/eval_data_context_and_answers.json")
```
I got an error
```
File ~/miniconda3/envs/datatrove/lib/python3.10/site-packages/datasets/data_files.py:380, in resolve_pattern(pattern, base_path, allowed_extensions, download_config)
378 if allowed_extensions is not None:
379 error_msg += f" with any supported extension {list(allowed_extensions)}"
--> 380 raise FileNotFoundError(error_msg)
381 return out
FileNotFoundError: Unable to find 'hf://datasets/HuggingFaceTB/eval_data/resolve/main/eval_data_context_and_answers.json'
(I'm logged in)
```
Fix: the correct path is
```
hf://datasets/HuggingFaceTB/eval_data/eval_data_context_and_answers.json
```
Proposal: raise a clearer error
### Motivation
Clearer error message
### Your contribution
Can open a PR
|
open
|
https://github.com/huggingface/datasets/issues/6549
| 2024-01-02T13:26:34
| 2024-01-02T14:06:49
| null |
{
"login": "thomwolf",
"id": 7353373,
"type": "User"
}
|
[
{
"name": "enhancement",
"color": "a2eeef"
}
] | false
|
[] |
2,061,047,984
| 6,548
|
Skip if a dataset has issues
|
### Describe the bug
Hello everyone,
I'm using **load_datasets** from **huggingface** to download the datasets and I'm facing an issue, the download starts but it reaches some state and then fails with the following error:
Couldn't reach https://huggingface.co/datasets/wikimedia/wikipedia/resolve/4cb9b0d719291f1a10f96f67d609c5d442980dc9/20231101.ext/train-00000-of-00001.parquet
Failed to resolve \'huggingface.co\' ([Errno -3] Temporary failure in name resolution)"))')))
so I was wondering is there a parameter to be passed to load_dataset() to skip files that can't be downloaded??
### Steps to reproduce the bug
Parameter to be passed to load_dataset() of huggingface to skip files that can't be downloaded??
### Expected behavior
load_dataset() finishes without error
### Environment info
None
|
open
|
https://github.com/huggingface/datasets/issues/6548
| 2023-12-31T12:41:26
| 2024-01-02T10:33:17
| null |
{
"login": "hadianasliwa",
"id": 143214684,
"type": "User"
}
|
[] | false
|
[] |
2,060,796,927
| 6,547
|
set dev version
| null |
closed
|
https://github.com/huggingface/datasets/pull/6547
| 2023-12-30T16:47:17
| 2023-12-30T16:53:38
| 2023-12-30T16:47:27
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,060,796,369
| 6,546
|
Release: 2.16.1
| null |
closed
|
https://github.com/huggingface/datasets/pull/6546
| 2023-12-30T16:44:51
| 2023-12-30T16:52:07
| 2023-12-30T16:45:52
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,060,789,507
| 6,545
|
`image` column not automatically inferred if image dataset only contains 1 image
|
### Describe the bug
By default, the standard Image Dataset maps out `file_name` to `image` when loading an Image Dataset.
However, if the dataset contains only 1 image, this does not take place
### Steps to reproduce the bug
Input
(dataset with one image `multimodalart/repro_1_image`)
```py
from datasets import load_dataset
dataset = load_dataset("multimodalart/repro_1_image")
dataset
```
Output:
```py
DatasetDict({
train: Dataset({
features: ['file_name', 'prompt'],
num_rows: 1
})
})
```
Input
(dataset with 2+ images `multimodalart/repro_2_image`)
```py
from datasets import load_dataset
dataset = load_dataset("multimodalart/repro_2_image")
dataset
```
Output:
```py
DatasetDict({
train: Dataset({
features: ['image', 'prompt'],
num_rows: 2
})
})
```
### Expected behavior
Expected to map `file_name` → `image` for all dataset sizes, including 1.
### Environment info
Both latest main and 2.16.0
|
closed
|
https://github.com/huggingface/datasets/issues/6545
| 2023-12-30T16:17:29
| 2024-01-09T13:06:31
| 2024-01-09T13:06:31
|
{
"login": "apolinario",
"id": 788417,
"type": "User"
}
|
[] | false
|
[] |
2,060,782,594
| 6,544
|
Fix custom configs from script
|
We should not use the parquet export when the user is passing config_kwargs
I also fixed a regression that would disallow creating a custom config when a dataset has multiple predefined configs
fix https://github.com/huggingface/datasets/issues/6533
|
closed
|
https://github.com/huggingface/datasets/pull/6544
| 2023-12-30T15:51:25
| 2024-01-02T11:02:39
| 2023-12-30T16:09:49
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,060,776,174
| 6,543
|
Fix dl_manager.extract returning FileNotFoundError
|
The dl_manager base path is remote (e.g. a hf:// path), so local cached paths should be passed as absolute paths.
This could happen if users provide a relative path as `cache_dir`
fix https://github.com/huggingface/datasets/issues/6536
|
closed
|
https://github.com/huggingface/datasets/pull/6543
| 2023-12-30T15:24:50
| 2023-12-30T16:00:06
| 2023-12-30T15:53:59
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,059,198,575
| 6,542
|
Datasets : wikipedia 20220301.en error
|
### Describe the bug
When I used load_dataset to download this data set, the following error occurred. The main problem was that the target data did not exist.
### Steps to reproduce the bug
1.I tried downloading directly.
```python
wiki_dataset = load_dataset("wikipedia", "20220301.en")
```
An exception occurred
```
MissingBeamOptions: Trying to generate a dataset using Apache Beam, yet no Beam Runner or PipelineOptions() has been provided in `load_dataset` or in the builder arguments. For big datasets it has to run on large-scale data processing tools like Dataflow, Spark, etc. More information about Apache Beam runners at https://beam.apache.org/documentation/runners/capability-matrix/
If you really want to run it locally because you feel like the Dataset is small enough, you can use the local beam runner called `DirectRunner` (you may run out of memory).
Example of usage:
`load_dataset('wikipedia', '20220301.en', beam_runner='DirectRunner')`
```
2.I modified the code as prompted.
```python
wiki_dataset = load_dataset('wikipedia', '20220301.en', beam_runner='DirectRunner')
```
An exception occurred:
```
FileNotFoundError: Couldn't find file at https://dumps.wikimedia.org/enwiki/20220301/dumpstatus.json
```
### Expected behavior
I searched in the parent directory of the corresponding URL, but there was no corresponding "20220301" directory.
I really need this data set and hope to provide a download method.
### Environment info
python 3.8
datasets 2.16.0
apache-beam 2.52.0
dill 0.3.7
|
closed
|
https://github.com/huggingface/datasets/issues/6542
| 2023-12-29T08:34:51
| 2024-01-02T13:21:06
| 2024-01-02T13:20:30
|
{
"login": "ppx666",
"id": 53203620,
"type": "User"
}
|
[] | false
|
[] |
2,058,983,826
| 6,541
|
Dataset not loading successfully.
|
### Describe the bug
When I run down the below code shows this error: AttributeError: module 'numpy' has no attribute '_no_nep50_warning'
I also added this issue in transformers library please check out: [link](https://github.com/huggingface/transformers/issues/28099)
### Steps to reproduce the bug
## Reproduction
Hi, please check this line of code, when I run Show attribute error.
```
from datasets import load_dataset
from transformers import WhisperProcessor, WhisperForConditionalGeneration
# Select an audio file and read it:
ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
audio_sample = ds[0]["audio"]
waveform = audio_sample["array"]
sampling_rate = audio_sample["sampling_rate"]
# Load the Whisper model in Hugging Face format:
processor = WhisperProcessor.from_pretrained("openai/whisper-tiny.en")
model = WhisperForConditionalGeneration.from_pretrained("openai/whisper-tiny.en")
# Use the model and processor to transcribe the audio:
input_features = processor(
waveform, sampling_rate=sampling_rate, return_tensors="pt"
).input_features
# Generate token ids
predicted_ids = model.generate(input_features)
# Decode token ids to text
transcription = processor.batch_decode(predicted_ids, skip_special_tokens=True)
transcription[0]
```
**Attribute Error**
```
AttributeError Traceback (most recent call last)
Cell In[9], line 6
4 # Select an audio file and read it:
5 ds = load_dataset("hf-internal-testing/librispeech_asr_dummy", "clean", split="validation")
----> 6 audio_sample = ds[0]["audio"]
7 waveform = audio_sample["array"]
8 sampling_rate = audio_sample["sampling_rate"]
File /opt/pytorch/lib/python3.8/site-packages/datasets/arrow_dataset.py:2795, in Dataset.__getitem__(self, key)
2793 def __getitem__(self, key): # noqa: F811
2794 """Can be used to index columns (by string names) or rows (by integer index or iterable of indices or bools)."""
-> 2795 return self._getitem(key)
File /opt/pytorch/lib/python3.8/site-packages/datasets/arrow_dataset.py:2780, in Dataset._getitem(self, key, **kwargs)
2778 formatter = get_formatter(format_type, features=self._info.features, **format_kwargs)
2779 pa_subtable = query_table(self._data, key, indices=self._indices if self._indices is not None else None)
-> 2780 formatted_output = format_table(
2781 pa_subtable, key, formatter=formatter, format_columns=format_columns, output_all_columns=output_all_columns
2782 )
2783 return formatted_output
File /opt/pytorch/lib/python3.8/site-packages/datasets/formatting/formatting.py:629, in format_table(table, key, formatter, format_columns, output_all_columns)
627 python_formatter = PythonFormatter(features=formatter.features)
628 if format_columns is None:
--> 629 return formatter(pa_table, query_type=query_type)
630 elif query_type == "column":
631 if key in format_columns:
File /opt/pytorch/lib/python3.8/site-packages/datasets/formatting/formatting.py:396, in Formatter.__call__(self, pa_table, query_type)
394 def __call__(self, pa_table: pa.Table, query_type: str) -> Union[RowFormat, ColumnFormat, BatchFormat]:
395 if query_type == "row":
--> 396 return self.format_row(pa_table)
397 elif query_type == "column":
398 return self.format_column(pa_table)
File /opt/pytorch/lib/python3.8/site-packages/datasets/formatting/formatting.py:437, in PythonFormatter.format_row(self, pa_table)
435 return LazyRow(pa_table, self)
436 row = self.python_arrow_extractor().extract_row(pa_table)
--> 437 row = self.python_features_decoder.decode_row(row)
438 return row
File /opt/pytorch/lib/python3.8/site-packages/datasets/formatting/formatting.py:215, in PythonFeaturesDecoder.decode_row(self, row)
214 def decode_row(self, row: dict) -> dict:
--> 215 return self.features.decode_example(row) if self.features else row
File /opt/pytorch/lib/python3.8/site-packages/datasets/features/features.py:1917, in Features.decode_example(self, example, token_per_repo_id)
1903 def decode_example(self, example: dict, token_per_repo_id: Optional[Dict[str, Union[str, bool, None]]] = None):
1904 """Decode example with custom feature decoding.
1905
1906 Args:
(...)
1914 `dict[str, Any]`
1915 """
-> 1917 return {
1918 column_name: decode_nested_example(feature, value, token_per_repo_id=token_per_repo_id)
1919 if self._column_requires_decoding[column_name]
1920 else value
1921 for column_name, (feature, value) in zip_dict(
1922 {key: value for key, value in self.items() if key in example}, example
1923 )
1924 }
File /opt/pytorch/lib/python3.8/site-packages/datasets/features/features.py:1918, in <dictcomp>(.0)
1903 def decode_example(self, example: dict, token_per_repo_id: Optional[Dict[str, Union[str, bool, None]]] = None):
1904 """Decode example with custom feature decoding.
1905
1906 Args:
(...)
1914 `dict[str, Any]`
1915 """
1917 return {
-> 1918 column_name: decode_nested_example(feature, value, token_per_repo_id=token_per_repo_id)
1919 if self._column_requires_decoding[column_name]
1920 else value
1921 for column_name, (feature, value) in zip_dict(
1922 {key: value for key, value in self.items() if key in example}, example
1923 )
1924 }
File /opt/pytorch/lib/python3.8/site-packages/datasets/features/features.py:1339, in decode_nested_example(schema, obj, token_per_repo_id)
1336 elif isinstance(schema, (Audio, Image)):
1337 # we pass the token to read and decode files from private repositories in streaming mode
1338 if obj is not None and schema.decode:
-> 1339 return schema.decode_example(obj, token_per_repo_id=token_per_repo_id)
1340 return obj
File /opt/pytorch/lib/python3.8/site-packages/datasets/features/audio.py:191, in Audio.decode_example(self, value, token_per_repo_id)
189 array = array.T
190 if self.mono:
--> 191 array = librosa.to_mono(array)
192 if self.sampling_rate and self.sampling_rate != sampling_rate:
193 array = librosa.resample(array, orig_sr=sampling_rate, target_sr=self.sampling_rate)
File /opt/pytorch/lib/python3.8/site-packages/lazy_loader/__init__.py:78, in attach.<locals>.__getattr__(name)
76 submod_path = f"{package_name}.{attr_to_modules[name]}"
77 submod = importlib.import_module(submod_path)
---> 78 attr = getattr(submod, name)
80 # If the attribute lives in a file (module) with the same
81 # name as the attribute, ensure that the attribute and *not*
82 # the module is accessible on the package.
83 if name == attr_to_modules[name]:
File /opt/pytorch/lib/python3.8/site-packages/lazy_loader/__init__.py:77, in attach.<locals>.__getattr__(name)
75 elif name in attr_to_modules:
76 submod_path = f"{package_name}.{attr_to_modules[name]}"
---> 77 submod = importlib.import_module(submod_path)
78 attr = getattr(submod, name)
80 # If the attribute lives in a file (module) with the same
81 # name as the attribute, ensure that the attribute and *not*
82 # the module is accessible on the package.
File /usr/lib/python3.8/importlib/__init__.py:127, in import_module(name, package)
125 break
126 level += 1
--> 127 return _bootstrap._gcd_import(name[level:], package, level)
File <frozen importlib._bootstrap>:1014, in _gcd_import(name, package, level)
File <frozen importlib._bootstrap>:991, in _find_and_load(name, import_)
File <frozen importlib._bootstrap>:975, in _find_and_load_unlocked(name, import_)
File <frozen importlib._bootstrap>:671, in _load_unlocked(spec)
File <frozen importlib._bootstrap_external>:848, in exec_module(self, module)
File <frozen importlib._bootstrap>:219, in _call_with_frames_removed(f, *args, **kwds)
File /opt/pytorch/lib/python3.8/site-packages/librosa/core/audio.py:13
11 import audioread
12 import numpy as np
---> 13 import scipy.signal
14 import soxr
15 import lazy_loader as lazy
File /opt/pytorch/lib/python3.8/site-packages/scipy/signal/__init__.py:323
314 from ._spline import ( # noqa: F401
315 cspline2d,
316 qspline2d,
(...)
319 symiirorder2,
320 )
322 from ._bsplines import *
--> 323 from ._filter_design import *
324 from ._fir_filter_design import *
325 from ._ltisys import *
File /opt/pytorch/lib/python3.8/site-packages/scipy/signal/_filter_design.py:16
13 from numpy.polynomial.polynomial import polyval as npp_polyval
14 from numpy.polynomial.polynomial import polyvalfromroots
---> 16 from scipy import special, optimize, fft as sp_fft
17 from scipy.special import comb
18 from scipy._lib._util import float_factorial
File /opt/pytorch/lib/python3.8/site-packages/scipy/optimize/__init__.py:405
1 """
2 =====================================================
3 Optimization and root finding (:mod:`scipy.optimize`)
(...)
401
402 """
404 from ._optimize import *
--> 405 from ._minimize import *
406 from ._root import *
407 from ._root_scalar import *
File /opt/pytorch/lib/python3.8/site-packages/scipy/optimize/_minimize.py:26
24 from ._trustregion_krylov import _minimize_trust_krylov
25 from ._trustregion_exact import _minimize_trustregion_exact
---> 26 from ._trustregion_constr import _minimize_trustregion_constr
28 # constrained minimization
29 from ._lbfgsb_py import _minimize_lbfgsb
File /opt/pytorch/lib/python3.8/site-packages/scipy/optimize/_trustregion_constr/__init__.py:4
1 """This module contains the equality constrained SQP solver."""
----> 4 from .minimize_trustregion_constr import _minimize_trustregion_constr
6 __all__ = ['_minimize_trustregion_constr']
File /opt/pytorch/lib/python3.8/site-packages/scipy/optimize/_trustregion_constr/minimize_trustregion_constr.py:5
3 from scipy.sparse.linalg import LinearOperator
4 from .._differentiable_functions import VectorFunction
----> 5 from .._constraints import (
6 NonlinearConstraint, LinearConstraint, PreparedConstraint, strict_bounds)
7 from .._hessian_update_strategy import BFGS
8 from .._optimize import OptimizeResult
File /opt/pytorch/lib/python3.8/site-packages/scipy/optimize/_constraints.py:8
6 from ._optimize import OptimizeWarning
7 from warnings import warn, catch_warnings, simplefilter
----> 8 from numpy.testing import suppress_warnings
9 from scipy.sparse import issparse
12 def _arr_to_scalar(x):
13 # If x is a numpy array, return x.item(). This will
14 # fail if the array has more than one element.
File /opt/pytorch/lib/python3.8/site-packages/numpy/testing/__init__.py:11
8 from unittest import TestCase
10 from . import _private
---> 11 from ._private.utils import *
12 from ._private.utils import (_assert_valid_refcount, _gen_alignment_data)
13 from ._private import extbuild, decorators as dec
File /opt/pytorch/lib/python3.8/site-packages/numpy/testing/_private/utils.py:480
476 pprint.pprint(desired, msg)
477 raise AssertionError(msg.getvalue())
--> 480 @np._no_nep50_warning()
481 def assert_almost_equal(actual,desired,decimal=7,err_msg='',verbose=True):
482 """
483 Raises an AssertionError if two items are not equal up to desired
484 precision.
(...)
548
549 """
550 __tracebackhide__ = True # Hide traceback for py.test
File /opt/pytorch/lib/python3.8/site-packages/numpy/__init__.py:313, in __getattr__(attr)
305 raise AttributeError(__former_attrs__[attr])
307 # Importing Tester requires importing all of UnitTest which is not a
308 # cheap import Since it is mainly used in test suits, we lazy import it
309 # here to save on the order of 10 ms of import time for most users
310 #
311 # The previous way Tester was imported also had a side effect of adding
312 # the full `numpy.testing` namespace
--> 313 if attr == 'testing':
314 import numpy.testing as testing
315 return testing
AttributeError: module 'numpy' has no attribute '_no_nep50_warning'
```
### Expected behavior
``` ' Mr. Quilter is the apostle of the middle classes, and we are glad to welcome his gospel.' ```
Also, make sure this script is provided for your official website so please update:
[script](https://huggingface.co/docs/transformers/model_doc/whisper)
### Environment info
**System Info**
* transformers -> 4.36.1
* datasets -> 2.15.0
* huggingface_hub -> 0.19.4
* python -> 3.8.10
* accelerate -> 0.25.0
* pytorch -> 2.0.1+cpu
* Using GPU in Script -> No
|
closed
|
https://github.com/huggingface/datasets/issues/6541
| 2023-12-29T01:35:47
| 2024-01-17T00:40:46
| 2024-01-17T00:40:45
|
{
"login": "hisushanta",
"id": 93595990,
"type": "User"
}
|
[] | false
|
[] |
2,058,965,157
| 6,540
|
Extreme inefficiency for `save_to_disk` when merging datasets
|
### Describe the bug
Hi, I tried to merge in total 22M sequences of data, where each sequence is of maximum length 2000. I found that merging these datasets and then `save_to_disk` is extremely slow because of flattening the indices. Wondering if you have any suggestions or guidance on this. Thank you very much!
### Steps to reproduce the bug
The source data is too big to demonstrate
### Expected behavior
The source data is too big to demonstrate
### Environment info
python 3.9.0
datasets 2.7.0
pytorch 2.0.0
tokenizers 0.13.1
transformers 4.31.0
|
open
|
https://github.com/huggingface/datasets/issues/6540
| 2023-12-29T00:44:35
| 2023-12-30T15:05:48
| null |
{
"login": "KatarinaYuan",
"id": 43512683,
"type": "User"
}
|
[] | false
|
[] |
2,058,493,960
| 6,539
|
'Repo card metadata block was not found' when loading a pragmeval dataset
|
### Describe the bug
I can't load dataset subsets of 'pragmeval'.
The funny thing is I ran the dataset author's [colab notebook](https://colab.research.google.com/drive/1sg--LF4z7XR1wxAOfp0-3d4J6kQ9nj_A?usp=sharing) and it works just fine. I tried to install exactly the same packages that are installed on colab using poetry, so my environment info only differs from the one from colab in linux version - I still get the same bug outside colab.
### Steps to reproduce the bug
Install dependencies with poetry
pyproject.toml
```
[tool.poetry]
name = "project"
version = "0.1.0"
description = ""
authors = []
[tool.poetry.dependencies]
python = "^3.10"
datasets = "2.16.0"
pandas = "1.5.3"
pyarrow = "10.0.1"
huggingface-hub = "0.19.4"
fsspec = "2023.6.0"
[build-system]
requires = ["poetry-core"]
build-backend = "poetry.core.masonry.api"
```
`poetry run python -c "import datasets; print(datasets.get_dataset_config_names('pragmeval'))`
prints ['default']
### Expected behavior
The command should print
```
['emergent',
'emobank-arousal',
'emobank-dominance',
'emobank-valence',
'gum',
'mrda',
'pdtb',
'persuasiveness-claimtype',
'persuasiveness-eloquence',
'persuasiveness-premisetype',
'persuasiveness-relevance',
'persuasiveness-specificity',
'persuasiveness-strength',
'sarcasm',
'squinky-formality',
'squinky-implicature',
'squinky-informativeness',
'stac',
'switchboard',
'verifiability']
```
### Environment info
- `datasets` version: 2.16.0
- Platform: Linux-6.2.0-37-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- `huggingface_hub` version: 0.19.4
- PyArrow version: 10.0.1
- Pandas version: 1.5.3
- `fsspec` version: 2023.6.0
|
open
|
https://github.com/huggingface/datasets/issues/6539
| 2023-12-28T14:18:25
| 2023-12-28T14:18:37
| null |
{
"login": "lambdaofgod",
"id": 3647577,
"type": "User"
}
|
[] | false
|
[] |
2,057,377,630
| 6,538
|
ImportError: cannot import name 'SchemaInferenceError' from 'datasets.arrow_writer' (/opt/conda/lib/python3.10/site-packages/datasets/arrow_writer.py)
|
### Describe the bug
While importing from packages getting the error
Code:
```
import os
import torch
from datasets import load_dataset, Dataset
from transformers import (
AutoModelForCausalLM,
AutoTokenizer,
BitsAndBytesConfig,
HfArgumentParser,
TrainingArguments,
pipeline,
logging
)
from peft import LoraConfig, PeftModel
from trl import SFTTrainer
from huggingface_hub import login
import pandas as pd
```
Error:
````
---------------------------------------------------------------------------
ImportError Traceback (most recent call last)
Cell In[5], line 14
4 from transformers import (
5 AutoModelForCausalLM,
6 AutoTokenizer,
(...)
11 logging
12 )
13 from peft import LoraConfig, PeftModel
---> 14 from trl import SFTTrainer
15 from huggingface_hub import login
16 import pandas as pd
File /opt/conda/lib/python3.10/site-packages/trl/__init__.py:21
8 from .import_utils import (
9 is_diffusers_available,
10 is_npu_available,
(...)
13 is_xpu_available,
14 )
15 from .models import (
16 AutoModelForCausalLMWithValueHead,
17 AutoModelForSeq2SeqLMWithValueHead,
18 PreTrainedModelWrapper,
19 create_reference_model,
20 )
---> 21 from .trainer import (
22 DataCollatorForCompletionOnlyLM,
23 DPOTrainer,
24 IterativeSFTTrainer,
25 PPOConfig,
26 PPOTrainer,
27 RewardConfig,
28 RewardTrainer,
29 SFTTrainer,
30 )
33 if is_diffusers_available():
34 from .models import (
35 DDPOPipelineOutput,
36 DDPOSchedulerOutput,
37 DDPOStableDiffusionPipeline,
38 DefaultDDPOStableDiffusionPipeline,
39 )
File /opt/conda/lib/python3.10/site-packages/trl/trainer/__init__.py:44
42 from .ppo_trainer import PPOTrainer
43 from .reward_trainer import RewardTrainer, compute_accuracy
---> 44 from .sft_trainer import SFTTrainer
45 from .training_configs import RewardConfig
File /opt/conda/lib/python3.10/site-packages/trl/trainer/sft_trainer.py:23
21 import torch.nn as nn
22 from datasets import Dataset
---> 23 from datasets.arrow_writer import SchemaInferenceError
24 from datasets.builder import DatasetGenerationError
25 from transformers import (
26 AutoModelForCausalLM,
27 AutoTokenizer,
(...)
33 TrainingArguments,
34 )
ImportError: cannot import name 'SchemaInferenceError' from 'datasets.arrow_writer' (/opt/conda/lib/python3.10/site-packages/datasets/arrow_writer.py
````
transformers version: 4.36.2
python version: 3.10.12
datasets version: 2.16.1
### Steps to reproduce the bug
1. Install packages
```
!pip install -U datasets trl accelerate peft bitsandbytes transformers trl huggingface_hub
```
2. import packages
```
import os
import torch
from datasets import load_dataset, Dataset
from transformers import (
AutoModelForCausalLM,
AutoTokenizer,
BitsAndBytesConfig,
HfArgumentParser,
TrainingArguments,
pipeline,
logging
)
from peft import LoraConfig, PeftModel
from trl import SFTTrainer
from huggingface_hub import login
import pandas as pd
```
### Expected behavior
No error while importing
### Environment info
- `datasets` version: 2.16.0
- Platform: Linux-5.15.133+-x86_64-with-glibc2.35
- Python version: 3.10.12
- `huggingface_hub` version: 0.20.1
- PyArrow version: 11.0.0
- Pandas version: 2.1.4
- `fsspec` version: 2023.10.0
|
closed
|
https://github.com/huggingface/datasets/issues/6538
| 2023-12-27T13:31:16
| 2024-01-03T10:06:47
| 2024-01-03T10:04:58
|
{
"login": "Sonali-Behera-TRT",
"id": 131662185,
"type": "User"
}
|
[] | false
|
[] |
2,057,132,173
| 6,537
|
Adding support for netCDF (*.nc) files
|
### Feature request
netCDF (*.nc) is a file format for storing multidimensional scientific data, which is used by packages like `xarray` (labelled multi-dimensional arrays in Python). It would be nice to have native support for netCDF in `datasets`.
### Motivation
When uploading *.nc files onto Huggingface Hub through the `datasets` API, I would like to be able to preview the dataset without converting it to another format.
### Your contribution
I can submit a PR, provided I have the time.
|
open
|
https://github.com/huggingface/datasets/issues/6537
| 2023-12-27T09:27:29
| 2023-12-27T20:46:53
| null |
{
"login": "shermansiu",
"id": 12627125,
"type": "User"
}
|
[
{
"name": "enhancement",
"color": "a2eeef"
}
] | false
|
[] |
2,056,863,239
| 6,536
|
datasets.load_dataset raises FileNotFoundError for datasets==2.16.0
|
### Describe the bug
Seems `datasets.load_dataset` raises FileNotFoundError for some hub datasets with the latest `datasets==2.16.0`
### Steps to reproduce the bug
For example `pip install datasets==2.16.0`
then
```python
import datasets
datasets.load_dataset("wentingzhao/anthropic-hh-first-prompt", cache_dir='cache1')["train"]
```
This will raise:
```bash
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/load.py", line 2545, in load_dataset
builder_instance.download_and_prepare(
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/builder.py", line 1003, in download_and_prepare
self._download_and_prepare(
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/builder.py", line 1076, in _download_and_prepare
split_generators = self._split_generators(dl_manager, **split_generators_kwargs)
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/packaged_modules/parquet/parquet.py", line 43, in _split_generators
data_files = dl_manager.download_and_extract(self.config.data_files)
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/download/download_manager.py", line 566, in download_and_extract
return self.extract(self.download(url_or_urls))
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/download/download_manager.py", line 539, in extract
extracted_paths = map_nested(
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/py_utils.py", line 466, in map_nested
mapped = [
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/py_utils.py", line 467, in <listcomp>
_single_map_nested((function, obj, types, None, True, None))
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/py_utils.py", line 387, in _single_map_nested
mapped = [_single_map_nested((function, v, types, None, True, None)) for v in pbar]
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/py_utils.py", line 387, in <listcomp>
mapped = [_single_map_nested((function, v, types, None, True, None)) for v in pbar]
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/py_utils.py", line 370, in _single_map_nested
return function(data_struct)
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/download/download_manager.py", line 451, in _download
out = cached_path(url_or_filename, download_config=download_config)
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/file_utils.py", line 188, in cached_path
output_path = get_from_cache(
File "/Users/xxx/miniconda3/envs/env/lib/python3.9/site-packages/datasets/utils/file_utils.py", line 570, in get_from_cache
raise FileNotFoundError(f"Couldn't find file at {url}")
FileNotFoundError: Couldn't find file at https://huggingface.co/datasets/wentingzhao/anthropic-hh-first-prompt/resolve/11b393a5545f706a357ebcd4a5285d93db176715/cache1/downloads/87d66c365626feca116cba323c4856c9aae056e4503f09f23e34aa085eb9de15
```
However, seems it works fine for some datasets, for example, if works fine for `datasets.load_dataset("ag_news", cache_dir='cache2')["test"]`
But the dataset works fine for datasets==2.15.0, for example `pip install datasets==2.15.0`,
then
```python
import datasets
datasets.load_dataset("wentingzhao/anthropic-hh-first-prompt", cache_dir='cache3')["train"]
Dataset({
features: ['user', 'system', 'source'],
num_rows: 8552
})
```
### Expected behavior
2.16.0 should work as same as 2.15.0 for all datasets
### Environment info
python3.9
conda env
tested on MacOS and Linux
|
closed
|
https://github.com/huggingface/datasets/issues/6536
| 2023-12-27T03:15:48
| 2023-12-30T18:58:04
| 2023-12-30T15:54:00
|
{
"login": "ArvinZhuang",
"id": 46237844,
"type": "User"
}
|
[] | false
|
[] |
2,056,264,339
| 6,535
|
IndexError: Invalid key: 47682 is out of bounds for size 0 while using PEFT
|
### Describe the bug
I am trying to fine-tune the t5 model on the paraphrasing task. While running the same code without-
model = get_peft_model(model, config)
the model trains without any issues. However, using the model returned from get_peft_model raises the following error due to datasets-
IndexError: Invalid key: 47682 is out of bounds for size 0.
I had raised this in https://github.com/huggingface/peft/issues/1299#issue-2056173386 and they suggested that I raise it here.
Here is the complete error-
IndexError Traceback (most recent call last)
in <cell line: 1>()
----> 1 trainer.train()
11 frames
[/usr/local/lib/python3.10/dist-packages/transformers/trainer.py](https://localhost:8080/#) in train(self, resume_from_checkpoint, trial, ignore_keys_for_eval, **kwargs)
1553 hf_hub_utils.enable_progress_bars()
1554 else:
-> 1555 return inner_training_loop(
1556 args=args,
1557 resume_from_checkpoint=resume_from_checkpoint,
[/usr/local/lib/python3.10/dist-packages/transformers/trainer.py](https://localhost:8080/#) in _inner_training_loop(self, batch_size, args, resume_from_checkpoint, trial, ignore_keys_for_eval)
1836
1837 step = -1
-> 1838 for step, inputs in enumerate(epoch_iterator):
1839 total_batched_samples += 1
1840 if rng_to_sync:
[/usr/local/lib/python3.10/dist-packages/accelerate/data_loader.py](https://localhost:8080/#) in iter(self)
446 # We iterate one batch ahead to check when we are at the end
447 try:
--> 448 current_batch = next(dataloader_iter)
449 except StopIteration:
450 yield
[/usr/local/lib/python3.10/dist-packages/torch/utils/data/dataloader.py](https://localhost:8080/#) in next(self)
628 # TODO(https://github.com/pytorch/pytorch/issues/76750)
629 self._reset() # type: ignore[call-arg]
--> 630 data = self._next_data()
631 self._num_yielded += 1
632 if self._dataset_kind == _DatasetKind.Iterable and \
[/usr/local/lib/python3.10/dist-packages/torch/utils/data/dataloader.py](https://localhost:8080/#) in _next_data(self)
672 def _next_data(self):
673 index = self._next_index() # may raise StopIteration
--> 674 data = self._dataset_fetcher.fetch(index) # may raise StopIteration
675 if self._pin_memory:
676 data = _utils.pin_memory.pin_memory(data, self._pin_memory_device)
[/usr/local/lib/python3.10/dist-packages/torch/utils/data/_utils/fetch.py](https://localhost:8080/#) in fetch(self, possibly_batched_index)
47 if self.auto_collation:
48 if hasattr(self.dataset, "getitems") and self.dataset.getitems:
---> 49 data = self.dataset.getitems(possibly_batched_index)
50 else:
51 data = [self.dataset[idx] for idx in possibly_batched_index]
[/usr/local/lib/python3.10/dist-packages/datasets/arrow_dataset.py](https://localhost:8080/#) in getitems(self, keys)
2802 def getitems(self, keys: List) -> List:
2803 """Can be used to get a batch using a list of integers indices."""
-> 2804 batch = self.getitem(keys)
2805 n_examples = len(batch[next(iter(batch))])
2806 return [{col: array[i] for col, array in batch.items()} for i in range(n_examples)]
[/usr/local/lib/python3.10/dist-packages/datasets/arrow_dataset.py](https://localhost:8080/#) in getitem(self, key)
2798 def getitem(self, key): # noqa: F811
2799 """Can be used to index columns (by string names) or rows (by integer index or iterable of indices or bools)."""
-> 2800 return self._getitem(key)
2801
2802 def getitems(self, keys: List) -> List:
[/usr/local/lib/python3.10/dist-packages/datasets/arrow_dataset.py](https://localhost:8080/#) in _getitem(self, key, **kwargs)
2782 format_kwargs = format_kwargs if format_kwargs is not None else {}
2783 formatter = get_formatter(format_type, features=self._info.features, **format_kwargs)
-> 2784 pa_subtable = query_table(self._data, key, indices=self._indices if self._indices is not None else None)
2785 formatted_output = format_table(
2786 pa_subtable, key, formatter=formatter, format_columns=format_columns, output_all_columns=output_all_columns
[/usr/local/lib/python3.10/dist-packages/datasets/formatting/formatting.py](https://localhost:8080/#) in query_table(table, key, indices)
581 else:
582 size = indices.num_rows if indices is not None else table.num_rows
--> 583 _check_valid_index_key(key, size)
584 # Query the main table
585 if indices is None:
[/usr/local/lib/python3.10/dist-packages/datasets/formatting/formatting.py](https://localhost:8080/#) in _check_valid_index_key(key, size)
534 elif isinstance(key, Iterable):
535 if len(key) > 0:
--> 536 _check_valid_index_key(int(max(key)), size=size)
537 _check_valid_index_key(int(min(key)), size=size)
538 else:
[/usr/local/lib/python3.10/dist-packages/datasets/formatting/formatting.py](https://localhost:8080/#) in _check_valid_index_key(key, size)
524 if isinstance(key, int):
525 if (key < 0 and key + size < 0) or (key >= size):
--> 526 raise IndexError(f"Invalid key: {key} is out of bounds for size {size}")
527 return
528 elif isinstance(key, slice):
IndexError: Invalid key: 47682 is out of bounds for size 0
### Steps to reproduce the bug
device = "cuda:0" if torch.cuda.is_available() else "cpu"
#defining model name for tokenizer and model loading
model_name= "t5-small"
#loading the tokenizer
tokenizer = AutoTokenizer.from_pretrained(model_name)
def preprocess_function(data, tokenizer):
inputs = [f"Paraphrase this sentence: {doc}" for doc in data["text"]]
model_inputs = tokenizer(inputs, max_length=150, truncation=True)
labels = [ast.literal_eval(i)[0] for i in data['paraphrases']]
labels = tokenizer(labels, max_length=150, truncation=True)
model_inputs["labels"] = labels["input_ids"]
return model_inputs
train_dataset = load_dataset("humarin/chatgpt-paraphrases", split="train").shuffle(seed=42).select(range(50000))
val_dataset = load_dataset("humarin/chatgpt-paraphrases", split="train").shuffle(seed=42).select(range(50000,55000))
tokenized_train = train_dataset.map(lambda batch: preprocess_function(batch, tokenizer), batched=True)
tokenized_val = val_dataset.map(lambda batch: preprocess_function(batch, tokenizer), batched=True)
def print_trainable_parameters(model):
"""
Prints the number of trainable parameters in the model.
"""
trainable_params = 0
all_param = 0
for _, param in model.named_parameters():
all_param += param.numel()
if param.requires_grad:
trainable_params += param.numel()
print(
f"trainable params: {trainable_params} || all params: {all_param} || trainable%: {100 * trainable_params / all_param}"
)
config = LoraConfig(
r=16, #attention heads
lora_alpha=32, #alpha scaling
lora_dropout=0.05,
bias="none",
task_type="Seq2Seq"
)
#loading the model
model = AutoModelForSeq2SeqLM.from_pretrained(model_name).to(device)
model = get_peft_model(model, config)
print_trainable_parameters(model)
#loading the data collator
data_collator = DataCollatorForSeq2Seq(
tokenizer=tokenizer,
model=model,
label_pad_token_id=-100,
padding="longest"
)
#defining the training arguments
training_args = Seq2SeqTrainingArguments(
output_dir=os.getcwd(),
evaluation_strategy="epoch",
save_strategy="epoch",
learning_rate=2e-5,
per_device_train_batch_size=16,
per_device_eval_batch_size=16,
weight_decay=1e-3,
save_total_limit=3,
load_best_model_at_end=True,
num_train_epochs=1,
predict_with_generate=True
)
def compute_metric_with_extra(tokenizer):
def compute_metrics(eval_preds):
metric = evaluate.load('rouge')
preds, labels = eval_preds
# decode preds and labels
labels = np.where(labels != -100, labels, tokenizer.pad_token_id)
decoded_preds = tokenizer.batch_decode(preds, skip_special_tokens=True)
decoded_labels = tokenizer.batch_decode(labels, skip_special_tokens=True)
# rougeLSum expects newline after each sentence
decoded_preds = ["\n".join(nltk.sent_tokenize(pred.strip())) for pred in decoded_preds]
decoded_labels = ["\n".join(nltk.sent_tokenize(label.strip())) for label in decoded_labels]
result = metric.compute(predictions=decoded_preds, references=decoded_labels, use_stemmer=True)
return result
return compute_metrics
trainer = Seq2SeqTrainer(
model=model,
args=training_args,
train_dataset=tokenized_train,
eval_dataset=tokenized_val,
tokenizer=tokenizer,
data_collator=data_collator,
compute_metrics= compute_metric_with_extra(tokenizer)
)
trainer.train()
### Expected behavior
I would want the trainer to train normally as it was before I used-
model = get_peft_model(model, config)
### Environment info
datasets version- 2.16.0
peft version- 0.7.1
transformers version- 4.35.2
accelerate version- 0.25.0
python- 3.10.12
enviroment- google colab
|
open
|
https://github.com/huggingface/datasets/issues/6535
| 2023-12-26T10:14:33
| 2024-02-05T08:42:31
| null |
{
"login": "MahavirDabas18",
"id": 57484266,
"type": "User"
}
|
[] | false
|
[] |
2,056,002,548
| 6,534
|
How to configure multiple folders in the same zip package
|
How should I write "config" in readme when all the data, such as train test, is in a zip file
train floder and test floder in data.zip
|
open
|
https://github.com/huggingface/datasets/issues/6534
| 2023-12-26T03:56:20
| 2023-12-26T06:31:16
| null |
{
"login": "d710055071",
"id": 12895488,
"type": "User"
}
|
[] | false
|
[] |
2,055,929,101
| 6,533
|
ted_talks_iwslt | Error: Config name is missing
|
### Describe the bug
Running load_dataset using the newest `datasets` library like below on the ted_talks_iwslt using year pair data will throw an error "Config name is missing"
see also:
https://huggingface.co/datasets/ted_talks_iwslt/discussions/3
likely caused by #6493, where the `and not config_kwargs` part in the if logic was removed
https://github.com/huggingface/datasets/blob/ef3b5dd3633995c95d77f35fb17f89ff44990bc4/src/datasets/builder.py#L512
### Steps to reproduce the bug
run:
```python
load_dataset("ted_talks_iwslt", language_pair=("ja", "en"), year="2015")
```
### Expected behavior
Load the data without error
### Environment info
datasets 2.16.0
|
closed
|
https://github.com/huggingface/datasets/issues/6533
| 2023-12-26T00:38:18
| 2023-12-30T18:58:21
| 2023-12-30T16:09:50
|
{
"login": "rayliuca",
"id": 35850903,
"type": "User"
}
|
[] | false
|
[] |
2,055,631,201
| 6,532
|
[Feature request] Indexing datasets by a customly-defined id field to enable random access dataset items via the id
|
### Feature request
Some datasets may contain an id-like field, for example the `id` field in [wikimedia/wikipedia](https://huggingface.co/datasets/wikimedia/wikipedia) and the `_id` field in [BeIR/dbpedia-entity](https://huggingface.co/datasets/BeIR/dbpedia-entity). HF datasets support efficient random access via row, but not via this kinds of id fields. I wonder if it is possible to add support for indexing by a custom "id-like" field to enable random access via such ids. The ids may be numbers or strings.
### Motivation
In some cases, especially during inference/evaluation, I may want to find out the item that has a specified id, defined by the dataset itself.
For example, in a typical re-ranking setting in information retrieval, the user may want to re-rank the set of candidate documents of each query. The input is usually presented in a TREC-style run file, with the following format:
```
<qid> Q0 <docno> <rank> <score> <tag>
```
The re-ranking program should be able to fetch the queries and documents according to the `<qid>` and `<docno>`, which are the original id defined in the query/document datasets. To accomplish this, I have to iterate over the whole HF dataset to get the mapping from real ids to row ids every time I start the program, which is time-consuming. Thus I want HF dataset to provide options for users to index by a custom id column, not by row.
### Your contribution
I'm not an expert in this project and I'm afraid that I'm not able to make contributions on the code.
|
open
|
https://github.com/huggingface/datasets/issues/6532
| 2023-12-25T11:37:10
| 2025-05-05T13:25:24
| null |
{
"login": "Yu-Shi",
"id": 3377221,
"type": "User"
}
|
[
{
"name": "enhancement",
"color": "a2eeef"
}
] | false
|
[] |
2,055,201,605
| 6,531
|
Add polars compatibility
|
Hey there,
I've just finished adding support to convert and format to `polars.DataFrame`. This was in response to the open issue about integrating Polars [#3334](https://github.com/huggingface/datasets/issues/3334). Datasets can be switched to Polars format via `Dataset.set_format("polars")`. I've also included `to_polars` and `from_polars`. All polars functions are checked via config.POLARS_AVAILABLE.
A few notes:
This only supports `DataFrames` and not `LazyFrames`. This probably could be integrated fairly easily via `is_lazy` args in `set_format`, and `to_polars`.
Let me know your feedbacks.
|
closed
|
https://github.com/huggingface/datasets/pull/6531
| 2023-12-24T20:03:23
| 2024-03-08T19:29:25
| 2024-03-08T15:22:58
|
{
"login": "psmyth94",
"id": 11325244,
"type": "User"
}
|
[] | true
|
[] |
2,054,817,609
| 6,530
|
Impossible to save a mapped dataset to disk
|
### Describe the bug
I want to play around with different hyperparameters when training but don't want to re-map my dataset with 3 million samples each time for tens of hours when I [fully fine-tune SDXL](https://github.com/huggingface/diffusers/blob/main/examples/text_to_image/train_text_to_image_sdxl.py).
After I do the mapping like this:
```
train_dataset = train_dataset.map(compute_embeddings_fn, batched=True)
train_dataset = train_dataset.map(
compute_vae_encodings_fn,
batched=True,
batch_size=16,
)
```
and try to save it like this:
`train_dataset.save_to_disk("test")`
i get this error ([full traceback](https://pastebin.com/kq3vt739)):
```
TypeError: Object of type function is not JSON serializable
The format kwargs must be JSON serializable, but key 'transform' isn't.
```
But what is interesting is that pushing to hub works like that:
`train_dataset.push_to_hub("kopyl/mapped-833-icons-sdxl-1024-dataset", token=True)`
Here is the link of the pushed dataset: https://huggingface.co/datasets/kopyl/mapped-833-icons-sdxl-1024-dataset
### Steps to reproduce the bug
Here is the self-contained notebook:
https://colab.research.google.com/drive/1RtCsEMVcwWcMwlWURk_cj_9xUBHz065M?usp=sharing
### Expected behavior
It should be easily saved to disk
### Environment info
NVIDIA A100, Linux (NC24ads A100 v4 from Azure), CUDA 12.2.
[pip freeze](https://pastebin.com/QTNb6iru)
|
open
|
https://github.com/huggingface/datasets/issues/6530
| 2023-12-23T15:18:27
| 2023-12-24T09:40:30
| null |
{
"login": "kopyl",
"id": 17604849,
"type": "User"
}
|
[] | false
|
[] |
2,054,209,449
| 6,529
|
Impossible to only download a test split
|
I've spent a significant amount of time trying to locate the split object inside my _split_generators() custom function.
Then after diving [in the code](https://github.com/huggingface/datasets/blob/5ff3670c18ed34fa8ddfa70a9aa403ae6cc9ad54/src/datasets/load.py#L2558) I realized that `download_and_prepare` is executed before! split is passed to the dataset builder in `as_dataset`.
If I'm not missing something, this seems like bad design, for the following use case:
> Imagine there is a huge dataset that has an evaluation test set and you want to just download and run just to compare your method.
Is there a current workaround that can help me achieve the same result?
Thank you,
|
open
|
https://github.com/huggingface/datasets/issues/6529
| 2023-12-22T16:56:32
| 2024-02-02T00:05:04
| null |
{
"login": "ysig",
"id": 28439529,
"type": "User"
}
|
[] | false
|
[] |
2,053,996,494
| 6,528
|
set dev version
| null |
closed
|
https://github.com/huggingface/datasets/pull/6528
| 2023-12-22T14:23:18
| 2023-12-22T14:31:42
| 2023-12-22T14:25:34
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,053,966,748
| 6,527
|
Release: 2.16.0
| null |
closed
|
https://github.com/huggingface/datasets/pull/6527
| 2023-12-22T13:59:56
| 2023-12-22T14:24:12
| 2023-12-22T14:17:55
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,053,726,451
| 6,526
|
Preserve order of configs and splits when using Parquet exports
|
Preserve order of configs and splits, as defined in dataset infos.
Fix #6521.
|
closed
|
https://github.com/huggingface/datasets/pull/6526
| 2023-12-22T10:35:56
| 2023-12-22T11:42:22
| 2023-12-22T11:36:14
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,053,119,357
| 6,525
|
BBox type
|
see [internal discussion](https://huggingface.slack.com/archives/C02EK7C3SHW/p1703097195609209)
Draft to get some feedback on a possible `BBox` feature type that can be used to get object detection bounding boxes data in one format or another.
```python
>>> from datasets import load_dataset, BBox
>>> ds = load_dataset("svhn", "full_numbers", split="train")
>>> ds[0]
{
'image': <PIL.PngImagePlugin.PngImageFile image mode=RGB size=107x46 at 0x126409BE0>,
'digits': {'bbox': [[38, 1, 21, 40], [57, 3, 16, 40]], 'label': [4, 6]}
}
>>> ds = ds.rename_column("digits", "annotations").cast_column("annotations", BBox(format="coco"))
>>> ds[0]
{
'image': <PIL.PngImagePlugin.PngImageFile image mode=RGB size=107x46 at 0x147730070>,
'annotations': [{'bbox': [38, 1, 21, 40], 'category_id': 4}, {'bbox': [57, 3, 16, 40], 'category_id': 6}]
}
```
note that it's a type for a list of bounding boxes, not just one - which would be needed to switch from a format to another using type casting.
|
closed
|
https://github.com/huggingface/datasets/pull/6525
| 2023-12-21T22:13:27
| 2024-01-11T06:34:51
| 2023-12-21T22:39:27
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,053,076,311
| 6,524
|
Streaming the Pile: Missing Files
|
### Describe the bug
The pile does not stream, a "File not Found error" is returned. It looks like the Pile's files have been moved.
### Steps to reproduce the bug
To reproduce run the following code:
```
from datasets import load_dataset
dataset = load_dataset('EleutherAI/pile', 'en', split='train', streaming=True)
next(iter(dataset))
```
I get the following error:
`FileNotFoundError: https://the-eye.eu/public/AI/pile/train/00.jsonl.zst`
### Expected behavior
Return the data in a stream.
### Environment info
- `datasets` version: 2.12.0
- Platform: Windows-10-10.0.22621-SP0
- Python version: 3.11.5
- Huggingface_hub version: 0.15.1
- PyArrow version: 11.0.0
- Pandas version: 2.0.3
|
closed
|
https://github.com/huggingface/datasets/issues/6524
| 2023-12-21T21:25:09
| 2023-12-22T09:17:05
| 2023-12-22T09:17:05
|
{
"login": "FelixLabelle",
"id": 23347756,
"type": "User"
}
|
[] | false
|
[] |
2,052,643,484
| 6,523
|
fix tests
| null |
closed
|
https://github.com/huggingface/datasets/pull/6523
| 2023-12-21T15:36:21
| 2023-12-21T15:56:54
| 2023-12-21T15:50:38
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,052,332,528
| 6,522
|
Loading HF Hub Dataset (private org repo) fails to load all features
|
### Describe the bug
When pushing a `Dataset` with multiple `Features` (`input`, `output`, `tags`) to Huggingface Hub (private org repo), and later downloading the `Dataset`, only `input` and `output` load - I believe the expected behavior is for all `Features` to be loaded by default?
### Steps to reproduce the bug
Pushing the data. `data_concat` is a `list` of `dict`s.
```python
for datum in data_concat:
datum_tags = {d["key"]: d["value"] for d in datum["tags"]}
split_fraction = # some logic that generates a train/test split number
if split_faction < test_fraction:
data_test.append(datum)
else:
data_train.append(datum)
dataset = DatasetDict(
{
"train": Dataset.from_list(data_train),
"test": Dataset.from_list(data_test),
"full": Dataset.from_list(data_concat),
},
)
dataset_shuffled = dataset.shuffle(seed=shuffle_seed)
dataset_shuffled.push_to_hub(
repo_id=hf_repo_id,
private=True,
config_name=m,
revision=revision,
token=hf_token,
)
```
Loading it later:
```python
dataset = datasets.load_dataset(
path=hf_repo_id,
name=name,
token=hf_token,
)
```
Produces:
```
DatasetDict({
train: Dataset({
features: ['input', 'output'],
num_rows: <obfuscated>
})
test: Dataset({
features: ['input', 'output'],
num_rows: <obfuscated>
})
full: Dataset({
features: ['input', 'output'],
num_rows: <obfuscated>
})
})
```
### Expected behavior
The expected result is below:
```
DatasetDict({
train: Dataset({
features: ['input', 'output', 'tags'],
num_rows: <obfuscated>
})
test: Dataset({
features: ['input', 'output', 'tags'],
num_rows: <obfuscated>
})
full: Dataset({
features: ['input', 'output', 'tags'],
num_rows: <obfuscated>
})
})
```
My workaround is as follows:
```python
dsinfo = datasets.get_dataset_config_info(
path=data_files,
config_name=data_config,
token=hf_token,
)
allfeatures = dsinfo.features.copy()
if "tags" not in allfeatures:
allfeatures["tags"] = [{"key": Value(dtype="string", id=None), "value": Value(dtype="string", id=None)}]
dataset = datasets.load_dataset(
path=data_files,
name=data_config,
features=allfeatures,
token=hf_token,
)
```
Interestingly enough (and perhaps a related bug?), if I don't add the `tags` to `allfeatures` above (i.e. only loading `input` and `output`), it throws an error when executing `load_dataset`:
```
ValueError: Couldn't cast
tags: list<element: struct<key: string, value: string>>
child 0, element: struct<key: string, value: string>
child 0, key: string
child 1, value: string
input: <obfuscated>
output: <obfuscated>
-- schema metadata --
huggingface: '{"info": {"features": {"tags": [{"key": {"dtype": "string",' + 532
to
{'input': <obfuscated>, 'output': <obfuscated>
because column names don't match
```
Traceback for this:
```
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/Users/bt/github/core/.venv/lib/python3.11/site-packages/datasets/load.py", line 2152, in load_dataset
builder_instance.download_and_prepare(
File "/Users/bt/github/core/.venv/lib/python3.11/site-packages/datasets/builder.py", line 948, in download_and_prepare
self._download_and_prepare(
File "/Users/bt/github/core/.venv/lib/python3.11/site-packages/datasets/builder.py", line 1043, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/Users/bt/github/core/.venv/lib/python3.11/site-packages/datasets/builder.py", line 1805, in _prepare_split
for job_id, done, content in self._prepare_split_single(
File "/Users/bt/github/core/.venv/lib/python3.11/site-packages/datasets/builder.py", line 1950, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.builder.DatasetGenerationError: An error occurred while generating the dataset
```
### Environment info
- `datasets` version: 2.15.0
- Platform: macOS-14.0-arm64-arm-64bit
- Python version: 3.11.5
- `huggingface_hub` version: 0.19.4
- PyArrow version: 14.0.1
- Pandas version: 2.1.4
- `fsspec` version: 2023.10.0
|
open
|
https://github.com/huggingface/datasets/issues/6522
| 2023-12-21T12:26:35
| 2023-12-21T13:24:31
| null |
{
"login": "versipellis",
"id": 6579034,
"type": "User"
}
|
[] | false
|
[] |
2,052,229,538
| 6,521
|
The order of the splits is not preserved
|
We had a regression and the order of the splits is not preserved. They are alphabetically sorted, instead of preserving original "train", "validation", "test" order.
Check: In branch "main"
```python
In [9]: dataset = load_dataset("adversarial_qa", '"adversarialQA")
In [10]: dataset
Out[10]:
DatasetDict({
test: Dataset({
features: ['id', 'title', 'context', 'question', 'answers', 'metadata'],
num_rows: 3000
})
train: Dataset({
features: ['id', 'title', 'context', 'question', 'answers', 'metadata'],
num_rows: 30000
})
validation: Dataset({
features: ['id', 'title', 'context', 'question', 'answers', 'metadata'],
num_rows: 3000
})
})
```
Before (2.15.0) it was:
```python
DatasetDict({
train: Dataset({
features: ['id', 'title', 'context', 'question', 'answers', 'metadata'],
num_rows: 30000
})
validation: Dataset({
features: ['id', 'title', 'context', 'question', 'answers', 'metadata'],
num_rows: 3000
})
test: Dataset({
features: ['id', 'title', 'context', 'question', 'answers', 'metadata'],
num_rows: 3000
})
})
```
See issues:
- https://huggingface.co/datasets/adversarial_qa/discussions/3
- https://huggingface.co/datasets/beans/discussions/4
This is a regression because it was previously fixed. See:
- #6196
- #5728
|
closed
|
https://github.com/huggingface/datasets/issues/6521
| 2023-12-21T11:17:27
| 2023-12-22T11:36:15
| 2023-12-22T11:36:15
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[
{
"name": "bug",
"color": "d73a4a"
}
] | false
|
[] |
2,052,059,078
| 6,520
|
Support commit_description parameter in push_to_hub
|
Support `commit_description` parameter in `push_to_hub`.
CC: @Wauplin
|
closed
|
https://github.com/huggingface/datasets/pull/6520
| 2023-12-21T09:36:11
| 2023-12-21T14:49:47
| 2023-12-21T14:43:35
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,050,759,824
| 6,519
|
Support push_to_hub canonical datasets
|
Support `push_to_hub` canonical datasets.
This is necessary in the Space to convert script-datasets to Parquet: https://huggingface.co/spaces/albertvillanova/convert-dataset-to-parquet
Note that before this PR, the `repo_id` "dataset_name" was transformed to "user/dataset_name". This behavior was introduced by:
- #6269
|
closed
|
https://github.com/huggingface/datasets/pull/6519
| 2023-12-20T15:16:45
| 2023-12-21T14:48:20
| 2023-12-21T14:40:57
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,050,137,038
| 6,518
|
fix get_metadata_patterns function args error
|
Bug get_metadata_patterns arg error https://github.com/huggingface/datasets/issues/6517
|
closed
|
https://github.com/huggingface/datasets/pull/6518
| 2023-12-20T09:06:22
| 2023-12-21T15:14:17
| 2023-12-21T15:07:57
|
{
"login": "d710055071",
"id": 12895488,
"type": "User"
}
|
[] | true
|
[] |
2,050,121,588
| 6,517
|
Bug get_metadata_patterns arg error
|
https://github.com/huggingface/datasets/blob/3f149204a2a5948287adcade5e90707aa5207a92/src/datasets/load.py#L1240C1-L1240C69
metadata_patterns = get_metadata_patterns(base_path, download_config=self.download_config)
|
closed
|
https://github.com/huggingface/datasets/issues/6517
| 2023-12-20T08:56:44
| 2023-12-22T00:24:23
| 2023-12-22T00:24:23
|
{
"login": "d710055071",
"id": 12895488,
"type": "User"
}
|
[] | false
|
[] |
2,050,033,322
| 6,516
|
Support huggingface-hub pre-releases
|
Support `huggingface-hub` pre-releases.
This way we will have our CI green when testing `huggingface-hub` release candidates. See: https://github.com/huggingface/datasets/tree/ci-test-huggingface-hub-v0.20.0.rc1
Close #6513.
|
closed
|
https://github.com/huggingface/datasets/pull/6516
| 2023-12-20T07:52:29
| 2023-12-20T08:51:34
| 2023-12-20T08:44:44
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,049,724,251
| 6,515
|
Why call http_head() when fsspec_head() succeeds
|
https://github.com/huggingface/datasets/blob/a91582de288d98e94bcb5ab634ca1cfeeff544c5/src/datasets/utils/file_utils.py#L510C1-L523C14
|
closed
|
https://github.com/huggingface/datasets/issues/6515
| 2023-12-20T02:25:51
| 2023-12-26T05:35:46
| 2023-12-26T05:35:46
|
{
"login": "d710055071",
"id": 12895488,
"type": "User"
}
|
[] | false
|
[] |
2,049,600,663
| 6,514
|
Cache backward compatibility with 2.15.0
|
...for datasets without scripts
It takes into account the changes in cache from
- https://github.com/huggingface/datasets/pull/6493: switch to `config/version/commit_sha` schema
- https://github.com/huggingface/datasets/pull/6454: fix `DataFilesDict` keys ordering when hashing
requires https://github.com/huggingface/datasets/pull/6493 to be merged
|
closed
|
https://github.com/huggingface/datasets/pull/6514
| 2023-12-19T23:52:25
| 2023-12-21T21:14:11
| 2023-12-21T21:07:55
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,048,869,151
| 6,513
|
Support huggingface-hub 0.20.0
|
CI to test the support of `huggingface-hub` 0.20.0: https://github.com/huggingface/datasets/compare/main...ci-test-huggingface-hub-v0.20.0.rc1
We need to merge:
- #6510
- #6512
- #6516
|
closed
|
https://github.com/huggingface/datasets/issues/6513
| 2023-12-19T15:15:46
| 2023-12-20T08:44:45
| 2023-12-20T08:44:45
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | false
|
[] |
2,048,795,819
| 6,512
|
Remove deprecated HfFolder
|
...and use `huggingface_hub.get_token()` instead
|
closed
|
https://github.com/huggingface/datasets/pull/6512
| 2023-12-19T14:40:49
| 2023-12-19T20:21:13
| 2023-12-19T20:14:30
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,048,465,958
| 6,511
|
Implement get dataset default config name
|
Implement `get_dataset_default_config_name`.
Now that we support setting a configuration as default in `push_to_hub` (see #6500), we need a programmatically way to know in advance which is the default configuration. This will be used in the Space to convert script-datasets to Parquet: https://huggingface.co/spaces/albertvillanova/convert-dataset-to-parquet
Follow-up of:
- #6500
CC: @severo
|
closed
|
https://github.com/huggingface/datasets/pull/6511
| 2023-12-19T11:26:19
| 2023-12-21T14:48:57
| 2023-12-21T14:42:41
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,046,928,742
| 6,510
|
Replace `list_files_info` with `list_repo_tree` in `push_to_hub`
|
Starting from `huggingface_hub` 0.20.0, `list_files_info` will be deprecated in favor of `list_repo_tree` (see https://github.com/huggingface/huggingface_hub/pull/1910)
|
closed
|
https://github.com/huggingface/datasets/pull/6510
| 2023-12-18T15:34:19
| 2023-12-19T18:05:47
| 2023-12-19T17:58:34
|
{
"login": "mariosasko",
"id": 47462742,
"type": "User"
}
|
[] | true
|
[] |
2,046,720,869
| 6,509
|
Better cast error when generating dataset
|
I want to improve the error message for datasets like https://huggingface.co/datasets/m-a-p/COIG-CQIA
Cc @albertvillanova @severo is this new error ok ? Or should I use a dedicated error class ?
New:
```python
Traceback (most recent call last):
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1920, in _prepare_split_single
writer.write_table(table)
File "/Users/quentinlhoest/hf/datasets/src/datasets/arrow_writer.py", line 574, in write_table
pa_table = table_cast(pa_table, self._schema)
File "/Users/quentinlhoest/hf/datasets/src/datasets/table.py", line 2322, in table_cast
return cast_table_to_schema(table, schema)
File "/Users/quentinlhoest/hf/datasets/src/datasets/table.py", line 2276, in cast_table_to_schema
raise CastError(
datasets.table.CastError: Couldn't cast
instruction: string
other: string
index: string
domain: list<item: string>
child 0, item: string
output: string
task_type: struct<major: list<item: string>, minor: list<item: string>>
child 0, major: list<item: string>
child 0, item: string
child 1, minor: list<item: string>
child 0, item: string
task_name_in_eng: string
input: string
to
{'answer_from': Value(dtype='string', id=None), 'instruction': Value(dtype='string', id=None), 'human_verified': Value(dtype='bool', id=None), 'domain': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'output': Value(dtype='string', id=None), 'task_type': {'major': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'minor': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}, 'copyright': Value(dtype='string', id=None), 'input': Value(dtype='string', id=None)}
because column names don't match
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/Users/quentinlhoest/hf/datasets/playground/ttest.py", line 74, in <module>
load_dataset("m-a-p/COIG-CQIA")
File "/Users/quentinlhoest/hf/datasets/src/datasets/load.py", line 2529, in load_dataset
builder_instance.download_and_prepare(
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 936, in download_and_prepare
self._download_and_prepare(
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1031, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1791, in _prepare_split
for job_id, done, content in self._prepare_split_single(
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1922, in _prepare_split_single
raise DatasetGenerationCastError.from_cast_error(
datasets.exceptions.DatasetGenerationCastError: An error occurred while generating the dataset
All the data files must have the same columns, but at some point there are 3 new columns (other, index, task_name_in_eng) and 3 missing columns (answer_from, copyright, human_verified).
This happened while the json dataset builder was generating data using
hf://datasets/m-a-p/COIG-CQIA/coig_pc/coig_pc_core_sample.json (at revision b7b7ecf290f6515036c7c04bd8537228ac2eb474)
Please either edit the data files to have matching columns, or separate them into different configurations (see docs at https://hf.co/docs/hub/datasets-manual-configuration#multiple-configurations)
```
Previously:
```python
Traceback (most recent call last):
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1931, in _prepare_split_single
writer.write_table(table)
File "/Users/quentinlhoest/hf/datasets/src/datasets/arrow_writer.py", line 574, in write_table
pa_table = table_cast(pa_table, self._schema)
File "/Users/quentinlhoest/hf/datasets/src/datasets/table.py", line 2295, in table_cast
return cast_table_to_schema(table, schema)
File "/Users/quentinlhoest/hf/datasets/src/datasets/table.py", line 2253, in cast_table_to_schema
raise ValueError(f"Couldn't cast\n{table.schema}\nto\n{features}\nbecause column names don't match")
ValueError: Couldn't cast
task_type: struct<major: list<item: string>, minor: list<item: string>>
child 0, major: list<item: string>
child 0, item: string
child 1, minor: list<item: string>
child 0, item: string
other: string
instruction: string
task_name_in_eng: string
domain: list<item: string>
child 0, item: string
index: string
output: string
input: string
to
{'human_verified': Value(dtype='bool', id=None), 'task_type': {'major': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'minor': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None)}, 'answer_from': Value(dtype='string', id=None), 'copyright': Value(dtype='string', id=None), 'instruction': Value(dtype='string', id=None), 'domain': Sequence(feature=Value(dtype='string', id=None), length=-1, id=None), 'output': Value(dtype='string', id=None), 'input': Value(dtype='string', id=None)}
because column names don't match
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "/Users/quentinlhoest/hf/datasets/playground/ttest.py", line 74, in <module>
load_dataset("m-a-p/COIG-CQIA")
File "/Users/quentinlhoest/hf/datasets/src/datasets/load.py", line 2529, in load_dataset
builder_instance.download_and_prepare(
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 949, in download_and_prepare
self._download_and_prepare(
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1044, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1804, in _prepare_split
for job_id, done, content in self._prepare_split_single(
File "/Users/quentinlhoest/hf/datasets/src/datasets/builder.py", line 1949, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.builder.DatasetGenerationError: An error occurred while generating the dataset
```
|
closed
|
https://github.com/huggingface/datasets/pull/6509
| 2023-12-18T13:57:24
| 2023-12-19T09:37:12
| 2023-12-19T09:31:03
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,045,733,273
| 6,508
|
Read GeoParquet files using parquet reader
|
Let GeoParquet files with the file extension `*.geoparquet` or `*.gpq` be readable by the default parquet reader.
Those two file extensions are the ones most commonly used for GeoParquet files, and is included in the `gpq` validator tool at https://github.com/planetlabs/gpq/blob/e5576b4ee7306b4d2259d56c879465a9364dab90/cmd/gpq/command/convert.go#L73-L75
Addresses https://github.com/huggingface/datasets/issues/6438
|
closed
|
https://github.com/huggingface/datasets/pull/6508
| 2023-12-18T04:50:37
| 2024-01-26T18:22:35
| 2024-01-26T16:18:41
|
{
"login": "weiji14",
"id": 23487320,
"type": "User"
}
|
[] | true
|
[] |
2,045,152,928
| 6,507
|
where is glue_metric.py> @Frankie123421 what was the resolution to this?
|
> @Frankie123421 what was the resolution to this?
use glue_metric.py instead of glue.py in load_metric
_Originally posted by @Frankie123421 in https://github.com/huggingface/datasets/issues/2117#issuecomment-905093763_
|
closed
|
https://github.com/huggingface/datasets/issues/6507
| 2023-12-17T09:58:25
| 2023-12-18T11:42:49
| 2023-12-18T11:42:49
|
{
"login": "Mcccccc1024",
"id": 119146162,
"type": "User"
}
|
[] | false
|
[] |
2,044,975,038
| 6,506
|
Incorrect test set labels for RTE and CoLA datasets via load_dataset
|
### Describe the bug
The test set labels for the RTE and CoLA datasets when loading via datasets load_dataset are all -1.
Edit: It appears this is also the case for every other dataset except for MRPC (stsb, sst2, qqp, mnli (both matched and mismatched), qnli, wnli, ax). Is this intended behavior to safeguard the test set for evaluation purposes?
### Steps to reproduce the bug
!pip install datasets
from datasets import load_dataset
rte_data = load_dataset('glue', 'rte')
cola_data = load_dataset('glue', 'cola')
print(rte_data['test'][0:30]['label'])
print(cola_data['test'][0:30]['label'])
Output:
[-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1]
[-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1]
The non-label test data seems to be fine:
e.g. rte_data['test'][1] is:
{'sentence1': "Authorities in Brazil say that more than 200 people are being held hostage in a prison in the country's remote, Amazonian-jungle state of Rondonia.",
'sentence2': 'Authorities in Brazil hold 200 people as hostage.',
'label': -1,
'idx': 1}
Training and validation data are also fine:
e.g. rte_data['train][0] is:
{'sentence1': 'No Weapons of Mass Destruction Found in Iraq Yet.',
'sentence2': 'Weapons of Mass Destruction Found in Iraq.',
'label': 1,
'idx': 0}
### Expected behavior
Expected the labels to be binary 0/1 values; Got all -1s instead
### Environment info
- `datasets` version: 2.15.0
- Platform: Linux-6.1.58+-x86_64-with-glibc2.35
- Python version: 3.10.12
- `huggingface_hub` version: 0.19.4
- PyArrow version: 10.0.1
- Pandas version: 1.5.3
- `fsspec` version: 2023.6.0
|
closed
|
https://github.com/huggingface/datasets/issues/6506
| 2023-12-16T22:06:08
| 2023-12-21T09:57:57
| 2023-12-21T09:57:57
|
{
"login": "emreonal11",
"id": 73316684,
"type": "User"
}
|
[] | false
|
[] |
2,044,721,288
| 6,505
|
Got stuck when I trying to load a dataset
|
### Describe the bug
Hello, everyone. I met a problem when I am trying to load a data file using load_dataset method on a Debian 10 system. The data file is not very large, only 1.63MB with 600 records.
Here is my code:
from datasets import load_dataset
dataset = load_dataset('json', data_files='mypath/oaast_rm_zh.json')
I waited it for 20 minutes. It still no response. I cannot using Ctrl+C to cancel the command. I have to use Ctrl+Z to kill it. I also try it with a txt file, it still no response in a long time.
I can load the same file successfully using my laptop (windows 10, python 3.8.5, datasets==2.14.5). I can also make it on another computer (Ubuntu 20.04.5 LTS, python 3.10.13, datasets 2.14.7). It only takes me 1-2 miniutes.
Could you give me some suggestions? Thank you.
### Steps to reproduce the bug
from datasets import load_dataset
dataset = load_dataset('json', data_files='mypath/oaast_rm_zh.json')
### Expected behavior
I hope it can load the file successfully.
### Environment info
OS: Debian GNU/Linux 10
Python: Python 3.10.13
Pip list:
Package Version
------------------------- ------------
accelerate 0.25.0
addict 2.4.0
aiofiles 23.2.1
aiohttp 3.9.1
aiosignal 1.3.1
aliyun-python-sdk-core 2.14.0
aliyun-python-sdk-kms 2.16.2
altair 5.2.0
annotated-types 0.6.0
anyio 3.7.1
async-timeout 4.0.3
attrs 23.1.0
certifi 2023.11.17
cffi 1.16.0
charset-normalizer 3.3.2
click 8.1.7
contourpy 1.2.0
crcmod 1.7
cryptography 41.0.7
cycler 0.12.1
datasets 2.14.7
dill 0.3.7
docstring-parser 0.15
einops 0.7.0
exceptiongroup 1.2.0
fastapi 0.105.0
ffmpy 0.3.1
filelock 3.13.1
fonttools 4.46.0
frozenlist 1.4.1
fsspec 2023.10.0
gast 0.5.4
gradio 3.50.2
gradio_client 0.6.1
h11 0.14.0
httpcore 1.0.2
httpx 0.25.2
huggingface-hub 0.19.4
idna 3.6
importlib-metadata 7.0.0
importlib-resources 6.1.1
jieba 0.42.1
Jinja2 3.1.2
jmespath 0.10.0
joblib 1.3.2
jsonschema 4.20.0
jsonschema-specifications 2023.11.2
kiwisolver 1.4.5
markdown-it-py 3.0.0
MarkupSafe 2.1.3
matplotlib 3.8.2
mdurl 0.1.2
modelscope 1.10.0
mpmath 1.3.0
multidict 6.0.4
multiprocess 0.70.15
networkx 3.2.1
nltk 3.8.1
numpy 1.26.2
nvidia-cublas-cu12 12.1.3.1
nvidia-cuda-cupti-cu12 12.1.105
nvidia-cuda-nvrtc-cu12 12.1.105
nvidia-cuda-runtime-cu12 12.1.105
nvidia-cudnn-cu12 8.9.2.26
nvidia-cufft-cu12 11.0.2.54
nvidia-curand-cu12 10.3.2.106
nvidia-cusolver-cu12 11.4.5.107
nvidia-cusparse-cu12 12.1.0.106
nvidia-nccl-cu12 2.18.1
nvidia-nvjitlink-cu12 12.3.101
nvidia-nvtx-cu12 12.1.105
orjson 3.9.10
oss2 2.18.3
packaging 23.2
pandas 2.1.4
peft 0.7.1
Pillow 10.1.0
pip 23.3.1
platformdirs 4.1.0
protobuf 4.25.1
psutil 5.9.6
pyarrow 14.0.1
pyarrow-hotfix 0.6
pycparser 2.21
pycryptodome 3.19.0
pydantic 2.5.2
pydantic_core 2.14.5
pydub 0.25.1
Pygments 2.17.2
pyparsing 3.1.1
python-dateutil 2.8.2
python-multipart 0.0.6
pytz 2023.3.post1
PyYAML 6.0.1
referencing 0.32.0
regex 2023.10.3
requests 2.31.0
rich 13.7.0
rouge-chinese 1.0.3
rpds-py 0.13.2
safetensors 0.4.1
scipy 1.11.4
semantic-version 2.10.0
sentencepiece 0.1.99
setuptools 68.2.2
shtab 1.6.5
simplejson 3.19.2
six 1.16.0
sniffio 1.3.0
sortedcontainers 2.4.0
sse-starlette 1.8.2
starlette 0.27.0
sympy 1.12
tiktoken 0.5.2
tokenizers 0.15.0
tomli 2.0.1
toolz 0.12.0
torch 2.1.2
tqdm 4.66.1
transformers 4.36.1
triton 2.1.0
trl 0.7.4
typing_extensions 4.9.0
tyro 0.6.0
tzdata 2023.3
urllib3 2.1.0
uvicorn 0.24.0.post1
websockets 11.0.3
wheel 0.41.2
xxhash 3.4.1
yapf 0.40.2
yarl 1.9.4
zipp 3.17.0
|
open
|
https://github.com/huggingface/datasets/issues/6505
| 2023-12-16T11:51:07
| 2024-12-24T16:45:52
| null |
{
"login": "yirenpingsheng",
"id": 18232551,
"type": "User"
}
|
[] | false
|
[] |
2,044,541,154
| 6,504
|
Error Pushing to Hub
|
### Describe the bug
Error when trying to push a dataset in a special format to hub
### Steps to reproduce the bug
```
import datasets
from datasets import Dataset
dataset_dict = {
"filename": ["apple", "banana"],
"token": [[[1,2],[3,4]],[[1,2],[3,4]]],
"label": [0, 1],
}
dataset = Dataset.from_dict(dataset_dict)
dataset = dataset.cast_column("token", datasets.features.features.Array2D(shape=(2, 2),dtype="int16"))
dataset.push_to_hub("SequenceModel/imagenet_val_256")
```
Error:
```
...
ConstructorError: could not determine a constructor for the tag 'tag:yaml.org,2002:python/tuple'
in "<unicode string>", line 8, column 16:
shape: !!python/tuple
^
```
### Expected behavior
Dataset being pushed to hub
### Environment info
- `datasets` version: 2.15.0
- Platform: Linux-5.19.0-1022-gcp-x86_64-with-glibc2.35
- Python version: 3.11.5
- `huggingface_hub` version: 0.19.4
- PyArrow version: 14.0.1
- Pandas version: 2.1.4
- `fsspec` version: 2023.10.0
|
closed
|
https://github.com/huggingface/datasets/issues/6504
| 2023-12-16T01:05:22
| 2023-12-16T06:20:53
| 2023-12-16T06:20:53
|
{
"login": "Jiayi-Pan",
"id": 55055083,
"type": "User"
}
|
[] | false
|
[] |
2,043,847,591
| 6,503
|
Fix streaming xnli
|
This code was failing
```python
In [1]: from datasets import load_dataset
In [2]:
...: ds = load_dataset("xnli", "all_languages", split="test", streaming=True)
...:
...: sample_data = next(iter(ds))["premise"] # pick up one data
...: input_text = list(sample_data.values())
```
```
File ~/hf/datasets/src/datasets/features/translation.py:104, in TranslationVariableLanguages.encode_example(self, translation_dict)
102 return translation_dict
103 elif self.languages and set(translation_dict) - lang_set:
--> 104 raise ValueError(
105 f'Some languages in example ({", ".join(sorted(set(translation_dict) - lang_set))}) are not in valid set ({", ".join(lang_set)}).'
106 )
108 # Convert dictionary into tuples, splitting out cases where there are
109 # multiple translations for a single language.
110 translation_tuples = []
ValueError: Some languages in example (language, translation) are not in valid set (ur, fr, hi, sw, vi, el, de, th, en, tr, zh, ar, bg, ru, es).
```
because in streaming mode we expect features encode methods to be no-ops if the example is already encoded.
I fixed `TranslationVariableLanguages` to account for that
|
closed
|
https://github.com/huggingface/datasets/pull/6503
| 2023-12-15T14:40:57
| 2023-12-15T14:51:06
| 2023-12-15T14:44:47
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,043,771,731
| 6,502
|
Pickle support for `torch.Generator` objects
|
Fix for https://discuss.huggingface.co/t/caching-a-dataset-processed-with-randomness/65616
|
closed
|
https://github.com/huggingface/datasets/pull/6502
| 2023-12-15T13:55:12
| 2023-12-15T15:04:33
| 2023-12-15T14:58:22
|
{
"login": "mariosasko",
"id": 47462742,
"type": "User"
}
|
[] | true
|
[] |
2,043,377,240
| 6,501
|
OverflowError: value too large to convert to int32_t
|
### Describe the bug
### Steps to reproduce the bug
just loading datasets
### Expected behavior
how can I fix it
### Environment info
pip install /mnt/cluster/zhangfan/study_info/LLaMA-Factory/peft-0.6.0-py3-none-any.whl
pip install huggingface_hub-0.19.4-py3-none-any.whl tokenizers-0.15.0-cp310-cp310-manylinux_2_17_x86_64.manylinux2014_x86_64.whl transformers-4.36.1-py3-none-any.whl pyarrow_hotfix-0.6-py3-none-any.whl datasets-2.15.0-py3-none-any.whl tyro-0.5.18-py3-none-any.whl trl-0.7.4-py3-none-any.whl
done
|
open
|
https://github.com/huggingface/datasets/issues/6501
| 2023-12-15T10:10:21
| 2025-06-27T04:27:14
| null |
{
"login": "zhangfan-algo",
"id": 47747764,
"type": "User"
}
|
[] | false
|
[] |
2,043,258,633
| 6,500
|
Enable setting config as default when push_to_hub
|
Fix #6497.
|
closed
|
https://github.com/huggingface/datasets/pull/6500
| 2023-12-15T09:17:41
| 2023-12-18T11:56:11
| 2023-12-18T11:50:03
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,043,166,976
| 6,499
|
docs: add reference Git over SSH
|
see https://discuss.huggingface.co/t/update-datasets-getting-started-to-new-git-security/65893
|
closed
|
https://github.com/huggingface/datasets/pull/6499
| 2023-12-15T08:38:31
| 2023-12-15T11:48:47
| 2023-12-15T11:42:38
|
{
"login": "severo",
"id": 1676121,
"type": "User"
}
|
[] | true
|
[] |
2,042,075,969
| 6,498
|
Fallback on dataset script if user wants to load default config
|
Right now this code is failing on `main`:
```python
load_dataset("openbookqa")
```
This is because it tries to load the dataset from the Parquet export but the dataset has multiple configurations and the Parquet export doesn't know which one is the default one.
I fixed this by simply falling back on using the dataset script (which tells the user to pass `trust_remote_code=True`):
```python
load_dataset("openbookqa", trust_remote_code=True)
```
Note that if the user happened to specify a config name I don't fall back on the script since we can use the Parquet export in this case (no need to know which config is the default)
```python
load_dataset("openbookqa", "main")
```
|
closed
|
https://github.com/huggingface/datasets/pull/6498
| 2023-12-14T16:46:01
| 2023-12-15T13:16:56
| 2023-12-15T13:10:48
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,041,994,274
| 6,497
|
Support setting a default config name in push_to_hub
|
In order to convert script-datasets to no-script datasets, we need to support setting a default config name for those scripts that set one.
|
closed
|
https://github.com/huggingface/datasets/issues/6497
| 2023-12-14T15:59:03
| 2023-12-18T11:50:04
| 2023-12-18T11:50:04
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[
{
"name": "enhancement",
"color": "a2eeef"
}
] | false
|
[] |
2,041,589,386
| 6,496
|
Error when writing a dataset to HF Hub: A commit has happened since. Please refresh and try again.
|
**Describe the bug**
Getting a `412 Client Error: Precondition Failed` when trying to write a dataset to the HF hub.
```
huggingface_hub.utils._errors.HfHubHTTPError: 412 Client Error: Precondition Failed for url: https://huggingface.co/api/datasets/GLorr/test-dask/commit/main (Request ID: Root=1-657ae26f-3bd92bf861bb254b2cc0826c;50a09ab7-9347-406a-ba49-69f98abee9cc)
A commit has happened since. Please refresh and try again.
```
**Steps to reproduce the bug**
This is a minimal reproducer:
```
import dask.dataframe as dd
import pandas as pd
import random
import os
import huggingface_hub
import datasets
huggingface_hub.login(token=os.getenv("HF_TOKEN"))
data = {"number": [random.randint(0,10) for _ in range(1000)]}
df = pd.DataFrame.from_dict(data)
dataframe = dd.from_pandas(df, npartitions=1)
dataframe = dataframe.repartition(npartitions=3)
schema = datasets.Features({"number": datasets.Value("int64")}).arrow_schema
repo_id = "GLorr/test-dask"
repo_path = f"hf://datasets/{repo_id}"
huggingface_hub.create_repo(repo_id=repo_id, repo_type="dataset", exist_ok=True)
dd.to_parquet(dataframe, path=f"{repo_path}/data", schema=schema)
```
**Expected behavior**
Would expect to write to the hub without any problem.
**Environment info**
```
datasets==2.15.0
huggingface-hub==0.19.4
```
|
open
|
https://github.com/huggingface/datasets/issues/6496
| 2023-12-14T11:24:54
| 2023-12-14T12:22:21
| null |
{
"login": "GeorgesLorre",
"id": 35808396,
"type": "User"
}
|
[] | false
|
[] |
2,039,684,839
| 6,494
|
Image Data loaded Twice
|
### Describe the bug
When I learn from https://huggingface.co/docs/datasets/image_load and try to load image data from a folder. I noticed that the image was read twice in the returned data. As you can see in the attached image, there are only four images in the train folder, but reading brings up eight images
### Steps to reproduce the bug
from datasets import Dataset, load_dataset
dataset = load_dataset("imagefolder", data_dir="data/", drop_labels=False)
# print(dataset["train"][0]["image"] == dataset["train"][1]["image"])
print(dataset)
print(dataset["train"]["image"])
print(len(dataset["train"]["image"]))
### Expected behavior
DatasetDict({
train: Dataset({
features: ['image', 'label'],
num_rows: 8
})
})
[<PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=2877x2129 at 0x1BD1D1CA8B0>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=2877x2129 at 0x1BD1D2452E0>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=4208x3120 at 0x1BD1D245310>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=4208x3120 at 0x1BD1D2453A0>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=2877x2129 at 0x1BD1D245460>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=2877x2129 at 0x1BD1D245430>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=4208x3120 at 0x1BD1D2454F0>, <PIL.JpegImagePlugin.JpegImageFile image mode=RGB size=4208x3120 at 0x1BD1D245550>]
8
### Environment info
- `datasets` version: 2.14.5
- Platform: Windows-10-10.0.22621-SP0
- Python version: 3.9.17
- Huggingface_hub version: 0.19.4
- PyArrow version: 13.0.0
- Pandas version: 2.0.3
|
open
|
https://github.com/huggingface/datasets/issues/6494
| 2023-12-13T13:11:42
| 2023-12-13T13:11:42
| null |
{
"login": "ArcaneLex",
"id": 28867010,
"type": "User"
}
|
[] | false
|
[] |
2,039,708,529
| 6,495
|
Newline characters don't behave as expected when calling dataset.info
|
### System Info
- `transformers` version: 4.32.1
- Platform: Windows-10-10.0.19045-SP0
- Python version: 3.11.5
- Huggingface_hub version: 0.15.1
- Safetensors version: 0.3.2
- Accelerate version: not installed
- Accelerate config: not found
- PyTorch version (GPU?): 2.1.1+cpu (False)
- Tensorflow version (GPU?): 2.15.0 (False)
- Flax version (CPU?/GPU?/TPU?): not installed (NA)
- Jax version: not installed
- JaxLib version: not installed
- Using GPU in script?: no
- Using distributed or parallel set-up in script?: no
### Who can help?
@marios
### Information
- [X] The official example scripts
- [ ] My own modified scripts
### Tasks
- [X] An officially supported task in the `examples` folder (such as GLUE/SQuAD, ...)
- [ ] My own task or dataset (give details below)
### Reproduction
[Source](https://huggingface.co/docs/datasets/v2.2.1/en/access)
```
from datasets import load_dataset
dataset = load_dataset('glue', 'mrpc', split='train')
dataset.info
```
DatasetInfo(description='GLUE, the General Language Understanding Evaluation benchmark\n(https://gluebenchmark.com/) is a collection of resources for training,\nevaluating, and analyzing natural language understanding systems.\n\n', citation='@inproceedings{dolan2005automatically,\n title={Automatically constructing a corpus of sentential paraphrases},\n author={Dolan, William B and Brockett, Chris},\n booktitle={Proceedings of the Third International Workshop on Paraphrasing (IWP2005)},\n year={2005}\n}\n@inproceedings{wang2019glue,\n title={{GLUE}: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding},\n author={Wang, Alex and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R.},\n note={In the Proceedings of ICLR.},\n year={2019}\n}\n', homepage='https://www.microsoft.com/en-us/download/details.aspx?id=52398', license='', features={'sentence1': Value(dtype='string', id=None), 'sentence2': Value(dtype='string', id=None), 'label': ClassLabel(names=['not_equivalent', 'equivalent'], id=None), 'idx': Value(dtype='int32', id=None)}, post_processed=None, supervised_keys=None, task_templates=None, builder_name='glue', dataset_name=None, config_name='mrpc', version=1.0.0, splits={'train': SplitInfo(name='train', num_bytes=943843, num_examples=3668, shard_lengths=None, dataset_name='glue'), 'validation': SplitInfo(name='validation', num_bytes=105879, num_examples=408, shard_lengths=None, dataset_name='glue'), 'test': SplitInfo(name='test', num_bytes=442410, num_examples=1725, shard_lengths=None, dataset_name='glue')}, download_checksums={'https://dl.fbaipublicfiles.com/glue/data/mrpc_dev_ids.tsv': {'num_bytes': 6222, 'checksum': None}, 'https://dl.fbaipublicfiles.com/senteval/senteval_data/msr_paraphrase_train.txt': {'num_bytes': 1047044, 'checksum': None}, 'https://dl.fbaipublicfiles.com/senteval/senteval_data/msr_paraphrase_test.txt': {'num_bytes': 441275, 'checksum': None}}, download_size=1494541, post_processing_size=None, dataset_size=1492132, size_in_bytes=2986673)
### Expected behavior
```
from datasets import load_dataset
dataset = load_dataset('glue', 'mrpc', split='train')
dataset.info
```
DatasetInfo(
description='GLUE, the General Language Understanding Evaluation benchmark\n(https://gluebenchmark.com/) is a collection of resources for training,\nevaluating, and analyzing natural language understanding systems.\n\n',
citation='@inproceedings{dolan2005automatically,\n title={Automatically constructing a corpus of sentential paraphrases},\n author={Dolan, William B and Brockett, Chris},\n booktitle={Proceedings of the Third International Workshop on Paraphrasing (IWP2005)},\n year={2005}\n}\n@inproceedings{wang2019glue,\n title={{GLUE}: A Multi-Task Benchmark and Analysis Platform for Natural Language Understanding},\n author={Wang, Alex and Singh, Amanpreet and Michael, Julian and Hill, Felix and Levy, Omer and Bowman, Samuel R.},\n note={In the Proceedings of ICLR.},\n year={2019}\n}\n', homepage='https://www.microsoft.com/en-us/download/details.aspx?id=52398',
license='',
features={'sentence1': Value(dtype='string', id=None), 'sentence2': Value(dtype='string', id=None), 'label': ClassLabel(num_classes=2, names=['not_equivalent', 'equivalent'], names_file=None, id=None), 'idx': Value(dtype='int32', id=None)}, post_processed=None, supervised_keys=None, builder_name='glue', config_name='mrpc', version=1.0.0, splits={'train': SplitInfo(name='train', num_bytes=943851, num_examples=3668, dataset_name='glue'), 'validation': SplitInfo(name='validation', num_bytes=105887, num_examples=408, dataset_name='glue'), 'test': SplitInfo(name='test', num_bytes=442418, num_examples=1725, dataset_name='glue')},
download_checksums={'https://dl.fbaipublicfiles.com/glue/data/mrpc_dev_ids.tsv': {'num_bytes': 6222, 'checksum': '971d7767d81b997fd9060ade0ec23c4fc31cbb226a55d1bd4a1bac474eb81dc7'}, 'https://dl.fbaipublicfiles.com/senteval/senteval_data/msr_paraphrase_train.txt': {'num_bytes': 1047044, 'checksum': '60a9b09084528f0673eedee2b69cb941920f0b8cd0eeccefc464a98768457f89'}, 'https://dl.fbaipublicfiles.com/senteval/senteval_data/msr_paraphrase_test.txt': {'num_bytes': 441275, 'checksum': 'a04e271090879aaba6423d65b94950c089298587d9c084bf9cd7439bd785f784'}},
download_size=1494541,
post_processing_size=None,
dataset_size=1492156,
size_in_bytes=2986697
)
|
open
|
https://github.com/huggingface/datasets/issues/6495
| 2023-12-12T23:07:51
| 2023-12-13T13:24:22
| null |
{
"login": "gerald-wrona",
"id": 32300890,
"type": "User"
}
|
[] | false
|
[] |
2,038,221,490
| 6,493
|
Lazy data files resolution and offline cache reload
|
Includes both https://github.com/huggingface/datasets/pull/6458 and https://github.com/huggingface/datasets/pull/6459
This PR should be merged instead of the two individually, since they are conflicting
## Offline cache reload
it can reload datasets that were pushed to hub if they exist in the cache.
example:
```python
>>> Dataset.from_dict({"a": [1, 2]}).push_to_hub("lhoestq/tmp")
>>> load_dataset("lhoestq/tmp")
DatasetDict({
train: Dataset({
features: ['a'],
num_rows: 2
})
})
```
and later, without connection:
```python
>>> load_dataset("lhoestq/tmp")
Using the latest cached version of the dataset since lhoestq/tmp couldn't be found on the Hugging Face Hub
Found the latest cached dataset configuration 'default' at /Users/quentinlhoest/.cache/huggingface/datasets/lhoestq___tmp/default/0.0.0/da0e902a945afeb9 (last modified on Wed Dec 13 14:55:52 2023).
DatasetDict({
train: Dataset({
features: ['a'],
num_rows: 2
})
})
```
- Updated `CachedDatasetModuleFactory` to look for datasets in the cache at `<namespace>___<dataset_name>/<config_id>`
- Since the metadata configs parameters are not available in offline mode, we don't know which folder to load (config_id and hash change), so I simply load the latest one
- I instantiate a BuilderConfig even if there is no metadata config with the right config_name
- Its config_id is equal to the config_name to be able to retrieve it in the cache (no more suffix for configs from metadata configs)
- We can reload this config if offline mode by specifying the right config_name (same as online !)
- Consequences of this change:
- Only when there are user's parameters it creates a custom builder config with config_id = config_name + user parameters hash
- the hash used to name the cache folder takes into account the metadata config and the dataset info, so that the right cache can be reloaded when there is internet connection without redownloading the data or resolving the data files. For local directories I hash the builder configs and dataset info, and for datasets on the hub I use the commit sha as hash.
- cache directories now look like `config/version/commit_sha` for hub datasets which is clean :)
Fix https://github.com/huggingface/datasets/issues/3547
## Lazy data files resolution
this makes this code run in 2sec instead of >10sec
```python
from datasets import load_dataset
ds = load_dataset("glue", "sst2", streaming=True, trust_remote_code=False)
```
For some datasets with many configs and files it can be up to 100x faster.
This is particularly important now that some datasets will be loaded from the Parquet export instead of the scripts.
The data files are only resolved in the builder `__init__`. To do so I added DataFilesPatternsList and DataFilesPatternsDict that have `.resolve()` to return resolved DataFilesList and DataFilesDict
|
closed
|
https://github.com/huggingface/datasets/pull/6493
| 2023-12-12T17:15:17
| 2023-12-21T15:19:20
| 2023-12-21T15:13:11
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,037,987,267
| 6,492
|
Make push_to_hub return CommitInfo
|
Make `push_to_hub` return `CommitInfo`.
This is useful, for example, if we pass `create_pr=True` and we want to know the created PR ID.
CC: @severo for the use case in https://huggingface.co/datasets/jmhessel/newyorker_caption_contest/discussions/4
|
closed
|
https://github.com/huggingface/datasets/pull/6492
| 2023-12-12T15:18:16
| 2023-12-13T14:29:01
| 2023-12-13T14:22:41
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,037,690,643
| 6,491
|
Fix metrics dead link
| null |
closed
|
https://github.com/huggingface/datasets/pull/6491
| 2023-12-12T12:51:49
| 2023-12-21T15:15:08
| 2023-12-21T15:08:53
|
{
"login": "qgallouedec",
"id": 45557362,
"type": "User"
}
|
[] | true
|
[] |
2,037,204,892
| 6,490
|
`load_dataset(...,save_infos=True)` not working without loading script
|
### Describe the bug
It seems that saving a dataset infos back into the card file is not working for datasets without a loading script.
After tracking the problem a bit it looks like saving the infos uses `Builder.get_imported_module_dir()` as its destination directory.
Internally this is a call to `inspect.getfile()` but since the actual builder class used is dynamically created (cf. `datasets.load.configure_builder_class`) this method actually return te path to the parent builder class (e.g. `datasets.packaged_modules.json.JSON`).
### Steps to reproduce the bug
1. Have a local dataset without any loading script
2. Make sure there are no dataset infos in the README.md
3. Load with `save_infos=True`
4. No change in the dataset README.md
5. A new README.md file is created in the directory of the parent builder class (e.g. for json in `.../site-packages/datasets/packaged_modules/json/README.md`)
### Expected behavior
The dataset README.md should be updated and no file should be created in the python environment.
### Environment info
- `datasets` version: 2.15.0
- Platform: Linux-6.2.0-37-generic-x86_64-with-glibc2.35
- Python version: 3.10.12
- `huggingface_hub` version: 0.19.4
- PyArrow version: 14.0.1
- Pandas version: 2.1.3
- `fsspec` version: 2023.6.0
|
open
|
https://github.com/huggingface/datasets/issues/6490
| 2023-12-12T08:09:18
| 2023-12-12T08:36:22
| null |
{
"login": "morganveyret",
"id": 114978051,
"type": "User"
}
|
[] | false
|
[] |
2,036,743,777
| 6,489
|
load_dataset imageflder for aws s3 path
|
### Feature request
I would like to load a dataset from S3 using the imagefolder option
something like
`dataset = datasets.load_dataset('imagefolder', data_dir='s3://.../lsun/train/bedroom', fs=S3FileSystem(), streaming=True) `
### Motivation
no need of data_files
### Your contribution
no experience with this
|
open
|
https://github.com/huggingface/datasets/issues/6489
| 2023-12-12T00:08:43
| 2023-12-12T00:09:27
| null |
{
"login": "segalinc",
"id": 9353106,
"type": "User"
}
|
[
{
"name": "enhancement",
"color": "a2eeef"
}
] | false
|
[] |
2,035,899,898
| 6,488
|
429 Client Error
|
Hello, I was downloading the following dataset and after 20% of data was downloaded, I started getting error 429. It is not resolved since a few days. How should I resolve it?
Thanks
Dataset:
https://huggingface.co/datasets/cerebras/SlimPajama-627B
Error:
`requests.exceptions.HTTPError: 429 Client Error: Too Many Requests for url: https://huggingface.co/datasets/cerebras/SlimPajama-627B/resolve/2d0accdd58c5d5511943ca1f5ff0e3eb5e293543/train/chunk1/example_train_3300.jsonl.zst`
|
open
|
https://github.com/huggingface/datasets/issues/6488
| 2023-12-11T15:06:01
| 2024-06-20T05:55:45
| null |
{
"login": "sasaadi",
"id": 7882383,
"type": "User"
}
|
[] | false
|
[] |
2,035,424,254
| 6,487
|
Update builder hash with info
|
Currently if you change the `dataset_info` of a dataset (e.g. in the YAML part of the README.md), the cache ignores this change.
This is problematic because you want to regenerate a dataset if you change the features or the split sizes for example (e.g. after push_to_hub)
Ideally we should take the resolved files into account as well but this will be for another PR
|
closed
|
https://github.com/huggingface/datasets/pull/6487
| 2023-12-11T11:09:16
| 2024-01-11T06:35:07
| 2023-12-11T11:41:34
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,035,206,206
| 6,486
|
Fix docs phrasing about supported formats when sharing a dataset
|
Fix docs phrasing.
|
closed
|
https://github.com/huggingface/datasets/pull/6486
| 2023-12-11T09:21:22
| 2023-12-13T14:21:29
| 2023-12-13T14:15:21
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,035,141,884
| 6,485
|
FileNotFoundError: [Errno 2] No such file or directory: 'nul'
|
### Describe the bug
it seems that sth wrong with my terrible "bug body" life, When i run this code, "import datasets"
i meet this error FileNotFoundError: [Errno 2] No such file or directory: 'nul'
### Steps to reproduce the bug
1.import datasets
### Expected behavior
i just run a single line code and stuct in this bug
### Environment info
OS: Windows10
Datasets==2.15.0
python=3.10
|
closed
|
https://github.com/huggingface/datasets/issues/6485
| 2023-12-11T08:52:13
| 2023-12-14T08:09:08
| 2023-12-14T08:09:08
|
{
"login": "amanyara",
"id": 73683903,
"type": "User"
}
|
[] | false
|
[] |
2,032,946,981
| 6,483
|
Iterable Dataset: rename column clashes with remove column
|
### Describe the bug
Suppose I have a two iterable datasets, one with the features:
* `{"audio", "text", "column_a"}`
And the other with the features:
* `{"audio", "sentence", "column_b"}`
I want to combine both datasets using `interleave_datasets`, which requires me to unify the column names. I would typically do this by:
1. Renaming the common columns to the same name (e.g. `"text"` -> `"sentence"`)
2. Removing the unwanted columns (e.g. `"column_a"`, `"column_b"`)
However, the process of renaming and removing columns in an iterable dataset doesn't work, since we need to preserve the original text column, meaning we can't combine the datasets.
### Steps to reproduce the bug
```python
from datasets import load_dataset
# load LS in streaming mode
dataset = load_dataset("librispeech_asr", "clean", split="validation", streaming=True)
# check original features
dataset_features = dataset.features.keys()
print("Original features: ", dataset_features)
# rename "text" -> "sentence"
dataset = dataset.rename_column("text", "sentence")
# remove unwanted columns
COLUMNS_TO_KEEP = {"audio", "sentence"}
dataset = dataset.remove_columns(set(dataset_features - COLUMNS_TO_KEEP))
# stream first sample, should return "audio" and "sentence" columns
print(next(iter(dataset)))
```
Traceback:
```python
---------------------------------------------------------------------------
KeyError Traceback (most recent call last)
Cell In[5], line 17
14 COLUMNS_TO_KEEP = {"audio", "sentence"}
15 dataset = dataset.remove_columns(set(dataset_features - COLUMNS_TO_KEEP))
---> 17 print(next(iter(dataset)))
File ~/datasets/src/datasets/iterable_dataset.py:1353, in IterableDataset.__iter__(self)
1350 yield formatter.format_row(pa_table)
1351 return
-> 1353 for key, example in ex_iterable:
1354 if self.features:
1355 # `IterableDataset` automatically fills missing columns with None.
1356 # This is done with `_apply_feature_types_on_example`.
1357 example = _apply_feature_types_on_example(
1358 example, self.features, token_per_repo_id=self._token_per_repo_id
1359 )
File ~/datasets/src/datasets/iterable_dataset.py:652, in MappedExamplesIterable.__iter__(self)
650 yield from ArrowExamplesIterable(self._iter_arrow, {})
651 else:
--> 652 yield from self._iter()
File ~/datasets/src/datasets/iterable_dataset.py:729, in MappedExamplesIterable._iter(self)
727 if self.remove_columns:
728 for c in self.remove_columns:
--> 729 del transformed_example[c]
730 yield key, transformed_example
731 current_idx += 1
KeyError: 'text'
```
=> we see that `datasets` is looking for the column "text", even though we've renamed this to "sentence" and then removed the un-wanted "text" column from our dataset.
### Expected behavior
Should be able to rename and remove columns from iterable dataset.
### Environment info
- `datasets` version: 2.15.1.dev0
- Platform: macOS-13.5.1-arm64-arm-64bit
- Python version: 3.11.6
- `huggingface_hub` version: 0.19.4
- PyArrow version: 14.0.1
- Pandas version: 2.1.2
- `fsspec` version: 2023.9.2
|
closed
|
https://github.com/huggingface/datasets/issues/6483
| 2023-12-08T16:11:30
| 2023-12-08T16:27:16
| 2023-12-08T16:27:04
|
{
"login": "sanchit-gandhi",
"id": 93869735,
"type": "User"
}
|
[
{
"name": "streaming",
"color": "fef2c0"
}
] | false
|
[] |
2,033,333,294
| 6,484
|
[Feature Request] Dataset versioning
|
**Is your feature request related to a problem? Please describe.**
I am working on a project, where I would like to test different preprocessing methods for my ML-data. Thus, I would like to work a lot with revisions and compare them. Currently, I was not able to make it work with the revision keyword because it was not redownloading the data, it was reading in some cached data, until I put `download_mode="force_redownload"`, even though the reversion was different.
Of course, I may have done something wrong or missed a setting somewhere!
**Describe the solution you'd like**
The solution would allow me to easily work with revisions:
- create a new dataset (by combining things, different preprocessing, ..) and give it a new revision (v.1.2.3), maybe like this:
`dataset_audio.push_to_hub('kenfus/xy', revision='v1.0.2')`
- then, get the current revision as follows:
```
dataset = load_dataset(
'kenfus/xy', revision='v1.0.2',
)
```
this downloads the new version and does not load in a different revision and all future map, filter, .. operations are done on this dataset and not loaded from cache produced from a different revision.
- if I rerun the run, the caching should be smart enough in every step to not reuse a mapping operation on a different revision.
**Describe alternatives you've considered**
I created my own caching, putting `download_mode="force_redownload"` and `load_from_cache_file=False,` everywhere.
**Additional context**
Thanks a lot for your great work! Creating NLP datasets and training a model with them is really easy and straightforward with huggingface.
This is the data loading in my script:
```
## CREATE PATHS
prepared_dataset_path = os.path.join(
DATA_FOLDER, str(DATA_VERSION), "prepared_dataset"
)
os.makedirs(os.path.join(DATA_FOLDER, str(DATA_VERSION)), exist_ok=True)
## LOAD DATASET
if os.path.exists(prepared_dataset_path):
print("Loading prepared dataset from disk...")
dataset_prepared = load_from_disk(prepared_dataset_path)
else:
print("Loading dataset from HuggingFace Datasets...")
dataset = load_dataset(
PATH_TO_DATASET, revision=DATA_VERSION, download_mode="force_redownload"
)
print("Preparing dataset...")
dataset_prepared = dataset.map(
prepare_dataset,
remove_columns=["audio", "transcription"],
num_proc=os.cpu_count(),
load_from_cache_file=False,
)
dataset_prepared.save_to_disk(prepared_dataset_path)
del dataset
if CHECK_DATASET:
## CHECK DATASET
dataset_prepared = dataset_prepared.map(
check_dimensions, num_proc=os.cpu_count(), load_from_cache_file=False
)
dataset_filtered = dataset_prepared.filter(
lambda example: not example["incorrect_dimension"],
load_from_cache_file=False,
)
for example in dataset_prepared.filter(
lambda example: example["incorrect_dimension"], load_from_cache_file=False
):
print(example["path"])
print(
f"Number of examples with incorrect dimension: {len(dataset_prepared) - len(dataset_filtered)}"
)
print("Number of examples train: ", len(dataset_filtered["train"]))
print("Number of examples test: ", len(dataset_filtered["test"]))
```
|
open
|
https://github.com/huggingface/datasets/issues/6484
| 2023-12-08T16:01:35
| 2023-12-11T19:13:46
| null |
{
"login": "kenfus",
"id": 47979198,
"type": "User"
}
|
[] | false
|
[] |
2,032,675,918
| 6,482
|
Fix max lock length on unix
|
reported in https://github.com/huggingface/datasets/pull/6482
|
closed
|
https://github.com/huggingface/datasets/pull/6482
| 2023-12-08T13:39:30
| 2023-12-12T11:53:32
| 2023-12-12T11:47:27
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,032,650,003
| 6,481
|
using torchrun, save_to_disk suddenly shows SIGTERM
|
### Describe the bug
When I run my code using the "torchrun" command, when the code reaches the "save_to_disk" part, suddenly I get the following warning and error messages:
Because the dataset is too large, the "save_to_disk" function splits it into 70 parts for saving. However, an error occurs suddenly when it reaches the 14th shard.
WARNING: torch.distributed.elastic.multiprocessing.api: Sending process 2224968 closing signal SIGTERM
ERROR: torch.distributed.elastic.multiprocessing.api: failed (exitcode: -7). traceback: Signal 7 (SIGBUS) received by PID 2224967.
### Steps to reproduce the bug
ds_shard = ds_shard.map(map_fn, *args, **kwargs)
ds_shard.save_to_disk(ds_shard_filepaths[rank])
Saving the dataset (14/70 shards): 20%|██ | 875350/4376702 [00:19<01:53, 30863.15 examples/s]
WARNING:torch.distributed.elastic.multiprocessing.api:Sending process 2224968 closing signal SIGTERM
ERROR:torch.distributed.elastic.multiprocessing.api:failed (exitcode: -7) local_rank: 0 (pid: 2224967) of binary: /home/bingxing2/home/scx6964/.conda/envs/ariya235/bin/python
Traceback (most recent call last):
File "/home/bingxing2/home/scx6964/.conda/envs/ariya235/bin/torchrun", line 8, in <module>
sys.exit(main())
File "/home/bingxing2/home/scx6964/.conda/envs/ariya235/lib/python3.10/site-packages/torch/distributed/elastic/multiprocessing/errors/__init__.py", line 346, in wrapper
return f(*args, **kwargs)
File "/home/bingxing2/home/scx6964/.conda/envs/ariya235/lib/python3.10/site-packages/torch/distributed/run.py", line 794, in main
run(args)
File "/home/bingxing2/home/scx6964/.conda/envs/ariya235/lib/python3.10/site-packages/torch/distributed/run.py", line 785, in run
elastic_launch(
File "/home/bingxing2/home/scx6964/.conda/envs/ariya235/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 134, in __call__
return launch_agent(self._config, self._entrypoint, list(args))
File "/home/bingxing2/home/scx6964/.conda/envs/ariya235/lib/python3.10/site-packages/torch/distributed/launcher/api.py", line 250, in launch_agent
raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:
==========================================================
run.py FAILED
----------------------------------------------------------
Failures:
<NO_OTHER_FAILURES>
----------------------------------------------------------
Root Cause (first observed failure):
[0]:
time : 2023-12-08_20:09:04
rank : 0 (local_rank: 0)
exitcode : -7 (pid: 2224967)
error_file: <N/A>
traceback : Signal 7 (SIGBUS) received by PID 2224967
### Expected behavior
I hope it can save successfully without any issues, but it seems there is a problem.
### Environment info
`datasets` version: 2.14.6
- Platform: Linux-4.19.90-24.4.v2101.ky10.aarch64-aarch64-with-glibc2.28
- Python version: 3.10.11
- Huggingface_hub version: 0.17.3
- PyArrow version: 14.0.0
- Pandas version: 2.1.2
|
open
|
https://github.com/huggingface/datasets/issues/6481
| 2023-12-08T13:22:03
| 2023-12-08T13:22:03
| null |
{
"login": "Ariya12138",
"id": 85916625,
"type": "User"
}
|
[] | false
|
[] |
2,031,116,653
| 6,480
|
Add IterableDataset `__repr__`
|
Example for glue sst2:
Dataset
```
DatasetDict({
test: Dataset({
features: ['sentence', 'label', 'idx'],
num_rows: 1821
})
train: Dataset({
features: ['sentence', 'label', 'idx'],
num_rows: 67349
})
validation: Dataset({
features: ['sentence', 'label', 'idx'],
num_rows: 872
})
})
```
IterableDataset (new)
```
IterableDatasetDict({
test: IterableDataset({
features: ['sentence', 'label', 'idx'],
n_shards: 1
})
train: IterableDataset({
features: ['sentence', 'label', 'idx'],
n_shards: 1
})
validation: IterableDataset({
features: ['sentence', 'label', 'idx'],
n_shards: 1
})
})
```
IterableDataset (before)
```
{'test': <datasets.iterable_dataset.IterableDataset object at 0x130d421f0>, 'train': <datasets.iterable_dataset.IterableDataset object at 0x136f3aaf0>, 'validation': <datasets.iterable_dataset.IterableDataset object at 0x136f4b100>}
{'sentence': 'hide new secretions from the parental units ', 'label': 0, 'idx': 0}
```
|
closed
|
https://github.com/huggingface/datasets/pull/6480
| 2023-12-07T16:31:50
| 2023-12-08T13:33:06
| 2023-12-08T13:26:54
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,029,040,121
| 6,479
|
More robust preupload retry mechanism
| null |
closed
|
https://github.com/huggingface/datasets/pull/6479
| 2023-12-06T17:19:38
| 2023-12-06T19:47:29
| 2023-12-06T19:41:06
|
{
"login": "mariosasko",
"id": 47462742,
"type": "User"
}
|
[] | true
|
[] |
2,028,071,596
| 6,478
|
How to load data from lakefs
|
My dataset is stored on the company's lakefs server. How can I write code to load the dataset? It would be great if I could provide code examples or provide some references
|
closed
|
https://github.com/huggingface/datasets/issues/6478
| 2023-12-06T09:04:11
| 2024-07-03T19:13:57
| 2024-07-03T19:13:56
|
{
"login": "d710055071",
"id": 12895488,
"type": "User"
}
|
[] | false
|
[] |
2,028,022,374
| 6,477
|
Fix PermissionError on Windows CI
|
Fix #6476.
|
closed
|
https://github.com/huggingface/datasets/pull/6477
| 2023-12-06T08:34:53
| 2023-12-06T09:24:11
| 2023-12-06T09:17:52
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,028,018,596
| 6,476
|
CI on windows is broken: PermissionError
|
See: https://github.com/huggingface/datasets/actions/runs/7104781624/job/19340572394
```
FAILED tests/test_load.py::test_loading_from_the_datasets_hub - NotADirectoryError: [WinError 267] The directory name is invalid: 'C:\\Users\\RUNNER~1\\AppData\\Local\\Temp\\tmpfcnps56i\\hf-internal-testing___dataset_with_script\\default\\0.0.0\\c240e2be3370bdbd\\dataset_with_script-train.arrow'
```
|
closed
|
https://github.com/huggingface/datasets/issues/6476
| 2023-12-06T08:32:53
| 2023-12-06T09:17:53
| 2023-12-06T09:17:53
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[
{
"name": "bug",
"color": "d73a4a"
}
] | false
|
[] |
2,027,373,734
| 6,475
|
laion2B-en failed to load on Windows with PrefetchVirtualMemory failed
|
### Describe the bug
I have downloaded laion2B-en, and I'm receiving the following error trying to load it:
```
Resolving data files: 100%|██████████| 128/128 [00:00<00:00, 1173.79it/s]
Traceback (most recent call last):
File "D:\Art-Workspace\src\artworkspace\tokeneval\compute_frequencies.py", line 31, in <module>
count = compute_frequencies()
^^^^^^^^^^^^^^^^^^^^^
File "D:\Art-Workspace\src\artworkspace\tokeneval\compute_frequencies.py", line 17, in compute_frequencies
laion2b_dataset = load_dataset("laion/laion2B-en", split="train", cache_dir=_CACHE_DIR, keep_in_memory=False)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\load.py", line 2165, in load_dataset
ds = builder_instance.as_dataset(split=split, verification_mode=verification_mode, in_memory=keep_in_memory)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\builder.py", line 1187, in as_dataset
datasets = map_nested(
^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\utils\py_utils.py", line 456, in map_nested
return function(data_struct)
^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\builder.py", line 1217, in _build_single_dataset
ds = self._as_dataset(
^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\builder.py", line 1291, in _as_dataset
dataset_kwargs = ArrowReader(cache_dir, self.info).read(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\arrow_reader.py", line 244, in read
return self.read_files(files=files, original_instructions=instructions, in_memory=in_memory)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\arrow_reader.py", line 265, in read_files
pa_table = self._read_files(files, in_memory=in_memory)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\arrow_reader.py", line 200, in _read_files
pa_table: Table = self._get_table_from_filename(f_dict, in_memory=in_memory)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\arrow_reader.py", line 336, in _get_table_from_filename
table = ArrowReader.read_table(filename, in_memory=in_memory)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\arrow_reader.py", line 357, in read_table
return table_cls.from_file(filename)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\table.py", line 1059, in from_file
table = _memory_mapped_arrow_table_from_file(filename)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "C:\Users\bberman\Documents\Art-Workspace\venv\Lib\site-packages\datasets\table.py", line 66, in _memory_mapped_arrow_table_from_file
pa_table = opened_stream.read_all()
^^^^^^^^^^^^^^^^^^^^^^^^
File "pyarrow\ipc.pxi", line 757, in pyarrow.lib.RecordBatchReader.read_all
File "pyarrow\error.pxi", line 91, in pyarrow.lib.check_status
OSError: [WinError 8] PrefetchVirtualMemory failed. Detail: [Windows error 8] Not enough memory resources are available to process this command.
```
This error is probably a red herring: https://stackoverflow.com/questions/50263929/numpy-memmap-returns-not-enough-memory-while-there-are-plenty-available In other words, the issue is related to asking for a memory mapping of length N > M the length of the file on Windows. This gracefully succeeds on Linux.
I have 1024 arrow files in my cache instead of 128 like in the repository for it. Probably related. I don't know why `datasets` reorganized/rewrote the dataset in my cache to be 1024 slices instead of the original 128.
### Steps to reproduce the bug
```
# as a huggingface developer, you may already have laion2B-en somewhere
_CACHE_DIR = "."
from datasets import load_dataset
load_dataset("laion/laion2B-en", split="train", cache_dir=_CACHE_DIR, keep_in_memory=False)
```
### Expected behavior
This should correctly load as a memory mapped Arrow dataset.
### Environment info
- `datasets` version: 2.15.0
- Platform: Windows-10-10.0.20348-SP0 (this is windows 2022)
- Python version: 3.11.4
- `huggingface_hub` version: 0.19.4
- PyArrow version: 14.0.1
- Pandas version: 2.1.2
- `fsspec` version: 2023.10.0
|
open
|
https://github.com/huggingface/datasets/issues/6475
| 2023-12-06T00:07:34
| 2023-12-06T23:26:23
| null |
{
"login": "doctorpangloss",
"id": 2229300,
"type": "User"
}
|
[] | false
|
[] |
2,027,006,715
| 6,474
|
Deprecate Beam API and download from HF GCS bucket
|
Deprecate the Beam API and download from the HF GCS bucked.
TODO:
- [x] Convert the Beam-based [`wikipedia`](https://huggingface.co/datasets/wikipedia) to an Arrow-based dataset ([Hub PR](https://huggingface.co/datasets/wikipedia/discussions/19))
- [x] Make [`natural_questions`](https://huggingface.co/datasets/natural_questions) a no-code dataset ([Hub PR](https://huggingface.co/datasets/natural_questions/discussions/7))
- [x] Make [`wiki40b`](https://huggingface.co/datasets/wiki40b) a no-code dataset ([Hub PR](https://huggingface.co/datasets/wiki40b/discussions/5))
- [x] Make [`wiki_dpr`](https://huggingface.co/datasets/wiki_dpr) an Arrow-based dataset ([Hub PR](https://huggingface.co/datasets/wiki_dpr/discussions/14))
|
closed
|
https://github.com/huggingface/datasets/pull/6474
| 2023-12-05T19:51:33
| 2024-03-12T14:56:25
| 2024-03-12T14:50:12
|
{
"login": "mariosasko",
"id": 47462742,
"type": "User"
}
|
[] | true
|
[] |
2,026,495,084
| 6,473
|
Fix CI quality
|
Fix #6472.
|
closed
|
https://github.com/huggingface/datasets/pull/6473
| 2023-12-05T15:36:23
| 2023-12-05T18:14:50
| 2023-12-05T18:08:41
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[] | true
|
[] |
2,026,493,439
| 6,472
|
CI quality is broken
|
See: https://github.com/huggingface/datasets/actions/runs/7100835633/job/19327734359
```
Would reformat: src/datasets/features/image.py
1 file would be reformatted, 253 files left unchanged
```
|
closed
|
https://github.com/huggingface/datasets/issues/6472
| 2023-12-05T15:35:34
| 2023-12-06T08:17:34
| 2023-12-05T18:08:43
|
{
"login": "albertvillanova",
"id": 8515462,
"type": "User"
}
|
[
{
"name": "bug",
"color": "d73a4a"
},
{
"name": "maintenance",
"color": "d4c5f9"
}
] | false
|
[] |
2,026,100,761
| 6,471
|
Remove delete doc CI
| null |
closed
|
https://github.com/huggingface/datasets/pull/6471
| 2023-12-05T12:37:50
| 2023-12-05T12:44:59
| 2023-12-05T12:38:50
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,024,724,319
| 6,470
|
If an image in a dataset is corrupted, we get unescapable error
|
### Describe the bug
Example discussed in detail here: https://huggingface.co/datasets/sasha/birdsnap/discussions/1
### Steps to reproduce the bug
```
from datasets import load_dataset, VerificationMode
dataset = load_dataset(
'sasha/birdsnap',
split="train",
verification_mode=VerificationMode.ALL_CHECKS,
streaming=True # I recommend using streaming=True when reproducing, as this dataset is large
)
for idx, row in enumerate(dataset):
# Iterating to 9287 took 7 minutes for me
# If you already have the data locally cached and set streaming=False, you see the same error just by with dataset[9287]
pass
# error at 9287 OSError: image file is truncated (45 bytes not processed)
# note that we can't avoid the error using a try/except + continue inside the loop
```
### Expected behavior
Able to escape errors in casting to Image() without killing the whole loop
### Environment info
- `datasets` version: 2.15.0
- Platform: Linux-5.15.0-84-generic-x86_64-with-glibc2.31
- Python version: 3.11.5
- `huggingface_hub` version: 0.19.4
- PyArrow version: 14.0.1
- Pandas version: 2.1.3
- `fsspec` version: 2023.10.0
|
open
|
https://github.com/huggingface/datasets/issues/6470
| 2023-12-04T20:58:49
| 2023-12-04T20:58:49
| null |
{
"login": "chigozienri",
"id": 14337872,
"type": "User"
}
|
[] | false
|
[] |
2,023,695,839
| 6,469
|
Don't expand_info in HF glob
|
Finally fix https://github.com/huggingface/datasets/issues/5537
|
closed
|
https://github.com/huggingface/datasets/pull/6469
| 2023-12-04T12:00:37
| 2023-12-15T13:18:37
| 2023-12-15T13:12:30
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,023,617,877
| 6,468
|
Use auth to get parquet export
|
added `token` to the `_datasets_server` functions
|
closed
|
https://github.com/huggingface/datasets/pull/6468
| 2023-12-04T11:18:27
| 2023-12-04T17:21:22
| 2023-12-04T17:15:11
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,023,174,233
| 6,467
|
New version release request
|
### Feature request
Hi!
I am using `datasets` in library `xtuner` and am highly interested in the features introduced since v2.15.0.
To avoid installation from source in our pypi wheels, we are eagerly waiting for the new release. So, Does your team have a new release plan for v2.15.1 and could you please share it with us?
Thanks very much!
### Motivation
.
### Your contribution
.
|
closed
|
https://github.com/huggingface/datasets/issues/6467
| 2023-12-04T07:08:26
| 2023-12-04T15:42:22
| 2023-12-04T15:42:22
|
{
"login": "LZHgrla",
"id": 36994684,
"type": "User"
}
|
[
{
"name": "enhancement",
"color": "a2eeef"
}
] | false
|
[] |
2,022,601,176
| 6,466
|
Can't align optional features of struct
|
### Describe the bug
Hello!
I'm currently experiencing an issue where I can't concatenate datasets if an inner field of a Feature is Optional.
I have a column named `speaker`, and this holds some information about a speaker.
```python
@dataclass
class Speaker:
name: str
email: Optional[str]
```
If I have two datasets, one happens to have `email` always None, then I get `The features can't be aligned because the key email of features`
### Steps to reproduce the bug
You can run the following script:
```python
ds = Dataset.from_dict({'speaker': [{'name': 'Ben', 'email': None}]})
ds2 = Dataset.from_dict({'speaker': [{'name': 'Fred', 'email': 'abc@aol.com'}]})
concatenate_datasets([ds, ds2])
>>>The features can't be aligned because the key speaker of features {'speaker': {'email': Value(dtype='string', id=None), 'name': Value(dtype='string', id=None)}} has unexpected type - {'email': Value(dtype='string', id=None), 'name': Value(dtype='string', id=None)} (expected either {'email': Value(dtype='null', id=None), 'name': Value(dtype='string', id=None)} or Value("null").
```
### Expected behavior
I think this should work; if two top-level columns were in the same situation it would properly cast to `string`.
```python
ds = Dataset.from_dict({'email': [None, None]})
ds2 = Dataset.from_dict({'email': ['abc@aol.com', 'one@yahoo.com']})
concatenate_datasets([ds, ds2])
>>> # Works!
```
### Environment info
- `datasets` version: 2.15.1.dev0
- Platform: Linux-5.15.0-89-generic-x86_64-with-glibc2.35
- Python version: 3.9.13
- `huggingface_hub` version: 0.19.4
- PyArrow version: 9.0.0
- Pandas version: 1.4.4
- `fsspec` version: 2023.6.0
I would be happy to fix this issue.
|
closed
|
https://github.com/huggingface/datasets/issues/6466
| 2023-12-03T15:57:07
| 2024-02-15T15:19:33
| 2024-02-08T14:38:34
|
{
"login": "Dref360",
"id": 8976546,
"type": "User"
}
|
[] | false
|
[] |
2,022,212,468
| 6,465
|
`load_dataset` uses out-of-date cache instead of re-downloading a changed dataset
|
### Describe the bug
When a dataset is updated on the hub, using `load_dataset` will load the locally cached dataset instead of re-downloading the updated dataset
### Steps to reproduce the bug
Here is a minimal example script to
1. create an initial dataset and upload
2. download it so it is stored in cache
3. change the dataset and re-upload
4. redownload
```python
import time
from datasets import Dataset, DatasetDict, DownloadMode, load_dataset
username = "YOUR_USERNAME_HERE"
initial = Dataset.from_dict({"foo": [1, 2, 3]})
print(f"Intial {initial['foo']}")
initial_ds = DatasetDict({"train": initial})
initial_ds.push_to_hub("test")
time.sleep(1)
download = load_dataset(f"{username}/test", split="train")
changed = download.map(lambda x: {"foo": x["foo"] + 1})
print(f"Changed {changed['foo']}")
changed.push_to_hub("test")
time.sleep(1)
download_again = load_dataset(f"{username}/test", split="train")
print(f"Download Changed {download_again['foo']}")
# >>> gives the out-dated [1,2,3] when it should be changed [2,3,4]
```
The redownloaded dataset should be the changed dataset but it is actually the cached, initial dataset. Force-redownloading gives the correct dataset
```python
download_again_force = load_dataset(f"{username}/test", split="train", download_mode=DownloadMode.FORCE_REDOWNLOAD)
print(f"Force Download Changed {download_again_force['foo']}")
# >>> [2,3,4]
```
### Expected behavior
I assumed there should be some sort of hashing that should check for changes in the dataset and re-download if the hashes don't match
### Environment info
- `datasets` version: 2.15.0 │
- Platform: Linux-5.15.0-1028-nvidia-x86_64-with-glibc2.17 │
- Python version: 3.8.17 │
- `huggingface_hub` version: 0.19.4 │
- PyArrow version: 13.0.0 │
- Pandas version: 2.0.3 │
- `fsspec` version: 2023.6.0
|
open
|
https://github.com/huggingface/datasets/issues/6465
| 2023-12-02T21:35:17
| 2024-08-20T08:32:11
| null |
{
"login": "mnoukhov",
"id": 3391297,
"type": "User"
}
|
[] | false
|
[] |
2,020,860,462
| 6,464
|
Add concurrent loading of shards to datasets.load_from_disk
|
In some file systems (like luster), memory mapping arrow files takes time. This can be accelerated by performing the mmap in parallel on processes or threads.
- Threads seem to be faster than processes when gathering the list of tables from the workers (see https://github.com/huggingface/datasets/issues/2252).
- I'm not sure if using threads would respect the `IN_MEMORY_MAX_SIZE` config.
- I'm not sure if we need to expose num_proc from `BaseReader.read` to `DatasetBuilder.as_dataset`. Since ` DatasetBuilder.as_dataset` is used in many places beside `load_dataset`.
### Tests on luster file system (on a shared partial node):
Loading 1231 shards of ~2GBs.
The files were pre-loaded in another process before the script runs (couldn't get a fresh node).
```python
import logging
from time import perf_counter
import datasets
logger = datasets.logging.get_logger(__name__)
datasets.logging.set_verbosity_info()
logging.basicConfig(level=logging.DEBUG, format="%(message)s")
class catchtime:
# context to measure loading time: https://stackoverflow.com/questions/33987060/python-context-manager-that-measures-time
def __init__(self, debug_print="Time", logger=logger):
self.debug_print = debug_print
self.logger = logger
def __enter__(self):
self.start = perf_counter()
return self
def __exit__(self, type, value, traceback):
self.time = perf_counter() - self.start
readout = f"{self.debug_print}: {self.time:.3f} seconds"
self.logger.info(readout)
dataset_path=""
# warmup
with catchtime("Loading in parallel", logger=logger):
ds = datasets.load_from_disk(dataset_path,num_proc=16)
# num_proc=16
with catchtime("Loading in parallel", logger=logger):
ds = datasets.load_from_disk(dataset_path,num_proc=16)
# num_proc=32
with catchtime("Loading in parallel", logger=logger):
ds = datasets.load_from_disk(dataset_path,num_proc=32)
# num_proc=1
with catchtime("Loading in conseq", logger=logger):
ds = datasets.load_from_disk(dataset_path,num_proc=1)
```
#### Run 1
```
open file: .../dataset_dict.json
Loading the dataset from disk using 16 threads: 100%|██████████| 1231/1231 [01:28<00:00, 13.96shards/s]
Loading in parallel: 88.690 seconds
open file: .../dataset_dict.json
Loading the dataset from disk using 16 threads: 100%|██████████| 1231/1231 [01:48<00:00, 11.31shards/s]
Loading in parallel: 109.339 seconds
open file: .../dataset_dict.json
Loading the dataset from disk using 32 threads: 100%|██████████| 1231/1231 [01:06<00:00, 18.56shards/s]
Loading in parallel: 66.931 seconds
open file: .../dataset_dict.json
Loading the dataset from disk: 100%|██████████| 1231/1231 [05:09<00:00, 3.98shards/s]
Loading in conseq: 309.792 seconds
```
#### Run 2
```
open file: .../dataset_dict.json
Loading the dataset from disk using 16 threads: 100%|██████████| 1231/1231 [01:38<00:00, 12.53shards/s]
Loading in parallel: 98.831 seconds
open file: .../dataset_dict.json
Loading the dataset from disk using 16 threads: 100%|██████████| 1231/1231 [02:01<00:00, 10.16shards/s]
Loading in parallel: 121.669 seconds
open file: .../dataset_dict.json
Loading the dataset from disk using 32 threads: 100%|██████████| 1231/1231 [01:07<00:00, 18.18shards/s]
Loading in parallel: 68.192 seconds
open file: .../dataset_dict.json
Loading the dataset from disk: 100%|██████████| 1231/1231 [05:19<00:00, 3.86shards/s]
Loading in conseq: 319.759 seconds
```
#### Run 3
```
open file: .../dataset_dict.json
Loading the dataset from disk using 16 threads: 100%|██████████| 1231/1231 [01:36<00:00, 12.74shards/s]
Loading in parallel: 96.936 seconds
open file: .../dataset_dict.json
Loading the dataset from disk using 16 threads: 100%|██████████| 1231/1231 [02:00<00:00, 10.24shards/s]
Loading in parallel: 120.761 seconds
open file: .../dataset_dict.json
Loading the dataset from disk using 32 threads: 100%|██████████| 1231/1231 [01:08<00:00, 18.04shards/s]
Loading in parallel: 68.666 seconds
open file: .../dataset_dict.json
Loading the dataset from disk: 100%|██████████| 1231/1231 [05:35<00:00, 3.67shards/s]
Loading in conseq: 335.777 seconds
```
fix #2252
|
closed
|
https://github.com/huggingface/datasets/pull/6464
| 2023-12-01T13:13:53
| 2024-01-26T15:17:43
| 2024-01-26T15:10:26
|
{
"login": "kkoutini",
"id": 51880718,
"type": "User"
}
|
[] | true
|
[] |
2,020,702,967
| 6,463
|
Disable benchmarks in PRs
|
In order to keep PR pages less spammy / more readable.
Having the benchmarks on commits on `main` is enough imo
|
closed
|
https://github.com/huggingface/datasets/pull/6463
| 2023-12-01T11:35:30
| 2023-12-01T12:09:09
| 2023-12-01T12:03:04
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,019,238,388
| 6,462
|
Missing DatasetNotFoundError
|
continuation of https://github.com/huggingface/datasets/pull/6431
this should fix the CI in https://github.com/huggingface/datasets/pull/6458 too
|
closed
|
https://github.com/huggingface/datasets/pull/6462
| 2023-11-30T18:09:43
| 2023-11-30T18:36:40
| 2023-11-30T18:30:30
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,018,850,731
| 6,461
|
Fix shard retry mechanism in `push_to_hub`
|
When it fails, `preupload_lfs_files` throws a [`RuntimeError`](https://github.com/huggingface/huggingface_hub/blob/5eefebee2c150a2df950ab710db350e96c711433/src/huggingface_hub/_commit_api.py#L402) error and chains the original HTTP error. This PR modifies the retry mechanism's error handling to account for that.
Fix https://github.com/huggingface/datasets/issues/6392
|
closed
|
https://github.com/huggingface/datasets/pull/6461
| 2023-11-30T14:57:14
| 2023-12-01T17:57:39
| 2023-12-01T17:51:33
|
{
"login": "mariosasko",
"id": 47462742,
"type": "User"
}
|
[] | true
|
[] |
2,017,433,899
| 6,460
|
jsonlines files don't load with `load_dataset`
|
### Describe the bug
While [the docs](https://huggingface.co/docs/datasets/upload_dataset#upload-dataset) seem to state that `.jsonl` is a supported extension for `datasets`, loading the dataset results in a `JSONDecodeError`.
### Steps to reproduce the bug
Code:
```
from datasets import load_dataset
dset = load_dataset('slotreck/pickle')
```
Traceback:
```
Downloading readme: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 925/925 [00:00<00:00, 3.11MB/s]
Downloading and preparing dataset json/slotreck--pickle to /mnt/home/lotrecks/.cache/huggingface/datasets/slotreck___json/slotreck--pickle-0c311f36ed032b04/0.0.0/8bb11242116d547c741b2e8a1f18598ffdd40a1d4f2a2872c7a28b697434bc96...
Downloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 589k/589k [00:00<00:00, 18.9MB/s]
Downloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 104k/104k [00:00<00:00, 4.61MB/s]
Downloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 170k/170k [00:00<00:00, 7.71MB/s]
Downloading data files: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 3.77it/s]
Extracting data files: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 523.92it/s]
Generating train split: 0 examples [00:00, ? examples/s]Failed to read file '/mnt/home/lotrecks/.cache/huggingface/datasets/downloads/6ec07bb2f279c9377036af6948532513fa8f48244c672d2644a2d7018ee5c9cb' with error <class 'pyarrow.lib.ArrowInvalid'>: JSON parse error: Column(/ner/[]/[]/[]) changed from number to string in row 0
Traceback (most recent call last):
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/packaged_modules/json/json.py", line 144, in _generate_tables
dataset = json.load(f)
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/json/__init__.py", line 296, in load
parse_constant=parse_constant, object_pairs_hook=object_pairs_hook, **kw)
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/json/__init__.py", line 348, in loads
return _default_decoder.decode(s)
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/json/decoder.py", line 340, in decode
raise JSONDecodeError("Extra data", s, end)
json.decoder.JSONDecodeError: Extra data: line 2 column 1 (char 3086)
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/builder.py", line 1879, in _prepare_split_single
for _, table in generator:
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/packaged_modules/json/json.py", line 147, in _generate_tables
raise e
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/packaged_modules/json/json.py", line 122, in _generate_tables
io.BytesIO(batch), read_options=paj.ReadOptions(block_size=block_size)
File "pyarrow/_json.pyx", line 259, in pyarrow._json.read_json
File "pyarrow/error.pxi", line 144, in pyarrow.lib.pyarrow_internal_check_status
File "pyarrow/error.pxi", line 100, in pyarrow.lib.check_status
pyarrow.lib.ArrowInvalid: JSON parse error: Column(/ner/[]/[]/[]) changed from number to string in row 0
The above exception was the direct cause of the following exception:
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/load.py", line 1815, in load_dataset
storage_options=storage_options,
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/builder.py", line 913, in download_and_prepare
**download_and_prepare_kwargs,
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/builder.py", line 1004, in _download_and_prepare
self._prepare_split(split_generator, **prepare_split_kwargs)
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/builder.py", line 1768, in _prepare_split
gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args
File "/mnt/home/lotrecks/anaconda3/envs/pickle/lib/python3.7/site-packages/datasets/builder.py", line 1912, in _prepare_split_single
raise DatasetGenerationError("An error occurred while generating the dataset") from e
datasets.builder.DatasetGenerationError: An error occurred while generating the dataset
```
### Expected behavior
For the dataset to be loaded without error.
### Environment info
- `datasets` version: 2.13.1
- Platform: Linux-3.10.0-1160.80.1.el7.x86_64-x86_64-with-centos-7.9.2009-Core
- Python version: 3.7.12
- Huggingface_hub version: 0.15.1
- PyArrow version: 8.0.0
- Pandas version: 1.3.5
|
closed
|
https://github.com/huggingface/datasets/issues/6460
| 2023-11-29T21:20:11
| 2023-12-29T02:58:29
| 2023-12-05T13:30:53
|
{
"login": "serenalotreck",
"id": 41377532,
"type": "User"
}
|
[] | false
|
[] |
2,017,029,380
| 6,459
|
Retrieve cached datasets that were pushed to hub when offline
|
I drafted the logic to retrieve a no-script dataset in the cache.
For example it can reload datasets that were pushed to hub if they exist in the cache.
example:
```python
>>> Dataset.from_dict({"a": [1, 2]}).push_to_hub("lhoestq/tmp")
>>> load_dataset("lhoestq/tmp")
DatasetDict({
train: Dataset({
features: ['a'],
num_rows: 2
})
})
```
and later, without connection:
```python
>>> load_dataset("lhoestq/tmp")
Using the latest cached version of the dataset from /Users/quentinlhoest/.cache/huggingface/datasets/lhoestq___tmp/*/*/0b3caccda1725efb(last modified on Wed Nov 29 16:50:27 2023) since it couldn't be found locally at lhoestq/tmp.
DatasetDict({
train: Dataset({
features: ['a'],
num_rows: 2
})
})
```
fix https://github.com/huggingface/datasets/issues/3547
## Implementation details (EDITED)
I continued in https://github.com/huggingface/datasets/pull/6493, see the changes there
TODO:
- [x] tests
- [ ] compatible with https://github.com/huggingface/datasets/pull/6458
|
closed
|
https://github.com/huggingface/datasets/pull/6459
| 2023-11-29T16:56:15
| 2024-03-25T13:55:42
| 2024-03-25T13:55:42
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,016,577,761
| 6,458
|
Lazy data files resolution
|
Related to discussion at https://github.com/huggingface/datasets/pull/6255
this makes this code run in 2sec instead of >10sec
```python
from datasets import load_dataset
ds = load_dataset("glue", "sst2", streaming=True, trust_remote_code=False)
```
For some datasets with many configs and files it can be up to 100x faster.
This is particularly important now that some datasets will be loaded from the Parquet export instead of the scripts.
The data files are only resolved in the builder `__init__`. To do so I added DataFilesPatternsList and DataFilesPatternsDict that have `.resolve()` to return resolved DataFilesList and DataFilesDict
|
closed
|
https://github.com/huggingface/datasets/pull/6458
| 2023-11-29T13:18:44
| 2024-02-08T14:41:35
| 2024-02-08T14:41:35
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,015,650,563
| 6,457
|
`TypeError`: huggingface_hub.hf_file_system.HfFileSystem.find() got multiple values for keyword argument 'maxdepth'
|
### Describe the bug
Please see https://github.com/huggingface/huggingface_hub/issues/1872
### Steps to reproduce the bug
Please see https://github.com/huggingface/huggingface_hub/issues/1872
### Expected behavior
Please see https://github.com/huggingface/huggingface_hub/issues/1872
### Environment info
Please see https://github.com/huggingface/huggingface_hub/issues/1872
|
closed
|
https://github.com/huggingface/datasets/issues/6457
| 2023-11-29T01:57:36
| 2023-11-29T15:39:03
| 2023-11-29T02:02:38
|
{
"login": "wasertech",
"id": 79070834,
"type": "User"
}
|
[] | false
|
[] |
2,015,186,090
| 6,456
|
Don't require trust_remote_code in inspect_dataset
|
don't require `trust_remote_code` in (deprecated) `inspect_dataset` (it defeats its purpose)
(not super important but we might as well keep it until the next major release)
this is needed to fix the tests in https://github.com/huggingface/datasets/pull/6448
|
closed
|
https://github.com/huggingface/datasets/pull/6456
| 2023-11-28T19:47:07
| 2023-11-30T10:40:23
| 2023-11-30T10:34:12
|
{
"login": "lhoestq",
"id": 42851186,
"type": "User"
}
|
[] | true
|
[] |
2,013,001,584
| 6,454
|
Refactor `dill` logic
|
Refactor the `dill` logic to make it easier to maintain (and fix some issues along the way)
It makes the following improvements to the serialization API:
* consistent order of a `dict`'s keys
* support for hashing `torch.compile`-ed modules and functions
* deprecates `datasets.fingerprint.hashregister` as the `hashregister`-ed reducers are never invoked anyways (does not support nested data as `pickle`/`dill` do)
~~TODO: optimize hashing of `pa.Table` and `datasets.table.Table`~~ The `pa_array.to_string` approach is faster for large arrays because it outputs the first 10 and last 10 elements (by default). The problem is that this can produce identical hashes for non-identical arrays if their differing elements get ellipsed...
Fix https://github.com/huggingface/datasets/issues/6440, fix https://github.com/huggingface/datasets/issues/5839
|
closed
|
https://github.com/huggingface/datasets/pull/6454
| 2023-11-27T20:01:25
| 2023-11-28T16:29:58
| 2023-11-28T16:29:31
|
{
"login": "mariosasko",
"id": 47462742,
"type": "User"
}
|
[] | true
|
[] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.