Datasets:

dalssy

/

data_del

like 0

https://api.github.com/repos/huggingface/datasets/issues/7087

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

https://api.github.com/repos/huggingface/datasets/issues/7087/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7087/comments

https://api.github.com/repos/huggingface/datasets/issues/7087/events

https://github.com/huggingface/datasets/issues/7087

I_kwDODunzps6R3K1z

Unable to create dataset card for Lushootseed language

closed

NONE

### Feature request While I was creating the dataset which contained all documents from the Lushootseed Wikipedia, the dataset card asked me to enter which language the dataset was in. Since Lushootseed is a critically endangered language, it was not available as one of the options. Is it possible to allow entering languages that aren't available in the options? ### Motivation I'd like to add more information about my dataset in the dataset card, and the language is one of the most important pieces of information, since the entire dataset is primarily concerned collecting Lushootseed documents. ### Your contribution I can submit a pull request

https://api.github.com/repos/huggingface/datasets/issues/7087/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7086

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

https://api.github.com/repos/huggingface/datasets/issues/7086/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7086/comments

https://api.github.com/repos/huggingface/datasets/issues/7086/events

https://github.com/huggingface/datasets/issues/7086

I_kwDODunzps6Rw6Ad

load_dataset ignores cached datasets and tries to hit HF Hub, resulting in API rate limit errors

open

NONE

### Describe the bug I have been running lm-eval-harness a lot which has results in an API rate limit. This seems strange, since all of the data should be cached locally. I have in fact verified this. ### Steps to reproduce the bug 1. Be Me 2. Run `load_dataset("TAUR-Lab/MuSR")` 3. Hit rate limit error 4. Dataset is in .cache/huggingface/datasets 5. ??? ### Expected behavior We should not run into API rate limits if we have cached the dataset ### Environment info datasets 2.16.0 python 3.10.4

https://api.github.com/repos/huggingface/datasets/issues/7086/timeline

https://api.github.com/repos/huggingface/datasets/issues/7085

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

https://api.github.com/repos/huggingface/datasets/issues/7085/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7085/comments

https://api.github.com/repos/huggingface/datasets/issues/7085/events

https://github.com/huggingface/datasets/issues/7085

I_kwDODunzps6Rb5Oq

[Regression] IterableDataset is broken on 2.20.0

closed

NONE

### Describe the bug In the latest version of datasets there is a major regression, after creating an `IterableDataset` from a generator and applying a few operations (`map`, `select`), you can no longer iterate through the dataset multiple times. The issue seems to stem from the recent addition of "resumable IterableDatasets" (#6658) (@lhoestq). It seems like it's keeping state when it shouldn't. ### Steps to reproduce the bug Minimal Reproducible Example (comparing `datasets==2.17.0` and `datasets==2.20.0`) ``` #!/bin/bash # List of dataset versions to test versions=("2.17.0" "2.20.0") # Loop through each version for version in "${versions[@]}"; do # Install the specific version of the datasets library pip3 install -q datasets=="$version" 2>/dev/null # Run the Python script python3 - <<EOF from datasets import IterableDataset from datasets.features.features import Features, Value def test_gen(): yield from [{"foo": i} for i in range(10)] features = Features([("foo", Value("int64"))]) d = IterableDataset.from_generator(test_gen, features=features) mapped = d.map(lambda row: {"foo": row["foo"] * 2}) column = mapped.select_columns(["foo"]) print("Version $version - Iterate Once:", list(column)) print("Version $version - Iterate Twice:", list(column)) EOF done ``` The output looks like this: ``` Version 2.17.0 - Iterate Once: [{'foo': 0}, {'foo': 2}, {'foo': 4}, {'foo': 6}, {'foo': 8}, {'foo': 10}, {'foo': 12}, {'foo': 14}, {'foo': 16}, {'foo': 18}] Version 2.17.0 - Iterate Twice: [{'foo': 0}, {'foo': 2}, {'foo': 4}, {'foo': 6}, {'foo': 8}, {'foo': 10}, {'foo': 12}, {'foo': 14}, {'foo': 16}, {'foo': 18}] Version 2.20.0 - Iterate Once: [{'foo': 0}, {'foo': 2}, {'foo': 4}, {'foo': 6}, {'foo': 8}, {'foo': 10}, {'foo': 12}, {'foo': 14}, {'foo': 16}, {'foo': 18}] Version 2.20.0 - Iterate Twice: [] ``` ### Expected behavior The expected behavior is it version 2.20.0 should behave the same as 2.17.0. ### Environment info `datasets==2.20.0` on any platform.

https://api.github.com/repos/huggingface/datasets/issues/7085/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7084

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

https://api.github.com/repos/huggingface/datasets/issues/7084/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7084/comments

https://api.github.com/repos/huggingface/datasets/issues/7084/events

https://github.com/huggingface/datasets/issues/7084

I_kwDODunzps6RaB0u

More easily support streaming local files

open

CONTRIBUTOR

### Feature request Simplify downloading and streaming datasets locally. Specifically, perhaps add an option to `load_dataset(..., streaming="download_first")` or add better support for streaming symlinked or arrow files. ### Motivation I have downloaded FineWeb-edu locally and currently trying to stream the dataset from the local files. I have both the raw parquet files using `hugginface-cli download --repo-type dataset HuggingFaceFW/fineweb-edu` and the processed arrow files using `load_dataset("HuggingFaceFW/fineweb-edu")`. Streaming the files locally does not work well for both file types for two different reasons. **Arrow files** When running `load_dataset("arrow", data_files={"train": "~/.cache/huggingface/datasets/HuggingFaceFW___fineweb-edu/default/0.0.0/5b89d1ea9319fe101b3cbdacd89a903aca1d6052/fineweb-edu-train-*.arrow"})` resolving the data files is fast, but because `arrow` is not included in the known [extensions file list](https://github.com/huggingface/datasets/blob/ce4a0c573920607bc6c814605734091b06b860e7/src/datasets/utils/file_utils.py#L738) , all files are opened and scanned to determine the compression type. Adding `arrow` to the known extension types resolves this issue. **Parquet files** When running `load_dataset("arrow", data_files={"train": "~/.cache/huggingface/hub/dataset-HuggingFaceFW___fineweb-edu/snapshots/5b89d1ea9319fe101b3cbdacd89a903aca1d6052/data/CC-MAIN-*/train-*.parquet"})` the paths do not get resolved because the parquet files are symlinked from the blobs (which contain all files in case there are different versions). This occurs because the [pattern matching](https://github.com/huggingface/datasets/blob/ce4a0c573920607bc6c814605734091b06b860e7/src/datasets/data_files.py#L389) checks if the path is a file and does not check for symlinks. Symlinks (at least on my machine) are of type "other". ### Your contribution I have created a PR for fixing arrow file streaming and symlinks. However, I have not checked locally if the tests work or new tests need to be added. IMO, the easiest option would be to add a `streaming=download_first` option, but I'm afraid that exceeds my current knowledge of how the datasets library works. https://github.com/huggingface/datasets/pull/7083

https://api.github.com/repos/huggingface/datasets/issues/7084/timeline

https://api.github.com/repos/huggingface/datasets/issues/7083

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

https://api.github.com/repos/huggingface/datasets/issues/7083/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7083/comments

https://api.github.com/repos/huggingface/datasets/issues/7083/events

https://github.com/huggingface/datasets/pull/7083

PR_kwDODunzps5292hC

fix streaming from arrow files

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/7083/timeline

https://api.github.com/repos/huggingface/datasets/issues/7082

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

https://api.github.com/repos/huggingface/datasets/issues/7082/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7082/comments

https://api.github.com/repos/huggingface/datasets/issues/7082/events

https://github.com/huggingface/datasets/pull/7082

PR_kwDODunzps522dTJ

Support HTTP authentication in non-streaming mode

closed

MEMBER

Support HTTP authentication in non-streaming mode, by support passing HTTP storage_options in non-streaming mode. - Note that currently, HTTP authentication is supported only in streaming mode. For example, this is necessary if a remote HTTP host requires authentication to download the data.

https://api.github.com/repos/huggingface/datasets/issues/7082/timeline

https://api.github.com/repos/huggingface/datasets/issues/7081

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

https://api.github.com/repos/huggingface/datasets/issues/7081/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7081/comments

https://api.github.com/repos/huggingface/datasets/issues/7081/events

https://github.com/huggingface/datasets/pull/7081

PR_kwDODunzps521cGm

Set load_from_disk path type as PathLike

closed

MEMBER

Set `load_from_disk` path type as `PathLike`. This way it is aligned with `save_to_disk`.

https://api.github.com/repos/huggingface/datasets/issues/7081/timeline

https://api.github.com/repos/huggingface/datasets/issues/7080

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

https://api.github.com/repos/huggingface/datasets/issues/7080/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7080/comments

https://api.github.com/repos/huggingface/datasets/issues/7080/events

https://github.com/huggingface/datasets/issues/7080

I_kwDODunzps6RGBlQ

Generating train split takes a long time

open

NONE

### Describe the bug Loading a simple webdataset takes ~45 minutes. ### Steps to reproduce the bug ``` from datasets import load_dataset dataset = load_dataset("PixArt-alpha/SAM-LLaVA-Captions10M") ``` ### Expected behavior The dataset should load immediately as it does when loaded through a normal indexed WebDataset loader. Generating splits should be optional and there should be a message showing how to disable it. ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-4.18.0-372.32.1.el8_6.x86_64-x86_64-with-glibc2.28 - Python version: 3.10.14 - `huggingface_hub` version: 0.24.1 - PyArrow version: 16.1.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7080/timeline

https://api.github.com/repos/huggingface/datasets/issues/7079

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

https://api.github.com/repos/huggingface/datasets/issues/7079/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7079/comments

https://api.github.com/repos/huggingface/datasets/issues/7079/events

https://github.com/huggingface/datasets/issues/7079

I_kwDODunzps6RCi1i

HfHubHTTPError: 500 Server Error: Internal Server Error for url:

closed

NONE

### Describe the bug newly uploaded datasets, since yesterday, yields an error. old datasets, works fine. Seems like the datasets api server returns a 500 I'm getting the same error, when I invoke `load_dataset` with my dataset. Long discussion about it here, but I'm not sure anyone from huggingface have seen it. https://discuss.huggingface.co/t/hfhubhttperror-500-server-error-internal-server-error-for-url/99580/1 ### Steps to reproduce the bug this api url: https://huggingface.co/api/datasets/neoneye/simon-arc-shape-v4-rev3 respond with: ``` {"error":"Internal Error - We're working hard to fix this as soon as possible!"} ``` ### Expected behavior return no error with newer datasets. With older datasets I can load the datasets fine. ### Environment info # Browser When I access the api in the browser: https://huggingface.co/api/datasets/neoneye/simon-arc-shape-v4-rev3 ``` {"error":"Internal Error - We're working hard to fix this as soon as possible!"} ``` ### Request headers ``` Accept text/html,application/xhtml+xml,application/xml;q=0.9,image/avif,image/webp,*/*;q=0.8 Accept-Encoding gzip, deflate, br, zstd Accept-Language en-US,en;q=0.5 Connection keep-alive Host huggingface.co Priority u=1 Sec-Fetch-Dest document Sec-Fetch-Mode navigate Sec-Fetch-Site cross-site Upgrade-Insecure-Requests 1 User-Agent Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:127.0) Gecko/20100101 Firefox/127.0 ``` ### Response headers ``` X-Firefox-Spdy h2 access-control-allow-origin https://huggingface.co access-control-expose-headers X-Repo-Commit,X-Request-Id,X-Error-Code,X-Error-Message,X-Total-Count,ETag,Link,Accept-Ranges,Content-Range content-length 80 content-type application/json; charset=utf-8 cross-origin-opener-policy same-origin date Fri, 26 Jul 2024 19:09:45 GMT etag W/"50-9qrwU+BNI4SD0Fe32p/nofkmv0c" referrer-policy strict-origin-when-cross-origin vary Origin via 1.1 1624c79cd07e6098196697a6a7907e4a.cloudfront.net (CloudFront) x-amz-cf-id SP8E7n5qRaP6i9c9G83dNAiOzJBU4GXSrDRAcVNTomY895K35H0nJQ== x-amz-cf-pop CPH50-C1 x-cache Error from cloudfront x-error-message Internal Error - We're working hard to fix this as soon as possible! x-powered-by huggingface-moon x-request-id Root=1-66a3f479-026417465ef42f49349fdca1 ```

https://api.github.com/repos/huggingface/datasets/issues/7079/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7078

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

https://api.github.com/repos/huggingface/datasets/issues/7078/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7078/comments

https://api.github.com/repos/huggingface/datasets/issues/7078/events

https://github.com/huggingface/datasets/pull/7078

PR_kwDODunzps52oq4n

Fix CI test_convert_to_parquet

closed

MEMBER

Fix `test_convert_to_parquet` by patching `HfApi.preupload_lfs_files` and revert temporary fix: - #7074

https://api.github.com/repos/huggingface/datasets/issues/7078/timeline

https://api.github.com/repos/huggingface/datasets/issues/7077

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

https://api.github.com/repos/huggingface/datasets/issues/7077/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7077/comments

https://api.github.com/repos/huggingface/datasets/issues/7077/events

https://github.com/huggingface/datasets/issues/7077

I_kwDODunzps6Q-qWR

column_names ignored by load_dataset() when loading CSV file

open

NONE

### Describe the bug load_dataset() ignores the column_names kwarg when loading a CSV file. Instead, it uses whatever values are on the first line of the file. ### Steps to reproduce the bug Call `load_dataset` to load data from a CSV file and specify `column_names` kwarg. ### Expected behavior The resulting dataset should have the specified column names **and** the first line of the file should be considered as data values. ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-5.10.0-30-cloud-amd64-x86_64-with-glibc2.31 - Python version: 3.9.2 - `huggingface_hub` version: 0.24.2 - PyArrow version: 17.0.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7077/timeline

https://api.github.com/repos/huggingface/datasets/issues/7076

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

https://api.github.com/repos/huggingface/datasets/issues/7076/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7076/comments

https://api.github.com/repos/huggingface/datasets/issues/7076/events

https://github.com/huggingface/datasets/pull/7076

PR_kwDODunzps52lTDe

🧪 Do not mock create_commit

closed

MEMBER

https://api.github.com/repos/huggingface/datasets/issues/7076/timeline

https://api.github.com/repos/huggingface/datasets/issues/7075

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

https://api.github.com/repos/huggingface/datasets/issues/7075/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7075/comments

https://api.github.com/repos/huggingface/datasets/issues/7075/events

https://github.com/huggingface/datasets/pull/7075

PR_kwDODunzps52kciD

Update required soxr version from pre-release to release

closed

MEMBER

Update required `soxr` version from pre-release to release 0.4.0: https://github.com/dofuuz/python-soxr/releases/tag/v0.4.0

https://api.github.com/repos/huggingface/datasets/issues/7075/timeline

https://api.github.com/repos/huggingface/datasets/issues/7074

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

https://api.github.com/repos/huggingface/datasets/issues/7074/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7074/comments

https://api.github.com/repos/huggingface/datasets/issues/7074/events

https://github.com/huggingface/datasets/pull/7074

PR_kwDODunzps52jkw4

Fix CI by temporarily marking test_convert_to_parquet as expected to fail

closed

MEMBER

As a hotfix for CI, temporarily mark test_convert_to_parquet as expected to fail. Fix #7073. Revert once root cause is fixed.

https://api.github.com/repos/huggingface/datasets/issues/7074/timeline

https://api.github.com/repos/huggingface/datasets/issues/7073

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

https://api.github.com/repos/huggingface/datasets/issues/7073/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7073/comments

https://api.github.com/repos/huggingface/datasets/issues/7073/events

https://github.com/huggingface/datasets/issues/7073

I_kwDODunzps6Q8OXI

CI is broken for convert_to_parquet: Invalid rev id: refs/pr/1 404 error causes RevisionNotFoundError

closed

MEMBER

See: https://github.com/huggingface/datasets/actions/runs/10095313567/job/27915185756 ``` FAILED tests/test_hub.py::test_convert_to_parquet - huggingface_hub.utils._errors.RevisionNotFoundError: 404 Client Error. (Request ID: Root=1-66a25839-31ce7b475e70e7db1e4d44c2;b0c8870f-d5ef-4bf2-a6ff-0191f3df0f64) Revision Not Found for url: https://hub-ci.huggingface.co/api/datasets/__DUMMY_TRANSFORMERS_USER__/test-dataset-5188a8-17219154347516/preupload/refs%2Fpr%2F1. Invalid rev id: refs/pr/1 ``` ``` /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/hub.py:86: in convert_to_parquet dataset.push_to_hub( /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/dataset_dict.py:1722: in push_to_hub split_additions, uploaded_size, dataset_nbytes = self[split]._push_parquet_shards_to_hub( /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/datasets/arrow_dataset.py:5511: in _push_parquet_shards_to_hub api.preupload_lfs_files( /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/huggingface_hub/hf_api.py:4231: in preupload_lfs_files _fetch_upload_modes( /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/huggingface_hub/utils/_validators.py:118: in _inner_fn return fn(*args, **kwargs) /opt/hostedtoolcache/Python/3.8.18/x64/lib/python3.8/site-packages/huggingface_hub/_commit_api.py:507: in _fetch_upload_modes hf_raise_for_status(resp) ```

https://api.github.com/repos/huggingface/datasets/issues/7073/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7072

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

https://api.github.com/repos/huggingface/datasets/issues/7072/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7072/comments

https://api.github.com/repos/huggingface/datasets/issues/7072/events

https://github.com/huggingface/datasets/issues/7072

I_kwDODunzps6Q36z8

nm

closed

NONE

https://api.github.com/repos/huggingface/datasets/issues/7072/timeline

not_planned

https://api.github.com/repos/huggingface/datasets/issues/7071

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

https://api.github.com/repos/huggingface/datasets/issues/7071/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7071/comments

https://api.github.com/repos/huggingface/datasets/issues/7071/events

https://github.com/huggingface/datasets/issues/7071

I_kwDODunzps6Q26Iz

Filter hangs

open

NONE

### Describe the bug When trying to filter my custom dataset, the process hangs, regardless of the lambda function used. It appears to be an issue with the way the Images are being handled. The dataset in question is a preprocessed version of https://huggingface.co/datasets/danaaubakirova/patfig where notably, I have converted the data to the Parquet format. ### Steps to reproduce the bug ```python from datasets import load_dataset ds = load_dataset('lcolonn/patfig', split='test') ds_filtered = ds.filter(lambda row: row['cpc_class'] != 'Y') ``` Eventually I ctrl+C and I obtain this stack trace: ``` >>> ds_filtered = ds.filter(lambda row: row['cpc_class'] != 'Y') Filter: 0%| | 0/998 [00:00<?, ? examples/s]Filter: 0%| | 0/998 [00:35<?, ? examples/s] Traceback (most recent call last): File "<stdin>", line 1, in <module> File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 567, in wrapper out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/fingerprint.py", line 482, in wrapper out = func(dataset, *args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 3714, in filter indices = self.map( ^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 602, in wrapper out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 567, in wrapper out: Union["Dataset", "DatasetDict"] = func(self, *args, **kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 3161, in map for rank, done, content in Dataset._map_single(**dataset_kwargs): File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 3552, in _map_single batch = apply_function_on_filtered_inputs( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 3421, in apply_function_on_filtered_inputs processed_inputs = function(*fn_args, *additional_args, **fn_kwargs) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/arrow_dataset.py", line 6478, in get_indices_from_mask_function num_examples = len(batch[next(iter(batch.keys()))]) ~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/formatting/formatting.py", line 273, in __getitem__ value = self.format(key) ^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/formatting/formatting.py", line 376, in format return self.formatter.format_column(self.pa_table.select([key])) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/formatting/formatting.py", line 443, in format_column column = self.python_features_decoder.decode_column(column, pa_table.column_names[0]) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/formatting/formatting.py", line 219, in decode_column return self.features.decode_column(column, column_name) if self.features else column ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/features/features.py", line 2008, in decode_column [decode_nested_example(self[column_name], value) if value is not None else None for value in column] File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/features/features.py", line 2008, in <listcomp> [decode_nested_example(self[column_name], value) if value is not None else None for value in column] ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/features/features.py", line 1351, in decode_nested_example return schema.decode_example(obj, token_per_repo_id=token_per_repo_id) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/datasets/features/image.py", line 188, in decode_example image.load() # to avoid "Too many open files" errors ^^^^^^^^^^^^ File "/home/l-walewski/miniconda3/envs/patentqa/lib/python3.11/site-packages/PIL/ImageFile.py", line 293, in load n, err_code = decoder.decode(b) ^^^^^^^^^^^^^^^^^ KeyboardInterrupt ``` Warning! This can even seem to cause some computers to crash. ### Expected behavior Should return the filtered dataset ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-6.5.0-41-generic-x86_64-with-glibc2.35 - Python version: 3.11.9 - `huggingface_hub` version: 0.24.0 - PyArrow version: 17.0.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7071/timeline

https://api.github.com/repos/huggingface/datasets/issues/7070

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

https://api.github.com/repos/huggingface/datasets/issues/7070/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7070/comments

https://api.github.com/repos/huggingface/datasets/issues/7070/events

https://github.com/huggingface/datasets/issues/7070

I_kwDODunzps6Q2zWz

how set_transform affects batch size?

open

NONE

### Describe the bug I am trying to fine-tune w2v-bert for ASR task. Since my dataset is so big, I preferred to use the on-the-fly method with set_transform. So i change the preprocessing function to this: ``` def prepare_dataset(batch): input_features = processor(batch["audio"], sampling_rate=16000).input_features[0] input_length = len(input_features) labels = processor.tokenizer(batch["text"], padding=False).input_ids batch = { "input_features": [input_features], "input_length": [input_length], "labels": [labels] } return batch train_ds.set_transform(prepare_dataset) val_ds.set_transform(prepare_dataset) ``` After this, I also had to change the DataCollatorCTCWithPadding class like this: ``` @dataclass class DataCollatorCTCWithPadding: processor: Wav2Vec2BertProcessor padding: Union[bool, str] = True def __call__(self, features: List[Dict[str, Union[List[int], torch.Tensor]]]) -> Dict[str, torch.Tensor]: # Separate input_features and labels input_features = [{"input_features": feature["input_features"][0]} for feature in features] labels = [feature["labels"][0] for feature in features] # Pad input features batch = self.processor.pad( input_features, padding=self.padding, return_tensors="pt", ) # Pad and process labels label_features = self.processor.tokenizer.pad( {"input_ids": labels}, padding=self.padding, return_tensors="pt", ) labels = label_features["input_ids"] attention_mask = label_features["attention_mask"] # Replace padding with -100 to ignore these tokens during loss calculation labels = labels.masked_fill(attention_mask.ne(1), -100) batch["labels"] = labels return batch ``` But now a strange thing is happening, no matter how much I increase the batch size, the amount of V-RAM GPU usage does not change, while the number of total steps in the progress-bar (logging) changes. Is this normal or have I made a mistake? ### Steps to reproduce the bug i can share my code if needed ### Expected behavior Equal to the batch size value, the set_transform function is applied to the dataset and given to the model as a batch. ### Environment info all updated versions

https://api.github.com/repos/huggingface/datasets/issues/7070/timeline

https://api.github.com/repos/huggingface/datasets/issues/7069

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

https://api.github.com/repos/huggingface/datasets/issues/7069/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7069/comments

https://api.github.com/repos/huggingface/datasets/issues/7069/events

https://github.com/huggingface/datasets/pull/7069

PR_kwDODunzps52betB

Fix push_to_hub by not calling create_branch if PR branch

closed

MEMBER

Fix push_to_hub by not calling create_branch if PR branch (e.g. `refs/pr/1`). Note that currently create_branch raises a 400 Bad Request error if the user passes a PR branch (e.g. `refs/pr/1`). EDIT: ~~Fix push_to_hub by not calling create_branch if branch exists.~~ Note that currently create_branch raises a 403 Forbidden error even if all these conditions are met: - exist_ok is passed - the branch already exists - the user does not have WRITE permission Fix #7067. Related issue: - https://github.com/huggingface/huggingface_hub/issues/2419

https://api.github.com/repos/huggingface/datasets/issues/7069/timeline

https://api.github.com/repos/huggingface/datasets/issues/7068

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

https://api.github.com/repos/huggingface/datasets/issues/7068/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7068/comments

https://api.github.com/repos/huggingface/datasets/issues/7068/events

https://github.com/huggingface/datasets/pull/7068

PR_kwDODunzps52SwXS

Fix prepare_single_hop_path_and_storage_options

closed

MEMBER

Fix `_prepare_single_hop_path_and_storage_options`: - Do not pass HF authentication headers and HF user-agent to non-HF HTTP URLs - Do not overwrite passed `storage_options` nested values: - Before, when passed ```DownloadConfig(storage_options={"https": {"client_kwargs": {"raise_for_status": True}}})```, it was overwritten to ```{"https": {"client_kwargs": {"trust_env": True}}}``` - Now, the result combines both: ```{"https": {"client_kwargs": {"trust_env": True, "raise_for_status": True}}}```

https://api.github.com/repos/huggingface/datasets/issues/7068/timeline

https://api.github.com/repos/huggingface/datasets/issues/7067

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

https://api.github.com/repos/huggingface/datasets/issues/7067/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7067/comments

https://api.github.com/repos/huggingface/datasets/issues/7067/events

https://github.com/huggingface/datasets/issues/7067

I_kwDODunzps6QkZXI

Convert_to_parquet fails for datasets with multiple configs

closed

NONE

If the dataset has multiple configs, when using the `datasets-cli convert_to_parquet` command to avoid issues with the data viewer caused by loading scripts, the conversion process only successfully converts the data corresponding to the first config. When it starts converting the second config, it throws an error: ``` Traceback (most recent call last): File "/opt/anaconda3/envs/dl/bin/datasets-cli", line 8, in <module> sys.exit(main()) File "/opt/anaconda3/envs/dl/lib/python3.10/site-packages/datasets/commands/datasets_cli.py", line 41, in main service.run() File "/opt/anaconda3/envs/dl/lib/python3.10/site-packages/datasets/commands/convert_to_parquet.py", line 83, in run dataset.push_to_hub( File "/opt/anaconda3/envs/dl/lib/python3.10/site-packages/datasets/dataset_dict.py", line 1713, in push_to_hub api.create_branch(repo_id, branch=revision, token=token, repo_type="dataset", exist_ok=True) File "/opt/anaconda3/envs/dl/lib/python3.10/site-packages/huggingface_hub/utils/_validators.py", line 114, in _inner_fn return fn(*args, **kwargs) File "/opt/anaconda3/envs/dl/lib/python3.10/site-packages/huggingface_hub/hf_api.py", line 5503, in create_branch hf_raise_for_status(response) File "/opt/anaconda3/envs/dl/lib/python3.10/site-packages/huggingface_hub/utils/_errors.py", line 358, in hf_raise_for_status raise BadRequestError(message, response=response) from e huggingface_hub.utils._errors.BadRequestError: (Request ID: Root=1-669fc665-7c2e80d75f4337496ee95402;731fcdc7-0950-4eec-99cf-ce047b8d003f) Bad request: Invalid reference for a branch: refs/pr/1 ```

https://api.github.com/repos/huggingface/datasets/issues/7067/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7066

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

https://api.github.com/repos/huggingface/datasets/issues/7066/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7066/comments

https://api.github.com/repos/huggingface/datasets/issues/7066/events

https://github.com/huggingface/datasets/issues/7066

I_kwDODunzps6QjHko

One subset per file in repo ?

open

MEMBER

Right now we consider all the files of a dataset to be the same data, e.g. ``` single_subset_dataset/ ├── train0.jsonl ├── train1.jsonl └── train2.jsonl ``` but in cases like this, each file is actually a different subset of the dataset and should be loaded separately ``` many_subsets_dataset/ ├── animals.jsonl ├── trees.jsonl └── metadata.jsonl ``` It would be nice to detect those subsets automatically using a simple heuristic. For example we can group files together if their paths names are the same except some digits ?

https://api.github.com/repos/huggingface/datasets/issues/7066/timeline

https://api.github.com/repos/huggingface/datasets/issues/7065

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

https://api.github.com/repos/huggingface/datasets/issues/7065/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7065/comments

https://api.github.com/repos/huggingface/datasets/issues/7065/events

https://github.com/huggingface/datasets/issues/7065

I_kwDODunzps6QhoTp

Cannot get item after loading from disk and then converting to iterable.

open

NONE

### Describe the bug The dataset generated from local file works fine. ```py root = "/home/data/train" file_list1 = glob(os.path.join(root, "*part1.flac")) file_list2 = glob(os.path.join(root, "*part2.flac")) ds = ( Dataset.from_dict({"part1": file_list1, "part2": file_list2}) .cast_column("part1", Audio(sampling_rate=None, mono=False)) .cast_column("part2", Audio(sampling_rate=None, mono=False)) ) ids = ds.to_iterable_dataset(128) ids = ids.shuffle(buffer_size=10000, seed=42) dataloader = DataLoader(ids, num_workers=4, batch_size=8, persistent_workers=True) for batch in dataloader: break ``` But after saving it to disk and then loading it from disk, I cannot get data as expected. ```py root = "/home/data/train" file_list1 = glob(os.path.join(root, "*part1.flac")) file_list2 = glob(os.path.join(root, "*part2.flac")) ds = ( Dataset.from_dict({"part1": file_list1, "part2": file_list2}) .cast_column("part1", Audio(sampling_rate=None, mono=False)) .cast_column("part2", Audio(sampling_rate=None, mono=False)) ) ds.save_to_disk("./train") ds = datasets.load_from_disk("./train") ids = ds.to_iterable_dataset(128) ids = ids.shuffle(buffer_size=10000, seed=42) dataloader = DataLoader(ids, num_workers=4, batch_size=8, persistent_workers=True) for batch in dataloader: break ``` After a long time waiting, an error occurs: ``` Loading dataset from disk: 100%|█████████████████████████████████████████████████████████████████████████| 165/165 [00:00<00:00, 6422.18it/s] Traceback (most recent call last): File "/home/hanzerui/.conda/envs/mss/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1133, in _try_get_data data = self._data_queue.get(timeout=timeout) File "/home/hanzerui/.conda/envs/mss/lib/python3.10/multiprocessing/queues.py", line 113, in get if not self._poll(timeout): File "/home/hanzerui/.conda/envs/mss/lib/python3.10/multiprocessing/connection.py", line 257, in poll return self._poll(timeout) File "/home/hanzerui/.conda/envs/mss/lib/python3.10/multiprocessing/connection.py", line 424, in _poll r = wait([self], timeout) File "/home/hanzerui/.conda/envs/mss/lib/python3.10/multiprocessing/connection.py", line 931, in wait ready = selector.select(timeout) File "/home/hanzerui/.conda/envs/mss/lib/python3.10/selectors.py", line 416, in select fd_event_list = self._selector.poll(timeout) File "/home/hanzerui/.conda/envs/mss/lib/python3.10/site-packages/torch/utils/data/_utils/signal_handling.py", line 66, in handler _error_if_any_worker_fails() RuntimeError: DataLoader worker (pid 3490529) is killed by signal: Killed. The above exception was the direct cause of the following exception: Traceback (most recent call last): File "/home/hanzerui/.conda/envs/mss/lib/python3.10/runpy.py", line 196, in _run_module_as_main return _run_code(code, main_globals, None, File "/home/hanzerui/.conda/envs/mss/lib/python3.10/runpy.py", line 86, in _run_code exec(code, run_globals) File "/home/hanzerui/.vscode-server/extensions/ms-python.debugpy-2024.9.12011011/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/__main__.py", line 39, in <module> cli.main() File "/home/hanzerui/.vscode-server/extensions/ms-python.debugpy-2024.9.12011011/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/../debugpy/server/cli.py", line 430, in main run() File "/home/hanzerui/.vscode-server/extensions/ms-python.debugpy-2024.9.12011011/bundled/libs/debugpy/adapter/../../debugpy/launcher/../../debugpy/../debugpy/server/cli.py", line 284, in run_file runpy.run_path(target, run_name="__main__") File "/home/hanzerui/.vscode-server/extensions/ms-python.debugpy-2024.9.12011011/bundled/libs/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 321, in run_path return _run_module_code(code, init_globals, run_name, File "/home/hanzerui/.vscode-server/extensions/ms-python.debugpy-2024.9.12011011/bundled/libs/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 135, in _run_module_code _run_code(code, mod_globals, init_globals, File "/home/hanzerui/.vscode-server/extensions/ms-python.debugpy-2024.9.12011011/bundled/libs/debugpy/_vendored/pydevd/_pydevd_bundle/pydevd_runpy.py", line 124, in _run_code exec(code, run_globals) File "/home/hanzerui/workspace/NetEase/test/test_datasets.py", line 60, in <module> for batch in dataloader: File "/home/hanzerui/.conda/envs/mss/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 631, in __next__ data = self._next_data() File "/home/hanzerui/.conda/envs/mss/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1329, in _next_data idx, data = self._get_data() File "/home/hanzerui/.conda/envs/mss/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1295, in _get_data success, data = self._try_get_data() File "/home/hanzerui/.conda/envs/mss/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 1146, in _try_get_data raise RuntimeError(f'DataLoader worker (pid(s) {pids_str}) exited unexpectedly') from e RuntimeError: DataLoader worker (pid(s) 3490529) exited unexpectedly ``` It seems that streaming is not supported by `laod_from_disk`, so does that mean I cannot convert it to iterable? ### Steps to reproduce the bug 1. Create a `Dataset` from local files with `from_dict` 2. Save it to disk with `save_to_disk` 3. Load it from disk with `load_from_disk` 4. Convert to iterable with `to_iterable_dataset` 5. Loop the dataset ### Expected behavior Get items faster than the original dataset generated from dict. ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-6.5.0-41-generic-x86_64-with-glibc2.35 - Python version: 3.10.14 - `huggingface_hub` version: 0.23.2 - PyArrow version: 17.0.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7065/timeline

https://api.github.com/repos/huggingface/datasets/issues/7064

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

https://api.github.com/repos/huggingface/datasets/issues/7064/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7064/comments

https://api.github.com/repos/huggingface/datasets/issues/7064/events

https://github.com/huggingface/datasets/pull/7064

PR_kwDODunzps52Lz2-

Add `batch` method to `Dataset` class

closed

CONTRIBUTOR

This PR introduces a new `batch` method to the `Dataset` class, aligning its functionality with the `IterableDataset.batch()` method (implemented in #7054). The implementation uses as well the existing `map` method for efficient batching of examples. Key changes: - Add `batch` method to `Dataset` class in `arrow_dataset.py` - Utilize `map` method for batching Closes #7063 Once the approach is approved, i will create the tests and update the documentation.

https://api.github.com/repos/huggingface/datasets/issues/7064/timeline

https://api.github.com/repos/huggingface/datasets/issues/7063

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

https://api.github.com/repos/huggingface/datasets/issues/7063/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7063/comments

https://api.github.com/repos/huggingface/datasets/issues/7063/events

https://github.com/huggingface/datasets/issues/7063

I_kwDODunzps6QgsLI

Add `batch` method to `Dataset`

closed

CONTRIBUTOR

### Feature request Add a `batch` method to the Dataset class, similar to the one recently implemented for `IterableDataset` in PR #7054. ### Motivation A batched iteration speeds up data loading significantly (see e.g. #6279) ### Your contribution I plan to open a PR to implement this.

https://api.github.com/repos/huggingface/datasets/issues/7063/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7062

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

https://api.github.com/repos/huggingface/datasets/issues/7062/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7062/comments

https://api.github.com/repos/huggingface/datasets/issues/7062/events

https://github.com/huggingface/datasets/pull/7062

PR_kwDODunzps52LUPR

Avoid calling http_head for non-HTTP URLs

closed

MEMBER

Avoid calling `http_head` for non-HTTP URLs, by adding and `else` statement. Currently, it makes an unnecessary HTTP call (which adds latency) for non-HTTP protocols, like FTP, S3,... I discovered this while working in an unrelated issue.

https://api.github.com/repos/huggingface/datasets/issues/7062/timeline

https://api.github.com/repos/huggingface/datasets/issues/7061

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

https://api.github.com/repos/huggingface/datasets/issues/7061/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7061/comments

https://api.github.com/repos/huggingface/datasets/issues/7061/events

https://github.com/huggingface/datasets/issues/7061

I_kwDODunzps6QeA2B

Custom Dataset | Still Raise Error while handling errors in _generate_examples

open

NONE

### Describe the bug I follow this [example](https://discuss.huggingface.co/t/error-handling-in-iterabledataset/72827/3) to handle errors in custom dataset. I am writing a dataset script which read jsonl files and i need to handle errors and continue reading files without raising exception and exit the execution. ``` def _generate_examples(self, filepaths): errors=[] id_ = 0 for filepath in filepaths: try: with open(filepath, 'r') as f: for line in f: json_obj = json.loads(line) yield id_, json_obj id_ += 1 except Exception as exc: logger.error(f"error occur at filepath: {filepath}") errors.append(error) ``` seems the logger.error is printed but still exception is raised the the run is exit. ``` Downloading and preparing dataset custom_dataset/default to /home/myuser/.cache/huggingface/datasets/custom_dataset/default-a14cdd566afee0a6/1.0.0/acfcc9fb9c57034b580c4252841 ERROR: datasets_modules.datasets.custom_dataset.acfcc9fb9c57034b580c4252841bb890a5617cbd28678dd4be5e52b81188ad02.custom_dataset: 2024-07-22 10:47:42,167: error occur at filepath: '/home/myuser/ds/corrupted-file.jsonl Traceback (most recent call last): File "/home/myuser/.cache/huggingface/modules/datasets_modules/datasets/custom_dataset/ac..2/custom_dataset.py", line 48, in _generate_examples json_obj = json.loads(line) File "myenv/lib/python3.8/json/__init__.py", line 357, in loads return _default_decoder.decode(s) File "myenv/lib/python3.8/json/decoder.py", line 337, in decode obj, end = self.raw_decode(s, idx=_w(s, 0).end()) File "myenv/lib/python3.8/json/decoder.py", line 353, in raw_decode obj, end = self.scan_once(s, idx) json.decoder.JSONDecodeError: Invalid control character at: line 1 column 4 (char 3) Generating train split: 0 examples [00:06, ? examples/s]> RemoteTraceback: """ Traceback (most recent call last): File "myenv/lib/python3.8/site-packages/datasets/builder.py", line 1637, in _prepare_split_single num_examples, num_bytes = writer.finalize() File "myenv/lib/python3.8/site-packages/datasets/arrow_writer.py", line 594, in finalize raise SchemaInferenceError("Please pass `features` or at least one example when writing data") datasets.arrow_writer.SchemaInferenceError: Please pass `features` or at least one example when writing data The above exception was the direct cause of the following exception: Traceback (most recent call last): File "myenv/lib/python3.8/site-packages/multiprocess/pool.py", line 125, in worker result = (True, func(*args, **kwds)) File "myenv/lib/python3.8/site-packages/datasets/utils/py_utils.py", line 1353, in _write_generator_to_queue for i, result in enumerate(func(**kwargs)): File "myenv/lib/python3.8/site-packages/datasets/builder.py", line 1646, 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 """ The above exception was the direct cause of the following exception: │ │ │ myenv/lib/python3.8/site-packages/datasets/utils/py_utils. │ │ py:1377 in <listcomp> │ │ │ │ 1374 │ │ │ │ if all(async_result.ready() for async_result in async_results) and queue │ │ 1375 │ │ │ │ │ break │ │ 1376 │ │ # we get the result in case there's an error to raise │ │ ❱ 1377 │ │ [async_result.get() for async_result in async_results] │ │ 1378 │ │ │ │ ╭──────────────────────────────── locals ─────────────────────────────────╮ │ │ │ .0 = <list_iterator object at 0x7f2cc1f0ce20> │ │ │ │ async_result = <multiprocess.pool.ApplyResult object at 0x7f2cc1f79c10> │ │ │ ╰─────────────────────────────────────────────────────────────────────────╯ │ │ │ │ myenv/lib/python3.8/site-packages/multiprocess/pool.py:771 │ │ in get │ │ │ │ 768 │ │ if self._success: │ │ 769 │ │ │ return self._value │ │ 770 │ │ else: │ │ ❱ 771 │ │ │ raise self._value │ │ 772 │ │ │ 773 │ def _set(self, i, obj): │ │ 774 │ │ self._success, self._value = obj │ │ │ │ ╭────────────────────────────── locals ──────────────────────────────╮ │ │ │ self = <multiprocess.pool.ApplyResult object at 0x7f2cc1f79c10> │ │ │ │ timeout = None │ │ │ ╰────────────────────────────────────────────────────────────────────╯ │ DatasetGenerationError: An error occurred while generating the dataset ``` ### Steps to reproduce the bug same as above ### Expected behavior should handle error and continue reading remaining files ### Environment info python 3.9

https://api.github.com/repos/huggingface/datasets/issues/7061/timeline

https://api.github.com/repos/huggingface/datasets/issues/7060

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

https://api.github.com/repos/huggingface/datasets/issues/7060/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7060/comments

https://api.github.com/repos/huggingface/datasets/issues/7060/events

https://github.com/huggingface/datasets/pull/7060

PR_kwDODunzps52G71g

WebDataset BuilderConfig

closed

NONE

This PR adds `WebDatasetConfig`. Closes #7055

https://api.github.com/repos/huggingface/datasets/issues/7060/timeline

https://api.github.com/repos/huggingface/datasets/issues/7059

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

https://api.github.com/repos/huggingface/datasets/issues/7059/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7059/comments

https://api.github.com/repos/huggingface/datasets/issues/7059/events

https://github.com/huggingface/datasets/issues/7059

I_kwDODunzps6QaWt0

None values are skipped when reading jsonl in subobjects

open

NONE

### Describe the bug I have been fighting against my machine since this morning only to find out this is some kind of a bug. When loading a dataset composed of `metadata.jsonl`, if you have nullable values (Optional[str]), they can be ignored by the parser, shifting things around. E.g., let's take this example Here are two version of a same dataset: [not-buggy.tar.gz](https://github.com/user-attachments/files/16333532/not-buggy.tar.gz) [buggy.tar.gz](https://github.com/user-attachments/files/16333553/buggy.tar.gz) ### Steps to reproduce the bug 1. Load the `buggy.tar.gz` dataset 2. Print baseline of `dts = load_dataset("./data")["train"][0]["baselines]` 3. Load the `not-buggy.tar.gz` dataset 4. Print baseline of `dts = load_dataset("./data")["train"][0]["baselines]` ### Expected behavior Both should have 4 baseline entries: 1. Buggy should have None followed by three lists 2. Non-Buggy should have four lists, and the first one should be an empty list. One does not work, 2 works. Despite accepting None in another position than the first one. ### Environment info - `datasets` version: 2.19.1 - Platform: Linux-6.5.0-44-generic-x86_64-with-glibc2.35 - Python version: 3.10.12 - `huggingface_hub` version: 0.23.0 - PyArrow version: 16.1.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.3.1

https://api.github.com/repos/huggingface/datasets/issues/7059/timeline

https://api.github.com/repos/huggingface/datasets/issues/7058

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

https://api.github.com/repos/huggingface/datasets/issues/7058/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7058/comments

https://api.github.com/repos/huggingface/datasets/issues/7058/events

https://github.com/huggingface/datasets/issues/7058

I_kwDODunzps6QZVZj

New feature type: Document

open

COLLABORATOR

It would be useful for PDF. https://github.com/huggingface/dataset-viewer/issues/2991#issuecomment-2242656069

https://api.github.com/repos/huggingface/datasets/issues/7058/timeline

https://api.github.com/repos/huggingface/datasets/issues/7057

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

https://api.github.com/repos/huggingface/datasets/issues/7057/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7057/comments

https://api.github.com/repos/huggingface/datasets/issues/7057/events

https://github.com/huggingface/datasets/pull/7057

PR_kwDODunzps52EjGC

Update load_hub.mdx

closed

COLLABORATOR

https://api.github.com/repos/huggingface/datasets/issues/7057/timeline

https://api.github.com/repos/huggingface/datasets/issues/7056

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

https://api.github.com/repos/huggingface/datasets/issues/7056/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7056/comments

https://api.github.com/repos/huggingface/datasets/issues/7056/events

https://github.com/huggingface/datasets/pull/7056

PR_kwDODunzps52DgOu

Make `BufferShuffledExamplesIterable` resumable

open

CONTRIBUTOR

This PR aims to implement a resumable `BufferShuffledExamplesIterable`. Instead of saving the entire buffer content, which is very memory-intensive, the newly implemented `BufferShuffledExamplesIterable` saves only the minimal state necessary for recovery, e.g., the random generator states and the state of the first example in the buffer dict. The idea is that since the buffer size is limited, even if the entire buffer is discarded, we can rebuild it as long as the state of the oldest example is recorded. For buffer size $B$, the expected distance between when an example is pushed and when it is yielded is $d = \sum_{k=1}^{\infty} k\frac{1}{B} (1 - \frac{1}{B} )^{k-1} =B$. Simulation experiments support these claims: ```py from random import randint BUFFER_SIZE = 1024 dists = [] buffer = [] for i in range(10000000): if i < BUFFER_SIZE: buffer.append(i) else: index = randint(0, BUFFER_SIZE - 1) dists.append(i - buffer[index]) buffer[index] = i print(f"MIN DIST: {min(dists)}\nMAX DIST: {max(dists)}\nAVG DIST: {sum(dists) / len(dists):.2f}\n") ``` which produces the following output: ```py MIN DIST: 1 MAX DIST: 15136 AVG DIST: 1023.95 ``` The overall time for reconstructing the buffer and recovery should not be too long. The following code mimics the cases of resuming online tokenization by `datasets` and `StatefulDataLoader` under distributed scenarios, ```py import pickle import time from itertools import chain from typing import Any, Dict, List import torch from datasets import load_dataset from torchdata.stateful_dataloader import StatefulDataLoader from tqdm import tqdm from transformers import AutoTokenizer, DataCollatorForLanguageModeling tokenizer = AutoTokenizer.from_pretrained('fla-hub/gla-1.3B-100B') tokenizer.pad_token = tokenizer.eos_token data_collator = DataCollatorForLanguageModeling(tokenizer=tokenizer, mlm=False) torch.manual_seed(42) def tokenize(examples: Dict[str, List[Any]]) -> Dict[str, List[List[int]]]: input_ids = tokenizer(examples['text'])['input_ids'] input_ids = list(chain(*input_ids)) total_length = len(input_ids) chunk_size = 2048 total_length = (total_length // chunk_size) * chunk_size # the last chunk smaller than chunk_size will be discarded return {'input_ids': [input_ids[i: i+chunk_size] for i in range(0, total_length, chunk_size)]} batch_size = 16 num_workers = 5 context_length = 2048 rank = 1 world_size = 32 prefetch_factor = 2 steps = 2048 path = 'fla-hub/slimpajama-test' dataset = load_dataset( path=path, split='train', streaming=True, trust_remote_code=True ) dataset = dataset.map(tokenize, batched=True, remove_columns=next(iter(dataset)).keys()) dataset = dataset.shuffle(seed=42) loader = StatefulDataLoader(dataset=dataset, batch_size=batch_size, collate_fn=data_collator, num_workers=num_workers, persistent_workers=False, prefetch_factor=prefetch_factor) start = time.time() for i, batch in tqdm(enumerate(loader)): if i == 0: print(f'{i}\n{batch["input_ids"]}') if i == steps - 1: print(f'{i}\n{batch["input_ids"]}') state_dict = loader.state_dict() if i == steps: print(f'{i}\n{batch["input_ids"]}') break print(f"{time.time() - start:.2f}s elapsed") print(f"{len(pickle.dumps(state_dict)) / 1024**2:.2f}MB states in total") for worker in state_dict['_snapshot']['_worker_snapshots'].keys(): print(f"{worker} {len(pickle.dumps(state_dict['_snapshot']['_worker_snapshots'][worker])) / 1024**2:.2f}MB") print(state_dict['_snapshot']['_worker_snapshots']['worker_0']['dataset_state']) loader = StatefulDataLoader(dataset=dataset, batch_size=batch_size, collate_fn=data_collator, num_workers=num_workers, persistent_workers=False, prefetch_factor=prefetch_factor) print("Loading state dict") loader.load_state_dict(state_dict) start = time.time() for batch in loader: print(batch['input_ids']) break print(f"{time.time() - start:.2f}s elapsed") ``` and the outputs are ```py 0 tensor([[ 909, 395, 19082, ..., 13088, 16232, 395], [ 601, 28705, 28770, ..., 28733, 923, 288], [21753, 15071, 13977, ..., 9369, 28723, 415], ..., [21763, 28751, 20300, ..., 28781, 28734, 4775], [ 354, 396, 10214, ..., 298, 429, 28770], [ 333, 6149, 28768, ..., 2773, 340, 351]]) 2047 tensor([[28723, 415, 3889, ..., 272, 3065, 2609], [ 403, 3214, 3629, ..., 403, 21163, 16434], [28723, 13, 28749, ..., 28705, 28750, 28734], ..., [ 2778, 2251, 28723, ..., 354, 684, 429], [ 5659, 298, 1038, ..., 5290, 297, 22153], [ 938, 28723, 1537, ..., 9123, 28733, 12154]]) 2048 tensor([[ 769, 278, 12531, ..., 28721, 19309, 28739], [ 415, 23347, 622, ..., 3937, 2426, 28725], [28745, 4345, 28723, ..., 338, 28725, 583], ..., [ 1670, 28709, 5809, ..., 28734, 28760, 393], [ 340, 1277, 624, ..., 325, 28790, 1329], [ 523, 1144, 3409, ..., 359, 359, 17422]]) 65.97s elapsed 0.00MB states in total worker_0 0.00MB worker_1 0.00MB worker_2 0.00MB worker_3 0.00MB worker_4 0.00MB {'ex_iterable': {'ex_iterable': {'shard_idx': 0, 'shard_example_idx': 14000}, 'num_examples_since_previous_state': 166, 'previous_state_example_idx': 7394, 'previous_state': {'shard_idx': 0, 'shard_example_idx': 13000}}, 'num_taken': 6560, 'global_example_idx': 7560, 'buffer_state_dict': {'num_taken': 6560, 'global_example_idx': 356, 'index_offset': 0, 'first_state': {'ex_iterable': {'shard_idx': 0, 'shard_example_idx': 1000}, 'num_examples_since_previous_state': 356, 'previous_state_example_idx': 0, 'previous_state': {'shard_idx': 0, 'shard_example_idx': 0}}, 'bit_generator_state': {'state': {'state': 274674114334540486603088602300644985544, 'inc': 332724090758049132448979897138935081983}, 'bit_generator': 'PCG64', 'has_uint32': 0, 'uinteger': 0}}} Loading state dict tensor([[ 769, 278, 12531, ..., 28721, 19309, 28739], [ 415, 23347, 622, ..., 3937, 2426, 28725], [28745, 4345, 28723, ..., 338, 28725, 583], ..., [ 1670, 28709, 5809, ..., 28734, 28760, 393], [ 340, 1277, 624, ..., 325, 28790, 1329], [ 523, 1144, 3409, ..., 359, 359, 17422]]) 24.60s elapsed ``` Not sure if this PR complies with the `datasets` code style. Looking for your help @lhoestq, also very willing to further improve the code if any suggestions are given.

https://api.github.com/repos/huggingface/datasets/issues/7056/timeline

https://api.github.com/repos/huggingface/datasets/issues/7055

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

https://api.github.com/repos/huggingface/datasets/issues/7055/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7055/comments

https://api.github.com/repos/huggingface/datasets/issues/7055/events

https://github.com/huggingface/datasets/issues/7055

I_kwDODunzps6QWFhb

WebDataset with different prefixes are unsupported

closed

NONE

### Describe the bug Consider a WebDataset with multiple images for each item where the number of images may vary: [example](https://huggingface.co/datasets/bigdata-pw/fashion-150k) Due to this [code](https://github.com/huggingface/datasets/blob/87f4c2088854ff33e817e724e75179e9975c1b02/src/datasets/packaged_modules/webdataset/webdataset.py#L76-L80) an error is given. ``` The TAR archives of the dataset should be in WebDataset format, but the files in the archive don't share the same prefix or the same types. ``` The purpose of this check is unclear because PyArrow supports different keys. Removing the check allows the dataset to be loaded and there's no issue when iterating through the dataset. ``` >>> from datasets import load_dataset >>> path = "shards/*.tar" >>> dataset = load_dataset("webdataset", data_files={"train": path}, split="train", streaming=True) Resolving data files: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 152/152 [00:00<00:00, 56458.93it/s] >>> dataset IterableDataset({ features: ['__key__', '__url__', '1.jpg', '2.jpg', '3.jpg', '4.jpg', 'json'], n_shards: 152 }) ``` ### Steps to reproduce the bug ```python from datasets import load_dataset load_dataset("bigdata-pw/fashion-150k") ``` ### Expected behavior Dataset loads without error ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-5.14.0-467.el9.x86_64-x86_64-with-glibc2.34 - Python version: 3.9.19 - `huggingface_hub` version: 0.23.4 - PyArrow version: 17.0.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7055/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7054

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

https://api.github.com/repos/huggingface/datasets/issues/7054/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7054/comments

https://api.github.com/repos/huggingface/datasets/issues/7054/events

https://github.com/huggingface/datasets/pull/7054

PR_kwDODunzps514T1f

Add batching to `IterableDataset`

closed

CONTRIBUTOR

I've taken a try at implementing a batched `IterableDataset` as requested in issue #6279. This PR adds a new `BatchedExamplesIterable` class and a `.batch()` method to the `IterableDataset` class. The main changes are: 1. A new `BatchedExamplesIterable` that groups examples into batches. 2. A `.batch()` method for `IterableDataset` to easily create batched versions. 3. Support for shuffling and sharding to work with PyTorch DataLoader and multiple workers. I'm not sure if this is exactly what you had in mind and also have not fully tested it atm, so I'd really appreciate your feedback. Does this seem like it's heading in the right direction? I'm happy to make any changes or explore different approaches if needed. Pinging @lhoestq

https://api.github.com/repos/huggingface/datasets/issues/7054/timeline

https://api.github.com/repos/huggingface/datasets/issues/7053

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

https://api.github.com/repos/huggingface/datasets/issues/7053/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7053/comments

https://api.github.com/repos/huggingface/datasets/issues/7053/events

https://github.com/huggingface/datasets/issues/7053

I_kwDODunzps6QB7Nv

Datasets.datafiles resolve_pattern `TypeError: can only concatenate tuple (not "str") to tuple`

closed

NONE

### Describe the bug in data_files.py, line 332, `fs, _, _ = get_fs_token_paths(pattern, storage_options=storage_options)` If we run the code on AWS, as fs.protocol will be a tuple like: `('file', 'local')` So, `isinstance(fs.protocol, str) == False` and `protocol_prefix = fs.protocol + "://" if fs.protocol != "file" else ""` will raise `TypeError: can only concatenate tuple (not "str") to tuple`. ### Steps to reproduce the bug Steps to reproduce: 1. Run on a cloud server like AWS, 2. `import datasets.data_files as datafile` 3. datafile.resolve_pattern('path/to/dataset', '.') 4. `TypeError: can only concatenate tuple (not "str") to tuple` ### Expected behavior Should return path of the dataset, with fs.protocol at the beginning ### Environment info - `datasets` version: 2.14.0 - Platform: Linux-3.10.0-1160.119.1.el7.x86_64-x86_64-with-glibc2.17 - Python version: 3.8.19 - Huggingface_hub version: 0.23.5 - PyArrow version: 16.1.0 - Pandas version: 1.1.5

https://api.github.com/repos/huggingface/datasets/issues/7053/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7052

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

https://api.github.com/repos/huggingface/datasets/issues/7052/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7052/comments

https://api.github.com/repos/huggingface/datasets/issues/7052/events

https://github.com/huggingface/datasets/pull/7052

PR_kwDODunzps51iuop

Adding `Music` feature for symbolic music modality (MIDI, abc)

closed

NONE

⚠️ (WIP) ⚠️ ### What this PR does This PR adds a `Music` feature for the symbolic music modality, in particular [MIDI](https://en.wikipedia.org/wiki/Musical_Instrument_Digital_Interface) and [abc](https://en.wikipedia.org/wiki/ABC_notation) files. ### Motivations These two file formats are widely used in the [Music Information Retrieval (MIR)](https://en.wikipedia.org/wiki/Music_information_retrieval) for tasks such as music generation, music transcription, music synthesis or music transcription. Having a dedicated feature in the datasets library would allow to both encourage researchers to share datasets of this modality as well as making them more easily usable for end users, benefitting from the perks of the library. These file formats are supported by [symusic](https://github.com/Yikai-Liao/symusic), a lightweight Python library with C bindings (using nanobind) allowing to efficiently read, write and manipulate them. The library is actively developed, and can in the future also implement other file formats such as [musicXML](https://en.wikipedia.org/wiki/MusicXML). As such, this PR relies on it. The music data can then easily be tokenized with appropriate tokenizers such as [MidiTok](https://github.com/Natooz/MidiTok) or converted to pianorolls matrices by symusic. **Jul 16th 2024:** * the tests for the `Music` feature are currently failing due to non-supported access to the LazyBatch in `test_dataset_with_music_feature_map` and `test_dataset_with_music_feature_map_resample_music` (see TODOs). I am a beginner with pyArrow, I'll take any advice to make this work; * additional tests including the `Music` feature with parquet and WebDataset should be implemented. As of right now, I am waiting for your feedback before taking further steps; * a `MusicFolder` should also be implemented to comply with the usages of the `Image` and `Audio` features, waiting for your feedback too. CCing @lhoestq and @albertvillanova

https://api.github.com/repos/huggingface/datasets/issues/7052/timeline

https://api.github.com/repos/huggingface/datasets/issues/7051

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

https://api.github.com/repos/huggingface/datasets/issues/7051/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7051/comments

https://api.github.com/repos/huggingface/datasets/issues/7051/events

https://github.com/huggingface/datasets/issues/7051

I_kwDODunzps6Pm9LJ

How to set_epoch with interleave_datasets?

closed

NONE

Let's say I have dataset A which has 100k examples, and dataset B which has 100m examples. I want to train on an interleaved dataset of A+B, with stopping_strategy='all_exhausted' so dataset B doesn't repeat any examples. But every time A is exhausted I want it to be reshuffled (eg. calling set_epoch) Of course I want to interleave as IterableDatasets / streaming mode so B doesn't have to get tokenized completely at the start. How could I achieve this? I was thinking something like, if I wrap dataset A in some new IterableDataset with from_generator() and manually call set_epoch before interleaving it? But I'm not sure how to keep the number of shards in that dataset... Something like ``` dataset_a = load_dataset(...) dataset_b = load_dataset(...) def epoch_shuffled_dataset(ds): # How to make this maintain the number of shards in ds?? for epoch in itertools.count(): ds.set_epoch(epoch) yield from iter(ds) shuffled_dataset_a = IterableDataset.from_generator(epoch_shuffled_dataset, gen_kwargs={'ds': dataset_a}) interleaved = interleave_datasets([shuffled_dataset_a, dataset_b], probs, stopping_strategy='all_exhausted') ```

https://api.github.com/repos/huggingface/datasets/issues/7051/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7050

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

https://api.github.com/repos/huggingface/datasets/issues/7050/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7050/comments

https://api.github.com/repos/huggingface/datasets/issues/7050/events

https://github.com/huggingface/datasets/pull/7050

PR_kwDODunzps51Z1Yp

add checkpoint and resume title in docs

closed

MEMBER

(minor) just to make it more prominent in the docs page for the soon-to-be-released new torchdata

https://api.github.com/repos/huggingface/datasets/issues/7050/timeline

https://api.github.com/repos/huggingface/datasets/issues/7049

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

https://api.github.com/repos/huggingface/datasets/issues/7049/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7049/comments

https://api.github.com/repos/huggingface/datasets/issues/7049/events

https://github.com/huggingface/datasets/issues/7049

I_kwDODunzps6PjwM-

Save nparray as list

closed

NONE

### Describe the bug When I use the `map` function to convert images into features, datasets saves nparray as a list. Some people use the `set_format` function to convert the column back, but doesn't this lose precision? ### Steps to reproduce the bug the map function ```python def convert_image_to_features(inst, processor, image_dir): image_file = inst["image_url"] file = image_file.split("/")[-1] image_path = os.path.join(image_dir, file) image = Image.open(image_path) image = image.convert("RGBA") inst["pixel_values"] = processor(images=image, return_tensors="np")["pixel_values"] return inst ``` main function ```python map_fun = partial( convert_image_to_features, processor=processor, image_dir=image_dir ) ds = ds.map(map_fun, batched=False, num_proc=20) print(type(ds[0]["pixel_values"]) ``` ### Expected behavior (type < list>) ### Environment info - `datasets` version: 2.16.1 - Platform: Linux-4.19.91-009.ali4000.alios7.x86_64-x86_64-with-glibc2.35 - Python version: 3.11.5 - `huggingface_hub` version: 0.23.4 - PyArrow version: 14.0.2 - Pandas version: 2.1.4 - `fsspec` version: 2023.10.0

https://api.github.com/repos/huggingface/datasets/issues/7049/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7048

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

https://api.github.com/repos/huggingface/datasets/issues/7048/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7048/comments

https://api.github.com/repos/huggingface/datasets/issues/7048/events

https://github.com/huggingface/datasets/issues/7048

I_kwDODunzps6Pjpp7

ImportError: numpy.core.multiarray when using `filter`

closed

NONE

### Describe the bug I can't apply the filter method on my dataset. ### Steps to reproduce the bug The following snippet generates a bug: ```python from datasets import load_dataset ami = load_dataset('kamilakesbi/ami', 'ihm') ami['train'].filter( lambda example: example["file_name"] == 'EN2001a' ) ``` I get the following error: `ImportError: numpy.core.multiarray failed to import (auto-generated because you didn't call 'numpy.import_array()' after cimporting numpy; use '<void>numpy._import_array' to disable if you are certain you don't need it).` ### Expected behavior It should work properly! ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-5.15.0-67-generic-x86_64-with-glibc2.35 - Python version: 3.10.6 - `huggingface_hub` version: 0.23.4 - PyArrow version: 16.1.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7048/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7047

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

https://api.github.com/repos/huggingface/datasets/issues/7047/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7047/comments

https://api.github.com/repos/huggingface/datasets/issues/7047/events

https://github.com/huggingface/datasets/issues/7047

I_kwDODunzps6PcDNs

Save Dataset as Sharded Parquet

open

NONE

### Feature request `to_parquet` currently saves the dataset as one massive, monolithic parquet file, rather than as several small parquet files. It should shard large datasets automatically. ### Motivation This default behavior makes me very sad because a program I ran for 6 hours saved its results using `to_parquet`, putting the entire billion+ row dataset into a 171 GB *single shard parquet file* which pyarrow, apache spark, etc. all cannot work with without completely exhausting the memory of my system. I was previously able to work with larger-than-memory parquet files, but not this one. I *assume* the reason why this is happening is because it is a single shard. Making sharding the default behavior puts datasets in parity with other frameworks, such as spark, which automatically shard when a large dataset is saved as parquet. ### Your contribution I could change the logic here https://github.com/huggingface/datasets/blob/bf6f41e94d9b2f1c620cf937a2e85e5754a8b960/src/datasets/io/parquet.py#L109-L158 to use `pyarrow.dataset.write_dataset`, which seems to support sharding, or periodically open new files. We would only shard if the user passed in a path rather than file handle.

https://api.github.com/repos/huggingface/datasets/issues/7047/timeline

https://api.github.com/repos/huggingface/datasets/issues/7046

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

https://api.github.com/repos/huggingface/datasets/issues/7046/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7046/comments

https://api.github.com/repos/huggingface/datasets/issues/7046/events

https://github.com/huggingface/datasets/pull/7046

PR_kwDODunzps51N05n

Support librosa and numpy 2.0 for Python 3.10

closed

MEMBER

Support librosa and numpy 2.0 for Python 3.10 by installing soxr 0.4.0b1 pre-release: - https://github.com/dofuuz/python-soxr/releases/tag/v0.4.0b1 - https://github.com/dofuuz/python-soxr/issues/28

https://api.github.com/repos/huggingface/datasets/issues/7046/timeline

https://api.github.com/repos/huggingface/datasets/issues/7045

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

https://api.github.com/repos/huggingface/datasets/issues/7045/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7045/comments

https://api.github.com/repos/huggingface/datasets/issues/7045/events

https://github.com/huggingface/datasets/pull/7045

PR_kwDODunzps51Nsie

Fix tensorflow min version depending on Python version

closed

MEMBER

Fix tensorflow min version depending on Python version. Related to: - #6991

https://api.github.com/repos/huggingface/datasets/issues/7045/timeline

https://api.github.com/repos/huggingface/datasets/issues/7044

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

https://api.github.com/repos/huggingface/datasets/issues/7044/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7044/comments

https://api.github.com/repos/huggingface/datasets/issues/7044/events

https://github.com/huggingface/datasets/pull/7044

PR_kwDODunzps51MLbh

Mark tests that require librosa

closed

MEMBER

Mark tests that require `librosa`. Note that `librosa` is an optional dependency (installed with `audio` option) and we should be able to test environments without that library installed. This is the case if we want to test Numpy 2.0, which is currently incompatible with `librosa` due to its dependency on `soxr`: - https://github.com/dofuuz/python-soxr/issues/28

https://api.github.com/repos/huggingface/datasets/issues/7044/timeline

https://api.github.com/repos/huggingface/datasets/issues/7043

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

https://api.github.com/repos/huggingface/datasets/issues/7043/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7043/comments

https://api.github.com/repos/huggingface/datasets/issues/7043/events

https://github.com/huggingface/datasets/pull/7043

PR_kwDODunzps51MAN0

Add decorator as explicit test dependency

closed

MEMBER

Add decorator as explicit test dependency. We use `decorator` library in our CI test since PR: - #4845 However we did not add it as an explicit test requirement, and we depended on it indirectly through other libraries' dependencies. I discovered this while testing Numpy 2.0 and removing incompatible libraries.

https://api.github.com/repos/huggingface/datasets/issues/7043/timeline

https://api.github.com/repos/huggingface/datasets/issues/7042

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

https://api.github.com/repos/huggingface/datasets/issues/7042/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7042/comments

https://api.github.com/repos/huggingface/datasets/issues/7042/events

https://github.com/huggingface/datasets/pull/7042

PR_kwDODunzps51K8CM

Improved the tutorial by adding a link for loading datasets

closed

CONTRIBUTOR

Improved the tutorial by letting readers know about loading datasets with common files and including a link. I left the local files section alone because the methods were already listed with code snippets.

https://api.github.com/repos/huggingface/datasets/issues/7042/timeline

https://api.github.com/repos/huggingface/datasets/issues/7041

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

https://api.github.com/repos/huggingface/datasets/issues/7041/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7041/comments

https://api.github.com/repos/huggingface/datasets/issues/7041/events

https://github.com/huggingface/datasets/issues/7041

I_kwDODunzps6PUusm

`sort` after `filter` unreasonably slow

open

NONE

### Describe the bug as the tittle says ... ### Steps to reproduce the bug `sort` seems to be normal. ```python from datasets import Dataset import random nums = [{"k":random.choice(range(0,1000))} for _ in range(100000)] ds = Dataset.from_list(nums) print("start sort") ds = ds.sort("k") print("finish sort") ``` but `sort` after `filter` is extremely slow. ```python from datasets import Dataset import random nums = [{"k":random.choice(range(0,1000))} for _ in range(100000)] ds = Dataset.from_list(nums) ds = ds.filter(lambda x:x > 100, input_columns="k") print("start sort") ds = ds.sort("k") print("finish sort") ``` ### Expected behavior Is this a bug, or is it a misuse of the `sort` function? ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-3.10.0-1127.19.1.el7.x86_64-x86_64-with-glibc2.17 - Python version: 3.10.13 - `huggingface_hub` version: 0.23.4 - PyArrow version: 16.1.0 - Pandas version: 2.2.2 - `fsspec` version: 2023.10.0

https://api.github.com/repos/huggingface/datasets/issues/7041/timeline

https://api.github.com/repos/huggingface/datasets/issues/7040

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

https://api.github.com/repos/huggingface/datasets/issues/7040/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7040/comments

https://api.github.com/repos/huggingface/datasets/issues/7040/events

https://github.com/huggingface/datasets/issues/7040

I_kwDODunzps6POZ-_

load `streaming=True` dataset with downloaded cache

open

NONE

### Describe the bug We build a dataset which contains several hdf5 files and write a script using `h5py` to generate the dataset. The hdf5 files are large and the processed dataset cache takes more disk space. So we hope to try streaming iterable dataset. Unfortunately, `h5py` can't convert a remote URL into a hdf5 file descriptor. So we use `fsspec` as an interface like below: ```python def _generate_examples(self, filepath, split): for file in filepath: with fsspec.open(file, "rb") as fs: with h5py.File(fs, "r") as fp: # for event_id in sorted(list(fp.keys())): event_ids = list(fp.keys()) ...... ``` ### Steps to reproduce the bug The `fsspec` works, but it takes 10+ min to print the first 10 examples, which is even longer than the downloading time. I'm not sure if it just caches the whole hdf5 file and generates the examples. ### Expected behavior So does the following make sense so far? 1. download the files ```python dataset = datasets.load('path/to/myscripts', split="train", name="event", trust_remote_code=True) ``` 2. load the iterable dataset faster (using the raw file cache at path `.cache/huggingface/datasets/downloads`) ```python dataset = datasets.load('path/to/myscripts', split="train", name="event", trust_remote_code=True, streaming=true) ``` I made some tests, but the code above can't get the expected result. I'm not sure if this is supported. I also find the issue #6327 . It seemed similar to mine, but I couldn't find a solution. ### Environment info - `datasets` = 2.18.0 - `h5py` = 3.10.0 - `fsspec` = 2023.10.0

https://api.github.com/repos/huggingface/datasets/issues/7040/timeline

https://api.github.com/repos/huggingface/datasets/issues/7039

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

https://api.github.com/repos/huggingface/datasets/issues/7039/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7039/comments

https://api.github.com/repos/huggingface/datasets/issues/7039/events

https://github.com/huggingface/datasets/pull/7039

PR_kwDODunzps51DgCY

Fix export to JSON when dataset larger than batch size

open

MEMBER

Fix export to JSON (`lines=False`) when dataset larger than batch size. Fix #7037.

https://api.github.com/repos/huggingface/datasets/issues/7039/timeline

https://api.github.com/repos/huggingface/datasets/issues/7037

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

https://api.github.com/repos/huggingface/datasets/issues/7037/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7037/comments

https://api.github.com/repos/huggingface/datasets/issues/7037/events

https://github.com/huggingface/datasets/issues/7037

I_kwDODunzps6PEAej

A bug of Dataset.to_json() function

open

NONE

### Describe the bug When using the Dataset.to_json() function, an unexpected error occurs if the parameter is set to lines=False. The stored data should be in the form of a list, but it actually turns into multiple lists, which causes an error when reading the data again. The reason is that to_json() writes to the file in several segments based on the batch size. This is not a problem when lines=True, but it is incorrect when lines=False, because writing in several times will produce multiple lists(when len(dataset) > batch_size). ### Steps to reproduce the bug try this code: ```python from datasets import load_dataset import json train_dataset = load_dataset("Anthropic/hh-rlhf", data_dir="harmless-base")["train"] output_path = "./harmless-base_hftojs.json" print(len(train_dataset)) train_dataset.to_json(output_path, lines=False, force_ascii=False, indent=2) with open(output_path, encoding="utf-8") as f: data = json.loads(f.read()) ``` it raise error: json.decoder.JSONDecodeError: Extra data: line 4003 column 1 (char 1373709) Extra square brackets have appeared here: <img width="265" alt="image" src="https://github.com/huggingface/datasets/assets/26499566/81492332-386d-42e8-88d1-b6d4ae3682cc"> ### Expected behavior The code runs normally. ### Environment info datasets=2.20.0

https://api.github.com/repos/huggingface/datasets/issues/7037/timeline

https://api.github.com/repos/huggingface/datasets/issues/7036

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

https://api.github.com/repos/huggingface/datasets/issues/7036/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7036/comments

https://api.github.com/repos/huggingface/datasets/issues/7036/events

https://github.com/huggingface/datasets/pull/7036

PR_kwDODunzps507bZk

Fix doc generation when NamedSplit is used as parameter default value

closed

MEMBER

Fix doc generation when `NamedSplit` is used as parameter default value. Fix #7035.

https://api.github.com/repos/huggingface/datasets/issues/7036/timeline

https://api.github.com/repos/huggingface/datasets/issues/7035

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

https://api.github.com/repos/huggingface/datasets/issues/7035/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7035/comments

https://api.github.com/repos/huggingface/datasets/issues/7035/events

https://github.com/huggingface/datasets/issues/7035

I_kwDODunzps6PDWrp

Docs are not generated when a parameter defaults to a NamedSplit value

closed

MEMBER

While generating the docs, we get an error when some parameter defaults to a `NamedSplit` value, like: ```python def call_function(split=Split.TRAIN): ... ``` The error is: ValueError: Equality not supported between split train and <class 'inspect._empty'> See: https://github.com/huggingface/datasets/actions/runs/9869660902/job/27254359863?pr=7015 ``` Building the MDX files: 97%|█████████▋| 58/60 [00:00<00:00, 91.94it/s] Traceback (most recent call last): File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/build_doc.py", line 197, in build_mdx_files content, new_anchors, source_files, errors = resolve_autodoc( File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/build_doc.py", line 123, in resolve_autodoc doc = autodoc( File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/autodoc.py", line 499, in autodoc method_doc, check = document_object( File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/autodoc.py", line 395, in document_object signature = format_signature(obj) File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/autodoc.py", line 126, in format_signature if param.default != inspect._empty: File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/datasets/splits.py", line 136, in __ne__ return not self.__eq__(other) File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/datasets/splits.py", line 379, in __eq__ raise ValueError(f"Equality not supported between split {self} and {other}") ValueError: Equality not supported between split train and <class 'inspect._empty'> The above exception was the direct cause of the following exception: Traceback (most recent call last): File "/home/runner/work/datasets/datasets/.venv/bin/doc-builder", line 8, in <module> sys.exit(main()) File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/commands/doc_builder_cli.py", line 47, in main args.func(args) File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/commands/build.py", line 102, in build_command build_doc( File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/build_doc.py", line 367, in build_doc anchors_mapping, source_files_mapping = build_mdx_files( File "/home/runner/work/datasets/datasets/.venv/lib/python3.10/site-packages/doc_builder/build_doc.py", line 230, in build_mdx_files raise type(e)(f"There was an error when converting {file} to the MDX format.\n" + e.args[0]) from e ValueError: There was an error when converting ../datasets/docs/source/package_reference/main_classes.mdx to the MDX format. Equality not supported between split train and <class 'inspect._empty'> ```

https://api.github.com/repos/huggingface/datasets/issues/7035/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7034

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

https://api.github.com/repos/huggingface/datasets/issues/7034/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7034/comments

https://api.github.com/repos/huggingface/datasets/issues/7034/events

https://github.com/huggingface/datasets/pull/7034

PR_kwDODunzps50y-ya

chore: fix typos in docs

closed

CONTRIBUTOR

https://api.github.com/repos/huggingface/datasets/issues/7034/timeline

https://api.github.com/repos/huggingface/datasets/issues/7033

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

https://api.github.com/repos/huggingface/datasets/issues/7033/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7033/comments

https://api.github.com/repos/huggingface/datasets/issues/7033/events

https://github.com/huggingface/datasets/issues/7033

I_kwDODunzps6O5bj4

`from_generator` does not allow to specify the split name

closed

CONTRIBUTOR

### Describe the bug I'm building train, dev, and test using `from_generator`; however, in all three cases, the logger prints `Generating train split:` It's not possible to change the split name since it seems to be hardcoded: https://github.com/huggingface/datasets/blob/main/src/datasets/packaged_modules/generator/generator.py ### Steps to reproduce the bug ``` In [1]: from datasets import Dataset In [2]: def gen(): ...: yield {"pokemon": "bulbasaur", "type": "grass"} ...: In [3]: ds = Dataset.from_generator(gen) Generating train split: 1 examples [00:00, 133.89 examples/s] ``` ### Expected behavior It should be possible to specify any split name ### Environment info - `datasets` version: 2.19.2 - Platform: macOS-10.16-x86_64-i386-64bit - Python version: 3.8.5 - `huggingface_hub` version: 0.23.3 - PyArrow version: 15.0.0 - Pandas version: 2.0.3 - `fsspec` version: 2023.10.0

https://api.github.com/repos/huggingface/datasets/issues/7033/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7032

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

https://api.github.com/repos/huggingface/datasets/issues/7032/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7032/comments

https://api.github.com/repos/huggingface/datasets/issues/7032/events

https://github.com/huggingface/datasets/pull/7032

PR_kwDODunzps50sJTq

Register `.zstd` extension for zstd-compressed files

closed

CONTRIBUTOR

For example, https://huggingface.co/datasets/mlfoundations/dclm-baseline-1.0 dataset files have `.zstd` extension which is currently ignored (only `.zst` is registered).

https://api.github.com/repos/huggingface/datasets/issues/7032/timeline

https://api.github.com/repos/huggingface/datasets/issues/7031

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

https://api.github.com/repos/huggingface/datasets/issues/7031/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7031/comments

https://api.github.com/repos/huggingface/datasets/issues/7031/events

https://github.com/huggingface/datasets/issues/7031

I_kwDODunzps6Oxu3c

CI quality is broken: use ruff check instead

closed

MEMBER

CI quality is broken: https://github.com/huggingface/datasets/actions/runs/9838873879/job/27159697027 ``` error: `ruff <path>` has been removed. Use `ruff check <path>` instead. ```

https://api.github.com/repos/huggingface/datasets/issues/7031/timeline

not_planned

https://api.github.com/repos/huggingface/datasets/issues/7030

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

https://api.github.com/repos/huggingface/datasets/issues/7030/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7030/comments

https://api.github.com/repos/huggingface/datasets/issues/7030/events

https://github.com/huggingface/datasets/issues/7030

I_kwDODunzps6OqJAv

Add option to disable progress bar when reading a dataset ("Loading dataset from disk")

closed

NONE

### Feature request Add an option in load_from_disk to disable the progress bar even if the number of files is larger than 16. ### Motivation I am reading a lot of datasets that it creates lots of logs. <img width="1432" alt="image" src="https://github.com/huggingface/datasets/assets/57996478/8d4bbf03-6b89-44b6-937c-932f01b4eb2a"> ### Your contribution Seems like an easy fix to make. I can create a PR if necessary.

https://api.github.com/repos/huggingface/datasets/issues/7030/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7029

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

https://api.github.com/repos/huggingface/datasets/issues/7029/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7029/comments

https://api.github.com/repos/huggingface/datasets/issues/7029/events

https://github.com/huggingface/datasets/issues/7029

I_kwDODunzps6OiVwo

load_dataset on AWS lambda throws OSError(30, 'Read-only file system') error

open

NONE

### Describe the bug I'm using AWS lambda to run a python application. I run the `load_dataset` function with cache_dir="/tmp" and is still throws the OSError(30, 'Read-only file system') error. Is even updated all the HF envs to point to /tmp dir but the issue still persists. I can confirm that the I can write to /tmp directory. ### Steps to reproduce the bug ```python d = load_dataset( path=hugging_face_link, split=split, token=token, cache_dir="/tmp/hugging_face_cache", ) ``` ### Expected behavior Everything written to the file system as part of the load_datasets function should be in the /tmp directory. ### Environment info datasets version: 2.16.1 Platform: Linux-5.10.216-225.855.amzn2.x86_64-x86_64-with-glibc2.26 Python version: 3.11.9 huggingface_hub version: 0.19.4 PyArrow version: 16.1.0 Pandas version: 2.2.2 fsspec version: 2023.10.0

https://api.github.com/repos/huggingface/datasets/issues/7029/timeline

https://api.github.com/repos/huggingface/datasets/issues/7028

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

https://api.github.com/repos/huggingface/datasets/issues/7028/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7028/comments

https://api.github.com/repos/huggingface/datasets/issues/7028/events

https://github.com/huggingface/datasets/pull/7028

PR_kwDODunzps50dQ1w

Fix ci

closed

MEMBER

...after last pr errors

https://api.github.com/repos/huggingface/datasets/issues/7028/timeline

https://api.github.com/repos/huggingface/datasets/issues/7027

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

https://api.github.com/repos/huggingface/datasets/issues/7027/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7027/comments

https://api.github.com/repos/huggingface/datasets/issues/7027/events

https://github.com/huggingface/datasets/pull/7027

PR_kwDODunzps50dCsE

Missing line from previous pr

closed

MEMBER

https://api.github.com/repos/huggingface/datasets/issues/7027/timeline

https://api.github.com/repos/huggingface/datasets/issues/7026

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

https://api.github.com/repos/huggingface/datasets/issues/7026/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7026/comments

https://api.github.com/repos/huggingface/datasets/issues/7026/events

https://github.com/huggingface/datasets/pull/7026

PR_kwDODunzps50c8Mf

Fix check_library_imports

closed

MEMBER

move it to after the `trust_remote_code` check Note that it only affects local datasets that already exist on disk, not datasets loaded from HF directly

https://api.github.com/repos/huggingface/datasets/issues/7026/timeline

https://api.github.com/repos/huggingface/datasets/issues/7025

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

https://api.github.com/repos/huggingface/datasets/issues/7025/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7025/comments

https://api.github.com/repos/huggingface/datasets/issues/7025/events

https://github.com/huggingface/datasets/pull/7025

PR_kwDODunzps50bSyD

feat: support non streamable arrow file binary format

closed

CONTRIBUTOR

Support Arrow files (`.arrow`) that are in non streamable binary file formats.

https://api.github.com/repos/huggingface/datasets/issues/7025/timeline

https://api.github.com/repos/huggingface/datasets/issues/7024

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

https://api.github.com/repos/huggingface/datasets/issues/7024/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7024/comments

https://api.github.com/repos/huggingface/datasets/issues/7024/events

https://github.com/huggingface/datasets/issues/7024

I_kwDODunzps6Odqq6

Streaming dataset not returning data

open

NONE

### Describe the bug I'm deciding to post here because I'm still not sure what the issue is, or if I am using IterableDatasets wrongly. I'm following the guide on here https://huggingface.co/learn/cookbook/en/fine_tuning_code_llm_on_single_gpu pretty much to a tee and have verified that it works when I'm fine-tuning on the provided dataset. However, I'm doing some data preprocessing steps (filtering out entries), when I try to swap out the dataset for mine, it fails to train. However, I eventually fixed this by simply setting `stream=False` in `load_dataset`. Coud this be some sort of network / firewall issue I'm facing? ### Steps to reproduce the bug I made a post with greater description about how I reproduced this problem before I found my workaround: https://discuss.huggingface.co/t/problem-with-custom-iterator-of-streaming-dataset-not-returning-anything/94551 Here is the problematic dataset snippet, which works when streaming=False (and with buffer keyword removed from shuffle) ``` commitpackft = load_dataset( "chargoddard/commitpack-ft-instruct", split="train", streaming=True ).filter(lambda example: example["language"] == "Python") def form_template(example): """Forms a template for each example following the alpaca format for CommitPack""" example["content"] = ( "### Human: " + example["instruction"] + " " + example["input"] + " ### Assistant: " + example["output"] ) return example dataset = commitpackft.map( form_template, remove_columns=["id", "language", "license", "instruction", "input", "output"], ).shuffle( seed=42, buffer_size=10000 ) # remove everything since its all inside "content" now validation_data = dataset.take(4000) train_data = dataset.skip(4000) ``` The annoying part about this is that it only fails during training and I don't know when it will fail, except that it always fails during evaluation. ### Expected behavior The expected behavior is that I should be able to get something from the iterator when called instead of getting nothing / stuck in a loop somewhere. ### Environment info - `datasets` version: 2.20.0 - Platform: Linux-5.4.0-121-generic-x86_64-with-glibc2.31 - Python version: 3.11.7 - `huggingface_hub` version: 0.23.4 - PyArrow version: 16.1.0 - Pandas version: 2.2.2 - `fsspec` version: 2024.5.0

https://api.github.com/repos/huggingface/datasets/issues/7024/timeline

https://api.github.com/repos/huggingface/datasets/issues/7023

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

https://api.github.com/repos/huggingface/datasets/issues/7023/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7023/comments

https://api.github.com/repos/huggingface/datasets/issues/7023/events

https://github.com/huggingface/datasets/pull/7023

PR_kwDODunzps50TDot

Remove dead code for pyarrow < 15.0.0

closed

MEMBER

Remove dead code for pyarrow < 15.0.0. Code is dead since the merge of: - #6892 Fix #7022.

https://api.github.com/repos/huggingface/datasets/issues/7023/timeline

https://api.github.com/repos/huggingface/datasets/issues/7022

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

https://api.github.com/repos/huggingface/datasets/issues/7022/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7022/comments

https://api.github.com/repos/huggingface/datasets/issues/7022/events

https://github.com/huggingface/datasets/issues/7022

I_kwDODunzps6OVvmK

There is dead code after we require pyarrow >= 15.0.0

closed

MEMBER

There are code lines specific for pyarrow versions < 15.0.0. However, we require pyarrow >= 15.0.0 since the merge of PR: - #6892 Those code lines are now dead code and should be removed.

https://api.github.com/repos/huggingface/datasets/issues/7022/timeline

completed

https://api.github.com/repos/huggingface/datasets/issues/7021

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

https://api.github.com/repos/huggingface/datasets/issues/7021/labels{/name}

https://api.github.com/repos/huggingface/datasets/issues/7021/comments

https://api.github.com/repos/huggingface/datasets/issues/7021/events

https://github.com/huggingface/datasets/pull/7021

PR_kwDODunzps50SlKR

Fix casting list array to fixed size list

closed