Datasets:

amannagrawall002

/

github-issues

like 1

[ "add remove_unused_columns=False to training_args\r\nhttps://github.com/huggingface/datasets/issues/6535#issuecomment-1874024704", "Closing this issue because it was a reported and fixed in transformers." ]

[]

[ "We're unlikely to add more features/support for datasets with python loading scripts, which include datasets with manual download. Sorry for the inconvenience" ]

[ "I can load public dataset, but for my private dataset it fails", "https://huggingface.co/docs/datasets/upload_dataset", "I have checked the API HTTP link. Repository Not Found for url: https://huggingface.co/api/datasets/xxx/xxx.\r\n\r\n\r\n\r\nIt just works fine.", "It seems that everything is in a mass huh....\r\n\r\n\r\n", "https://huggingface.co/datasets/rajpurkar/squad/blob/main/squad.py fails again", "https://github.com/huggingface/datasets/blob/main/templates/new_dataset_script.py#L81 can not use this, too complex. I just need a def to load my file to a dict", "I am facing the same issue. Did you find a fix?", "You should authenticate to be able to access private or gated repos: https://huggingface.co/docs/hub/datasets-gated#access-gated-datasets-as-a-user" ]

[ "Canonical datasets are no longer mentioned in the docs." ]

[ "@xianbaoqian ", "Feel free to open a PR in `m-a-p/COIG-CQIA` to define a default subset. Currently there is no default.\r\n\r\nYou can find some documentation at https://huggingface.co/docs/hub/datasets-manual-configuration#multiple-configurations", "@lhoestq \r\n\r\nWhilst having a default subset readily available (e.g. `all`) by the dataset author is an ideal solution, it is not always the reality.\r\n\r\nWithout the ability to fork the dataset, this can be problematic.\r\n\r\nAs far as I know, it is not possible at all to specify multiple subsets in a generalized programmatic way without hard coding subset names for a specific dataset.\r\n\r\nEven the ability to fetch subset names and loop over them would be sufficient.", "Please note that each subset can have different feature columns, thus making it impossible to load them all into a unique Dataset instance.\r\n\r\nThat is why subsets were created: to support different but related datasets to coexist in a single dataset repository.\r\n\r\nIf you would like to programmatically get the list of subset names, you can use `datasets.get_dataset_config_names`: https://huggingface.co/docs/datasets/v2.20.0/en/load_hub#configurations" ]

[ "Hi ! It seems the documentation was outdated in this paragraph\r\n\r\nI fixed it here: https://github.com/huggingface/datasets/pull/6956", "Fixed." ]

[ "Hi, @lion-ops.\r\n\r\nIn our Continuous Integration we have many tests on loading JSON files and all of them work properly.\r\n\r\nCould you please share your \"train.json\" file, so that we can try to reproduce the issue you have? ", "> Hi, @lion-ops.\r\n> \r\n> In our Continuous Integration we have many tests on loading JSON files and all of them work properly.\r\n> \r\n> Could you please share your \"train.json\" file, so that we can try to reproduce the issue you have?\r\n\r\nThank you for your reply. I can load it normally in another server. Is it possible that the disk of my server is a network disk in the LAN, so it will be downloaded from the LAN and get stuck?" ]

[]

[ "same problem here", "Hello,\r\n\r\nAre you sure you are really using datasets version 2.19.2? We just made the patch release yesterday specifically to fix this issue:\r\n- #6925\r\n\r\nI can't reproduce the error:\r\n```python\r\nIn [1]: from datasets import load_dataset\r\n\r\nIn [2]: ds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation')\r\nDownloading readme: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 41.1k/41.1k [00:00<00:00, 596kB/s]\r\nDownloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 40.7M/40.7M [00:04<00:00, 8.50MB/s]\r\nGenerating validation split: 45576 examples [00:01, 44956.75 examples/s]\r\n\r\nIn [3]: ds\r\nOut[3]: \r\nDataset({\r\n features: ['text', 'timestamp', 'url'],\r\n num_rows: 45576\r\n})\r\n```", "> Hello,\r\n> \r\n> Are you sure you are really using datasets version 2.19.2? We just made the patch release yesterday specifically to fix this issue:\r\n> \r\n> * [Fix NonMatchingSplitsSizesError/ExpectedMoreSplits when passing data_dir/data_files in no-code Hub datasets #6925](https://github.com/huggingface/datasets/pull/6925)\r\n> \r\n> I can't reproduce the error:\r\n> \r\n> ```python\r\n> In [1]: from datasets import load_dataset\r\n> \r\n> In [2]: ds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation')\r\n> Downloading readme: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 41.1k/41.1k [00:00<00:00, 596kB/s]\r\n> Downloading data: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 40.7M/40.7M [00:04<00:00, 8.50MB/s]\r\n> Generating validation split: 45576 examples [00:01, 44956.75 examples/s]\r\n> \r\n> In [3]: ds\r\n> Out[3]: \r\n> Dataset({\r\n> features: ['text', 'timestamp', 'url'],\r\n> num_rows: 45576\r\n> })\r\n> ```\r\nThank you for your reply,ExpectedMoreSplits was encountered in datasets version 2.12.2. After I updated the version, that is, datasets version 2.19.2, I encountered the FileNotFoundError problem mentioned above.", "That might be due to a corrupted cache.\r\n\r\nPlease, retry loading the dataset passing: `download_mode=\"force_redownload\"`\r\n```python\r\nds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation', download_mode=\"force_redownload\")\r\n```\r\n\r\nIt the above command does not fix the issue, then you will need to fix the cache manually, by removing the corresponding directory inside `~/.cache/huggingface/`.\r\n", "> That might be due to a corrupted cache.\r\n> \r\n> Please, retry loading the dataset passing: `download_mode=\"force_redownload\"`\r\n> \r\n> ```python\r\n> ds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation', download_mode=\"force_redownload\")\r\n> ```\r\n> \r\n> It the above command does not fix the issue, then you will need to fix the cache manually, by removing the corresponding directory inside `~/.cache/huggingface/`.\r\n\r\nThe two methods you mentioned above can not solve this problem, but the command line interface shows Downloading readme: 41.1kB [00:00, 281kB/s], and then FileNotFoundError appears. It is worth noting that I have no problem loading other datasets with the initial method, such as wikitext datasets", "> The two methods you mentioned above can not solve this problem, but the command line interface shows Downloading readme: 41.1kB [00:00, 281kB/s], and then FileNotFoundError appears.\r\n\r\nSame issue encountered.\r\n", "I really think the issue is caused by a corrupted cache, between versions 2.12.0 (there does not exist 2.12.2 version) and 2.19.2.\r\n\r\nAre you sure you removed all the corresponding corrupted directories within the cache?\r\n\r\nYou can easily check if the issue is caused by a corrupted cache by removing the entire cache:\r\n```shell\r\nmv ~/.cache/huggingface ~/.cache/huggingface.bak\r\n```\r\nand then reloading the dataset:\r\n```python\r\nds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation', download_mode=\"force_redownload\")\r\n```", "@albertvillanova Thanks for the reply. I tried removing the entire cache and reloading the dataset as you suggest. However, the same issue still exists. \r\n\r\nAs a test, I switch to a new platform, which (is a Windows system and) hasn't downloaded huggingface dataset before, and the dataset is loaded successfully. So I think \"a corrupted cache\" explanation makes sense. I wonder, besides `~/.cache/huggingface`, is there any other directory that may save the cache thing?\r\n\r\nAs a side note, I am using `datasets==2.20.0` and proxy `export HF_ENDPOINT=https://hf-mirror.com`.", "Ho @ZhangGe6,\r\n\r\nAs far as I know, that directory is the only one where the cache is saved, unless you configured another one. You can check it:\r\n```python\r\nimport datasets.config\r\n\r\nprint(datasets.config.HF_CACHE_HOME)\r\n# ~/.cache/huggingface\r\n\r\nprint(datasets.config.HF_DATASETS_CACHE)\r\n# ~/.cache/huggingface/datasets\r\n\r\nprint(datasets.config.HF_MODULES_CACHE)\r\n# ~/.cache/huggingface/modules\r\n\r\nprint(datasets.config.DOWNLOADED_DATASETS_PATH)\r\n# ~/.cache/huggingface/datasets/downloads\r\n\r\nprint(datasets.config.EXTRACTED_DATASETS_PATH)\r\n# ~/.cache/huggingface/datasets/downloads/extracted\r\n```\r\n\r\nAdditionally, `datasets` uses `huggingface_hub`, but its cache directory should also be inside `~/.cache/huggingface`, unless you configured another one. You can check it:\r\n```python\r\nimport huggingface_hub.constants\r\n\r\nprint(huggingface_hub.constants.HF_HOME)\r\n# ~/.cache/huggingface\r\n\r\nprint(huggingface_hub.constants.HF_HUB_CACHE)\r\n# ~/.cache/huggingface/hub\r\n```", "@albertvillanova I checked the directories you listed, and find that they are the same as the ones you provided. I am going to find more clues and will update what I find here.", "I've had a similar problem, and for some reason decreasing the number of workers in the dataloader solved it", "Same issue.\r\n", "Hi folks. Finally, I find it is a network issue that causes huggingface hub unreachable (in China).\r\n\r\nTo run the following script \r\n```python\r\nfrom datasets import load_dataset\r\n\r\nds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation', download_mode=\"force_redownload\")\r\n```\r\nWithout setting `export HF_ENDPOINT=https://hf-mirror.com`, I get the following error log\r\n```bash\r\nTraceback (most recent call last):\r\n File \".\\demo.py\", line 8, in <module>\r\n ds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation', download_mode=\"force_redownload\")\r\n File \"D:\\SoftwareInstall\\Python\\lib\\site-packages\\datasets\\load.py\", line 2594, in load_dataset\r\n builder_instance = load_dataset_builder(\r\n File \"D:\\SoftwareInstall\\Python\\lib\\site-packages\\datasets\\load.py\", line 2266, in load_dataset_builder\r\n dataset_module = dataset_module_factory(\r\n File \"D:\\SoftwareInstall\\Python\\lib\\site-packages\\datasets\\load.py\", line 1914, in dataset_module_factory\r\n raise e1 from None\r\n File \"D:\\SoftwareInstall\\Python\\lib\\site-packages\\datasets\\load.py\", line 1845, in dataset_module_factory\r\n raise ConnectionError(f\"Couldn't reach '{path}' on the Hub ({e.__class__.__name__})\") from e\r\nConnectionError: Couldn't reach 'allenai/c4' on the Hub (ConnectionError)\r\n```\r\nAfter setting `export HF_ENDPOINT=https://hf-mirror.com`, I get the following error, which is exactly the same as what we are debugging in this issue\r\n```bash\r\nDownloading readme: 41.1kB [00:00, 41.1MB/s]\r\nTraceback (most recent call last):\r\n File \".\\demo.py\", line 8, in <module>\r\n ds = load_dataset('allenai/c4', data_files={'validation': 'en/c4-validation.00003-of-00008.json.gz'}, split='validation', download_mode=\"force_redownload\")\r\n File \"D:\\SoftwareInstall\\Python\\lib\\site-packages\\datasets\\load.py\", line 2594, in loa builder_instance = load_dataset_builder(\r\n File \"D:\\SoftwareInstall\\Python\\lib\\site-packages\\datasets\\load.py\", line 2266, in load_dataset_builder\r\n dataset_module = dataset_module_factory(\r\n raise FileNotFoundError(\r\nFileNotFoundError: Couldn't find a dataset script at C:\\Users\\ZhangGe\\Desktop\\allenai\\c4\\c4.py or any data file in the same directory. Couldn't find 'allenai/c4' on the Hugging Face Hub either: FileNotFoundError: Unable to find 'hf://datasets/allenai/c4@1588ec454eed extension ['.csv', '.tsv', '.json', '.jsonl', '.parquet', '.geoparquet', '.gpq', '.arrow', '.txt', '.tar', '.blp', '.bmp', '.dib', '.bufr', '.cur', '.pcx', '.dcx', '.dds', '.ps', '.eps', '.fit', '.fits', '.fli', '.flc', '.ftc', '.ftu', '.gbr', '.gif', '.grib', \r\n'.h5', '.hdf', '.png', '.apng', '.jp2', '.j2k', '.jpc', '.jpf', '.jpx', '.j2c', '.icns',pm', '.BLP', '.BMP', '.DIB', '.BUFR', '.CUR', '.PCX', '.DCX', '.DDS', '.PS', '.EPS', '.FIT', '.FITS', '.FLI', '.FLC', '.FTC', '.FTU', '.GBR', '.GIF', '.GRIB', '.H5', '.HDF', '.PNG', '.APNG', '.JP2', '.J2K', '.JPC', '.JPF', '.JPX', '.J2C', '.ICNS', '.ICO', '.IM', '.IIM', '.TIF', '.TIFF', '.JFIF', '.JPE', '.JPG', '.JPEG', '.MPG', '.MPEG', '.MSP', '.PCD', '.PXR', '.PBM', '.PGM', '.PPM', '.PNM', '.PSD', '.BW', '.RGB', '.RGBA', '.SGI', '.RAS', '.TGA', '.ICB', '.VDA', '.VST', '.WEBP', '.WMF', '.EMF', '.XBM', '.XPM', '.aiff', '.au', '.avr', '.caf', '.flac', '.htk', '.svx', '.mat4', '.mat5', '.mpc2k', '.ogg', '.paf', '.pvf', '.raw', '.rf64', '.sd2', '.sds', '.ircam', '.voc', '.w64', '.wav', '.nist', '.wavex', '.wve', '.xi', '.mp3', '.opus', '.AIFF', '.AU', '.AVR', '.CAF', '.FLAC', '.HTK', \r\n'.SVX', '.MAT4', '.MAT5', '.MPC2K', '.OGG', '.PAF', '.PVF', '.RAW', '.RF64', '.SD2', '.SDS', '.IRCAM', '.VOC', '.W64', '.WAV', '.NIST', '.WAVEX', '.WVE', '.XI', '.MP3', '.OPUS', '.zip']\r\n```\r\n\r\n**Using a proxy software that avoids the internet access restrictions imposed by China, I can download the dataset using the same script**\r\n```bash\r\nDownloading readme: 100%|███████████████████████████████████████████| 41.1k/41.1k [00:00<00:00, 312kB/s] \r\nDownloading data: 100%|████████████████████████████████████████████| 40.7M/40.7M [00:19<00:00, 2.07MB/s] \r\nGenerating validation split: 45576 examples [00:00, 54883.48 examples/s]\r\n```\r\nSo `allenai/c4` is still unreachable even after setting `export HF_ENDPOINT=https://hf-mirror.com`.", "I have created an issue to inform the maintainers of `hf-mirror`：https://github.com/padeoe/hf-mirror-site/issues/30", "Thanks for the investigation: so finally it is an issue with the specific endpoint you are using.\r\n\r\nYou properly opened an issue in their repo, so they can fix it.\r\n\r\nI am closing this issue here." ]

[]

[]

[]

[]

[]

[ "I got the same issue. Any updates so far for this issue?" ]

[]

[ "How do you solve it ?\r\n", "> How do you solve it ?\r\n\r\nPlease check your Python environment and dataset version. I have just resolved the issue, which was caused by a Python environment switching error\r\n" ]

[ "you're right, we're tackling this here: https://github.com/huggingface/dataset-viewer/issues/2757", "@severo : great !" ]

[]

[]

[]

[ "Thanks for reporting, @srehaag.\r\n\r\nWe are investigating this issue.", "I confirm there is a bug for data-based Hub datasets when the user passes `data_dir`, which was introduced by PR:\r\n- #6714" ]

[]

[]

[]

[ "@mariosasko, @lhoestq, @albertvillanova\r\nHello! Can anyone help? or can you guys suggest who can help with this?", "Hi ! Feel free to download the dataset and create a `Dataset` object with it.\r\n\r\nThen your'll be able to use `push_to_hub()` to upload the dataset to HF in Parquet format and make it streamable :)", "> Hi ! Feel free to download the dataset and create a `Dataset` object with it.\r\n> \r\n> Then your'll be able to use `push_to_hub()` to upload the dataset to HF in Parquet format and make it streamable :)\r\n\r\nThe dataset is several TB in total, which I do not have the resources to handle." ]

[ "I am not able to reproduce the error with datasets 2.19.1:\r\n```python\r\nIn [1]: from datasets import load_dataset; ds = load_dataset(\"stas/c4-en-10k\", streaming=True); item = next(iter(ds[\"train\"])); item\r\nOut[1]: {'text': 'Beginners BBQ Class Taking Place in Missoula!\\nDo you want to get better at making delicious BBQ? You will have the opportunity, put this on your calendar now. Thursday, September 22nd join World Class BBQ Champion, Tony Balay from Lonestar Smoke Rangers. He will be teaching a beginner level class for everyone who wants to get better with their culinary skills.\\nHe will teach you everything you need to know to compete in a KCBS BBQ competition, including techniques, recipes, timelines, meat selection and trimming, plus smoker and fire information.\\nThe cost to be in the class is $35 per person, and for spectators it is free. Included in the cost will be either a t-shirt or apron and you will be tasting samples of each meat that is prepared.'}\r\n\r\nIn [2]: from datasets import load_dataset; ds = load_dataset(\"stas/c4-en-10k\", download_mode=\"force_redownload\"); ds\r\nDownloading data: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 13.3M/13.3M [00:00<00:00, 18.7MB/s]\r\nGenerating train split: 100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 10000/10000 [00:00<00:00, 78548.55 examples/s]\r\nOut[2]: \r\nDatasetDict({\r\n train: Dataset({\r\n features: ['text'],\r\n num_rows: 10000\r\n })\r\n})\r\n```\r\n\r\nLooking at your error traceback, I notice that the code line numbers do not correspond to the ones of datasets 2.19.1.\r\n\r\nAdditionally, I can't reproduce the issue with `HfFileSystem`:\r\n```python\r\nIn [1]: from huggingface_hub import HfFileSystem\r\n\r\nIn [2]: fs = HfFileSystem()\r\n\r\nIn [3]: with fs.open(\"datasets/stas/c4-en-10k/c4-en-10k.py\", \"rb\") as f:\r\n ...: data = f.read()\r\n ...: \r\n\r\nIn [4]: data[:20]\r\nOut[4]: b'# coding=utf-8\\n# Cop'\r\n```\r\n\r\nCould you please verify the `datasets` and `huggingface_hub` versions you are indeed using?\r\n```python\r\nimport datasets; print(datasets.__version__)\r\n\r\nimport huggingface_hub; print(huggingface_hub.__version__)\r\n```", "Thanks for your reply! After I update the datasets version from 2.15.0 back to 2.19.1 again, it seems everything work well. Sorry for bordering you!" ]

[ "Update: I ended up deciding to go back to use lines of dictionaries instead of arrays, not because of this issue as my users would be capable of downloading my corpus without `datasets`, but the speed and storage savings are not currently worth breaking my API and harming the backwards compatibility of each new revision.\r\n\r\nWith that said, for a static dataset that is not regularly updated like mine, and particularly for extremely large datasets with millions or billions of rows, using arrays could have a meaningful impact, and so there is probably still value in supporting this structure, provided the effort is not too much." ]

[ "Thank you I will try this out!\r\n\r\nOn Tue, Jun 11, 2024 at 3:55 AM Vincent Lau ***@***.***>\r\nwrote:\r\n\r\n> I add a \"streaming=True\" after the name of the dataset, and it\r\n> works.....hope it can help you\r\n>\r\n> And if you install the version datasets==2.15.0, this bug will not happen.\r\n> I don't know why, but all of them works\r\n>\r\n> —\r\n> Reply to this email directly, view it on GitHub\r\n> <https://github.com/huggingface/datasets/issues/6906#issuecomment-2160041812>,\r\n> or unsubscribe\r\n> <https://github.com/notifications/unsubscribe-auth/A3HXU7AMBT2MNO34SC3Z5G3ZG2UOXAVCNFSM6AAAAABH45CNPWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRQGA2DCOBRGI>\r\n> .\r\n> You are receiving this because you authored the thread.Message ID:\r\n> ***@***.***>\r\n>\r\n", "I still find out that there are some strange bug in v2.15.0 of datasets. it seems like that the *.arrow file cannot be established. it may be an index of the subsets. well I still try to debug it. but, one of the most efficient way may be using the google colab to build this index in the ~/huggingface/datasets, and than download them to replace the local file.....lol......it works!", "Yeah I did try what you suggested and it didn’t work. I was able to get it\r\non a local from someone who access the dataset in the past. Let me know\r\nwhen you end up fixing this bug.\r\n\r\nOn Tue, Jun 11, 2024 at 10:33 PM Vincent Lau ***@***.***>\r\nwrote:\r\n\r\n> I still find out that there are some strange bug in v2.15.0 of datasets.\r\n> it seems like that the *.arrow file cannot be established. it may be an\r\n> index of the subsets. well I still try to debug it. but, one of the most\r\n> efficient way may be using the google colab to build this index in the\r\n> ~/huggingface/datasets, and than download them to replace the local\r\n> file.....lol......it works!\r\n>\r\n> —\r\n> Reply to this email directly, view it on GitHub\r\n> <https://github.com/huggingface/datasets/issues/6906#issuecomment-2161988798>,\r\n> or unsubscribe\r\n> <https://github.com/notifications/unsubscribe-auth/A3HXU7BCJE2LOCWRVWPMNODZG6XPJAVCNFSM6AAAAABH45CNPWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDCNRRHE4DQNZZHA>\r\n> .\r\n> You are receiving this because you authored the thread.Message ID:\r\n> ***@***.***>\r\n>\r\n", "Could you please provide more information, as required by the Bug template: https://github.com/huggingface/datasets/issues/new?assignees=&labels=&projects=&template=bug-report.yml\r\n\r\nWithout all that information, it is very difficult for us to understand the underlying issue and to give a pertinent answer.\r\n\r\nWhat are the versions of the libraries you are using? Datasets, pyarrow, fsspec,...\r\n> Environment info\r\n> Please share your environemnt info with us. You can run the command datasets-cli env and copy-paste its output below.\r\n\r\nWhat is the output you get after executing these code lines?\r\n```python\r\nimport datasets\r\nds = datasets.load_dataset('irc_disentangle')\r\nds\r\n```\r\n\r\n", "We have made the following fixes:\r\n- [Fix source data URL](https://huggingface.co/datasets/jkkummerfeld/irc_disentangle/discussions/4)\r\n- [Convert dataset to Parquet](https://huggingface.co/datasets/jkkummerfeld/irc_disentangle/discussions/5)", "Thank you for the fixes. Sorry I lost this conversation in my inbox.\r\n\r\nOn Mon, Jul 8, 2024 at 2:18 AM Albert Villanova del Moral <\r\n***@***.***> wrote:\r\n\r\n> Closed #6906 <https://github.com/huggingface/datasets/issues/6906> as\r\n> completed.\r\n>\r\n> —\r\n> Reply to this email directly, view it on GitHub\r\n> <https://github.com/huggingface/datasets/issues/6906#event-13418330895>,\r\n> or unsubscribe\r\n> <https://github.com/notifications/unsubscribe-auth/A3HXU7HREJDE5BZSOEJFJI3ZLIVLNAVCNFSM6AAAAABH45CNPWVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJTGQYTQMZTGA4DSNI>\r\n> .\r\n> You are receiving this because you authored the thread.Message ID:\r\n> ***@***.***>\r\n>\r\n" ]

[]

[ "I think [`Dataset.to_parquet`](https://huggingface.co/docs/datasets/v1.10.2/package_reference/main_classes.html#datasets.Dataset.to_parquet) is what you're looking for.\r\n\r\nLet me know if I'm wrong ", "No, it does not save the metadata json.\r\n\r\nWe have to recode all meta json load/save\r\nwith another custome functions.\r\n\r\nsave_to_disk\r\nand load should have option with\r\n“Parquet” instead of “arrow”\r\n\r\nsince “arrow” is never user for production \r\n(only parquet).\r\n\r\nThanks !\r\n\r\n> On May 17, 2024, at 5:38, Frédéric Branchaud-Charron ***@***.***> wrote:\r\n> \r\n> \r\n> I think Dataset.to_parquet is what you're looking for.\r\n> \r\n> Let me know if I'm wrong\r\n> \r\n> —\r\n> Reply to this email directly, view it on GitHub, or unsubscribe.\r\n> You are receiving this because you authored the thread.\r\n", "You can use `to_parquet` and `ds.info.write_to_directory()` to save the dataset info", "Ok,\r\n\r\nWhat about loading ?\r\n\r\nShould we do in 2 steps ?\r\n\r\n\r\n\r\n> On Jun 14, 2024, at 1:09, Quentin Lhoest ***@***.***> wrote:\r\n> \r\n> \r\n> You can use to_parquet and ds.info.write_to_directory() to save the dataset info\r\n> \r\n> —\r\n> Reply to this email directly, view it on GitHub, or unsubscribe.\r\n> You are receiving this because you authored the thread.\r\n", "Yes, and there is DatasetInfo.from_directory(). to reload the info", "Isn’t easier to combine both\r\ninto load_dataset and save_dataset\r\nwith parquet options.\r\n\r\n2) another question,\r\nHow can we download large dataset into disk directly without loading all in memory (!)\r\n\r\n\r\n\r\n\r\n> On Jun 14, 2024, at 19:54, Quentin Lhoest ***@***.***> wrote:\r\n> \r\n> \r\n> Yes, and there is DatasetInfo.from_directory(). to reload the info\r\n> \r\n> —\r\n> Reply to this email directly, view it on GitHub, or unsubscribe.\r\n> You are receiving this because you authored the thread.\r\n", "`load_dataset` doesn't load the dataset in memory, it progressively writes to disk in Arrow format and then memory maps the Arrow files. This allows to load datasets bigger than memory and without filling your RAM", "Sure.\r\nHow memory map is managed ?\r\nManaged by the OS ?\r\n\r\nWhy the need of save_dataset() ?\r\n\r\n\r\n\r\n> On Jun 15, 2024, at 0:06, Quentin Lhoest ***@***.***> wrote:\r\n> \r\n> \r\n> load_dataset doesn't load the dataset in memory, it progressively writes to disk in Arrow format and then memory maps the Arrow files. This allows to load datasets bigger than memory and without filling your RAM\r\n> \r\n> —\r\n> Reply to this email directly, view it on GitHub, or unsubscribe.\r\n> You are receiving this because you authored the thread.\r\n" ]

[]

[ "@lhoestq How difficult of fix is this?", "It shouldn't be difficult, I think it's just a matter of adding the missing fields from `self.config.features` in `example` here: before it iterates on image_field_names and audio_field_names. A missing field should have a value set to None\r\n\r\nhttps://github.com/huggingface/datasets/blob/768cb35ede5a6c35fa7545aa3671f3e321c96440/src/datasets/packaged_modules/webdataset/webdataset.py#L113-L116", "@lhoestq So like this then?\r\n\r\n``` \r\ndef _generate_examples(self, tar_paths, tar_iterators):\r\n image_field_names = [\r\n field_name for field_name, feature in self.info.features.items() if isinstance(feature, datasets.Image)\r\n ]\r\n audio_field_names = [\r\n field_name for field_name, feature in self.info.features.items() if isinstance(feature, datasets.Audio)\r\n ]\r\n\t\r\n all_field_names = list(self.config.features.keys())\r\n \r\n for tar_idx, (tar_path, tar_iterator) in enumerate(zip(tar_paths, tar_iterators)):\r\n for example_idx, example in enumerate(self._get_pipeline_from_tar(tar_path, tar_iterator)):\r\n for field_name in all_field_names:\r\n if field_name not in example:\r\n if field_name in self.config.features:\r\n example[field_name] = self.config.features[field_name]\r\n else:\r\n example[field_name] = None\r\n \r\n # Process image and audio fields\r\n for field_name in image_field_names + audio_field_names:\r\n if example[field_name] is not None:\r\n example[field_name] = {\"path\": example[\"__key__\"] + \".\" + field_name, \"bytes\": example[field_name]}\r\n \r\n yield f\"{tar_idx}_{example_idx}\", example\r\n```\r\n\r\nOr should we avoid trying add the missing values and just set them to None?\r\n\r\n```\r\n for field_name in all_field_names:\r\n if field_name not in example:\r\n example[field_name] = None\r\n```", "Yup this is the solution !\r\n\r\n```python\r\n for field_name in all_field_names:\r\n if field_name not in example:\r\n example[field_name] = None\r\n```", "@lhoestq Awesome, thanks! I made a PR with the fixes" ]

[]

[ "Hello, @bghira.\r\n\r\nThanks for reporting. Please note that the text originating the error is not supposed to be valid YAML: it contains the instructions to generate the actual YAML content, that should replace the instructions comment.\r\n\r\nOn the other hand, I agree that it is not nice to have that YAML error message at the top of the page: \r\n\r\n\r\nI am proposing a change to make the YAML error disappear.", "thanks albert! i looked at it for a while to figure it out. i think the `raw` view option is the correct way to look at it?" ]

[]

[]

[ "Hi @DarshanDeshpande,\r\n\r\nPlease note that the default format of the method `Dataset.to_json` is [JSON-Lines](https://jsonlines.org/): it passes `orient=\"records\", lines=True` to `pandas.DataFrame.to_json`. This format is specially useful for large datasets, since unlike regular JSON files, it does not require loading all the data into memory at once, but can be done iteratively by batches.\r\n\r\nIn order to read this file using the `json` library, you should parse line by line:\r\n```python\r\nwith open(\"full_dataset.json\", \"r\") as f:\r\n data = [json.loads(line) for line in f]\r\nlen(data)\r\n```\r\nMaybe we should explain this better in our docs.", "Now we explain this better in out docs:\r\n- #6895" ]

[]

[ "Hello,\r\n\r\nI'm not sure I understand. \r\nThe return value of `ds.load_faiss_index` is None as expected.\r\n\r\nI see that loading an Index on a dataset that doesn't have an `embedding` column doesn't raise an Issue. Is that the issue?\r\n\r\nSo `ds` doesn't have an `embedding` column, but we load an index that looks for it. But this will raise an issue only when calling `ds.search`." ]

[]

[]

[ "Changing the supplier of the proxy will solve this problem, or you can visit and follow the instructions in https://hf-mirror.com " ]

[ "@albertvillanova @lhoestq just ran into it and requiring newer pillow isn't a solution as it breaks Pillow-SIMD which is behind Pillow quite a few versions but necessary for training with reasonable throughput. \r\n\r\nA couple things here... \r\n\r\n1. This can be done with a method that isn't an issue for any somewhat recent Pillow\r\n`image = ImageOps.exif_transpose(image)`\r\n\r\n2. I'd rather this not be done for me automatically. Sometimes exif data is correct, sometimes it's not. Sometimes I might want to correct the orientation, sometimes I might not. \r\n\r\nIn any case if I've preprocessed the images properly myself I don't want to incur overhead, possible further fp seeks, parsing, to load the exif that's not loaded and parsed when you just open and decode the image.", "Hi @rwightman, thanks for your feedback.\r\n\r\nFirst, as a side note comment, please note that you are depending on Pillow-SIMD and that library seems no longer maintained:\r\n- it has not been updated for more than a year: last commit to main was on June 20, 2023: https://github.com/uploadcare/pillow-simd/commit/faae977a00472275690664fe27e21df4e4e8ce07\r\n- in PyPI, the last release was more than 2 years ago, on January 4, 2022: https://pypi.org/project/Pillow-SIMD/#history\r\n\r\nIn relation with your suggestions for the `datasets` library, the changes were introduced by this PR:\r\n- #6739\r\n\r\nI agree maybe we should have given the option whether to perform this operation or not.", "@albertvillanova \r\n\r\nHuh, thought I'd just installed the current datasets when I ran into this, maybe it was behind...\r\n\r\nI'm aware the support for SIMD is a problem, but it's up to 8x faster than non SIMD Pillow and really necessary in many training situations or you have lots of idle GPUs. The current situation is unfortunate but most changes since 9.0 aren't all that important for 'decoding jpegs and resizing'" ]

[ "The error is caused by malformed basenames of the files within the TARs:\r\n- `15_Cohen_1-s2.0-S0929664620300449-gr3_lrg-b.png` becomes `15_Cohen_1-s2` as the grouping `__key__`, and `0-S0929664620300449-gr3_lrg-b.png` as the additional key to be added to the example\r\n- whereas the intended behavior was to use `15_Cohen_1-s2.0-S0929664620300449-gr3_lrg-b` as the grouping `__key__`, and `png` as the additional key to be added to the example\r\n\r\nTo get the expected behavior, the basenames of the files within the TARs should be fixed so that they only contain a single dot, the one separating the file extension.", "I reopen it because I think we should try to give a clearer error message with a specific error code.\r\n\r\nFor now, it's hard for the user to understand where the error comes from (not everybody knows the subtleties of the webdataset filename structure).\r\n\r\n(we can transfer it to https://github.com/huggingface/dataset-viewer if it fits better there)", "same with .jpg -> https://huggingface.co/datasets/ProGamerGov/synthetic-dataset-1m-dalle3-high-quality-captions\r\n\r\n```\r\nError code: DatasetGenerationError\r\nException: DatasetGenerationError\r\nMessage: An error occurred while generating the dataset\r\nTraceback: Traceback (most recent call last):\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py\", line 1748, in _prepare_split_single\r\n for key, record in generator:\r\n File \"/src/services/worker/src/worker/job_runners/config/parquet_and_info.py\", line 818, in wrapped\r\n for item in generator(*args, **kwargs):\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/packaged_modules/webdataset/webdataset.py\", line 109, in _generate_examples\r\n example[field_name] = {\"path\": example[\"__key__\"] + \".\" + field_name, \"bytes\": example[field_name]}\r\n KeyError: 'jpg'\r\n \r\n The above exception was the direct cause of the following exception:\r\n \r\n Traceback (most recent call last):\r\n File \"/src/services/worker/src/worker/job_runners/config/parquet_and_info.py\", line 1316, in compute_config_parquet_and_info_response\r\n parquet_operations, partial = stream_convert_to_parquet(\r\n File \"/src/services/worker/src/worker/job_runners/config/parquet_and_info.py\", line 909, in stream_convert_to_parquet\r\n builder._prepare_split(\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py\", line 1627, in _prepare_split\r\n for job_id, done, content in self._prepare_split_single(\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/builder.py\", line 1784, in _prepare_split_single\r\n raise DatasetGenerationError(\"An error occurred while generating the dataset\") from e\r\n datasets.exceptions.DatasetGenerationError: An error occurred while generating the dataset\r\n```\r\n", "More details in the spec (https://docs.google.com/document/d/18OdLjruFNX74ILmgrdiCI9J1fQZuhzzRBCHV9URWto0/edit#heading=h.hkptaq2kct2s)\r\n\r\n> The prefix of a file is all directory components of the file plus the file name component up to the first “.” in the file name.\r\n> The last extension (i.e., the portion after the last “.”) in a file name determines the file type.\r\n\r\n> Example:\r\n\timages17/image194.left.jpg\r\n\timages17/image194.right.jpg\r\n\timages17/image194.json\r\n\timages17/image12.left.jpg\r\n\timages17/image12.json\r\n\timages17/image12.right.jpg\r\n\timages3/image1459.left.jpg\r\n> \t…\r\n> When reading this with a WebDataset library, you would get the following two dictionaries back in sequence:\r\n\r\n { “__key__”: “images17/image194”, “left.jpg”: b”...”, “right.jpg”: b”...”, “json”: b”...”}\r\n { “__key__”: “images17/image12”, “left.jpg”: b”...”, “right.jpg”: b”...”, “json”: b”...”}\r\n", "OK, the issue is different in the latter case: some files are suffixed as `.jpeg`, and others as `.jpg` :)\r\n\r\nIs it a limitation of the webdataset format, or of the datasets library @lhoestq? And could we be able to give a clearer error?" ]

[]

[ "ulimit -n 8192 can solve this problem", "> ulimit -n 8192 can solve this problem\r\n\r\nWould there be a systematic way to do this ? The data loading is part of the [MTEB](https://github.com/embeddings-benchmark/mteb) library", "> > ulimit -n 8192 can solve this problem\r\n> \r\n> Would there be a systematic way to do this ? The data loading is part of the [MTEB](https://github.com/embeddings-benchmark/mteb) library\r\n\r\n I think we could modify the _prepare_split_single function", "I fixed it with https://github.com/huggingface/datasets/pull/6893, feel free to re-open if you're still having the issue :)", "> I fixed it with #6893, feel free to re-open if you're still having the issue :)\r\n\r\nThanks a lot!" ]

[]

[ "I guess the issue is caused by the customization of BuilderConfig that you use from the repo [https://github.com/BeyonderXX/InstructUIE](https://github.com/BeyonderXX/InstructUIE/blob/master/src/uie_dataset.py). You should report to them.\r\n\r\nI see you already opened an issue in their repo:\r\n- https://github.com/BeyonderXX/InstructUIE/issues/40" ]

[ "Thanks! Feel free to ping me for examples. May not respond immediately because we're all busy but would like to help.", "Hi @natolambert, could you please give some examples of JSON files to benchmark?\r\n\r\nPlease note that this JSON file (https://huggingface.co/datasets/allenai/reward-bench-results/blob/main/eval-set-scores/Ray2333/reward-model-Mistral-7B-instruct-Unified-Feedback.json) is not in \"records\" orient; instead it has the following structure:\r\n```json\r\n{\r\n \"chat_template\": \"tulu\",\r\n \"id\": [30, 34, 35,...],\r\n \"model\": \"Ray2333/reward-model-Mistral-7B-instruct-Unified-Feedback\",\r\n \"model_type\": \"Seq. Classifier\",\r\n \"results\": [1, 1, 1, ...],\r\n \"scores_chosen\": [4.421875, 1.8916015625, 3.8515625,...],\r\n \"scores_rejected\": [-2.416015625, -1.47265625, -0.9912109375,...],\r\n \"subset\": [\"alpacaeval-easy\", \"alpacaeval-easy\", \"alpacaeval-easy\",...]\r\n \"text_chosen\": [\"<s>[INST] How do I detail a...\",...],\r\n \"text_rejected\": [\"<s>[INST] How do I detail a...\",...]\r\n}\r\n```\r\n\r\nNote that \"records\" orient should be a list (not a dict) with each row as one item of the list:\r\n```json\r\n[\r\n {\"chat_template\": \"tulu\", \"id\": 30,... },\r\n {\"chat_template\": \"tulu\", \"id\": 34,... },\r\n ...\r\n]\r\n```", "We use a mix (which is a mess), here's an example with the records orient\r\nhttps://huggingface.co/datasets/allenai/reward-bench-results/blob/main/best-of-n/alpaca_eval/tulu-13b/OpenAssistant/oasst-rm-2.1-pythia-1.4b-epoch-2.5.json\r\n\r\nThere are more in that folder, ~40mb maybe?", "@albertvillanova here's a snippet so you don't need to click\r\n```\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 0\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 3.076171875\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 1\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 3.87890625\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 2\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 3.287109375\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 3\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 1.6337890625\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 4\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 5.27734375\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 5\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 3.0625\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 6\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 2.29296875\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 7\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 6.77734375\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 8\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 3.853515625\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 9\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 4.86328125\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 10\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 2.890625\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 11\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 4.70703125\r\n}\r\n{\r\n \"config\": \"top_p=0.9;temp=1.0\",\r\n \"dataset_details\": \"helpful_base\",\r\n \"id\": [\r\n 0,\r\n 12\r\n ],\r\n \"model\": \"allenai/tulu-2-dpo-13b\",\r\n \"scores\": 4.45703125\r\n}\r\n```", "Thanks again for your feedback, @natolambert.\r\n\r\nHowever, strictly speaking, the last file is not in JSON format but in kind of JSON-Lines like format (although not properly either because there are multiple newline characters within each object). Not even pandas can read that file format.\r\n\r\nAnyway, for JSON-Lines, I would expect that `datasets` and `pandas` have the same performance for JSON Lines files, as both use `pyarrow` under the hood...\r\n\r\nA proper JSON file in records orient should be a list (a JSON array): the first character should be `[`.\r\n\r\nAnyway, I am generating a JSON file from your JSON-Lines file to test performance." ]

[ "Potentially related:\r\n* #6864\r\n* #6850\r\n* #6848\r\n* #6819", "Hi @jerome-white, thnaks for reporting.\r\n\r\nHowever, I cannot reproduce your issue:\r\n```python\r\n>>> from datasets import get_dataset_config_names\r\n\r\n>>> get_dataset_config_names(\"open-llm-leaderboard/details_davidkim205__Rhea-72b-v0.5\")\r\n['harness_arc_challenge_25',\r\n 'harness_gsm8k_5',\r\n 'harness_hellaswag_10',\r\n 'harness_hendrycksTest_5',\r\n 'harness_hendrycksTest_abstract_algebra_5',\r\n 'harness_hendrycksTest_anatomy_5',\r\n 'harness_hendrycksTest_astronomy_5',\r\n 'harness_hendrycksTest_business_ethics_5',\r\n 'harness_hendrycksTest_clinical_knowledge_5',\r\n 'harness_hendrycksTest_college_biology_5',\r\n 'harness_hendrycksTest_college_chemistry_5',\r\n 'harness_hendrycksTest_college_computer_science_5',\r\n 'harness_hendrycksTest_college_mathematics_5',\r\n 'harness_hendrycksTest_college_medicine_5',\r\n 'harness_hendrycksTest_college_physics_5',\r\n 'harness_hendrycksTest_computer_security_5',\r\n 'harness_hendrycksTest_conceptual_physics_5',\r\n 'harness_hendrycksTest_econometrics_5',\r\n 'harness_hendrycksTest_electrical_engineering_5',\r\n 'harness_hendrycksTest_elementary_mathematics_5',\r\n 'harness_hendrycksTest_formal_logic_5',\r\n 'harness_hendrycksTest_global_facts_5',\r\n 'harness_hendrycksTest_high_school_biology_5',\r\n 'harness_hendrycksTest_high_school_chemistry_5',\r\n 'harness_hendrycksTest_high_school_computer_science_5',\r\n 'harness_hendrycksTest_high_school_european_history_5',\r\n 'harness_hendrycksTest_high_school_geography_5',\r\n 'harness_hendrycksTest_high_school_government_and_politics_5',\r\n 'harness_hendrycksTest_high_school_macroeconomics_5',\r\n 'harness_hendrycksTest_high_school_mathematics_5',\r\n 'harness_hendrycksTest_high_school_microeconomics_5',\r\n 'harness_hendrycksTest_high_school_physics_5',\r\n 'harness_hendrycksTest_high_school_psychology_5',\r\n 'harness_hendrycksTest_high_school_statistics_5',\r\n 'harness_hendrycksTest_high_school_us_history_5',\r\n 'harness_hendrycksTest_high_school_world_history_5',\r\n 'harness_hendrycksTest_human_aging_5',\r\n 'harness_hendrycksTest_human_sexuality_5',\r\n 'harness_hendrycksTest_international_law_5',\r\n 'harness_hendrycksTest_jurisprudence_5',\r\n 'harness_hendrycksTest_logical_fallacies_5',\r\n 'harness_hendrycksTest_machine_learning_5',\r\n 'harness_hendrycksTest_management_5',\r\n 'harness_hendrycksTest_marketing_5',\r\n 'harness_hendrycksTest_medical_genetics_5',\r\n 'harness_hendrycksTest_miscellaneous_5',\r\n 'harness_hendrycksTest_moral_disputes_5',\r\n 'harness_hendrycksTest_moral_scenarios_5',\r\n 'harness_hendrycksTest_nutrition_5',\r\n 'harness_hendrycksTest_philosophy_5',\r\n 'harness_hendrycksTest_prehistory_5',\r\n 'harness_hendrycksTest_professional_accounting_5',\r\n 'harness_hendrycksTest_professional_law_5',\r\n 'harness_hendrycksTest_professional_medicine_5',\r\n 'harness_hendrycksTest_professional_psychology_5',\r\n 'harness_hendrycksTest_public_relations_5',\r\n 'harness_hendrycksTest_security_studies_5',\r\n 'harness_hendrycksTest_sociology_5',\r\n 'harness_hendrycksTest_us_foreign_policy_5',\r\n 'harness_hendrycksTest_virology_5',\r\n 'harness_hendrycksTest_world_religions_5',\r\n 'harness_truthfulqa_mc_0',\r\n 'harness_winogrande_5',\r\n 'results']\r\n```\r\n\r\nMaybe it was just a temporary issue...", "> Maybe it was just a temporary issue...\r\n\r\nPerhaps. I've changed my workflow to use the hub's `HfFileSystem`, so for now this is no longer a blocker for me. I'll reopen the issue if that changes." ]

[]

[ "Hi @vinodrajendran001, thanks for reporting.\r\n\r\nIndeed the dataset no longer exists on the Hub. The URL https://huggingface.co/datasets/rewardsignal/reddit_writing_prompts gives 404 Not Found error." ]

[]

[ "I think this needs to be fixed on transformers.\r\n\r\nCC: @Wauplin ", "See:\r\n- https://github.com/huggingface/transformers/issues/30618", "Opened https://github.com/huggingface/transformers/pull/30620" ]

[ "I downloaded the jsonl file and extract it manually. \r\nThe issue seems to be related to pyarrow.json \r\n\r\n\r\n\r\npython3 -q -X faulthandler -c \"from datasets import load_dataset; load_dataset('json', data_files='/Users/scampion/Downloads/1998-09.jsonl')\"\r\nGenerating train split: 0 examples [00:00, ? examples/s]Fatal Python error: Segmentation fault\r\n\r\nThread 0x00007000000c1000 (most recent call first):\r\n <no Python frame>\r\n\r\nThread 0x00007000024df000 (most recent call first):\r\n File \"/usr/local/Cellar/[email protected]/3.11.7/Frameworks/Python.framework/Versions/3.11/lib/python3.11/threading.py\", line 331 in wait\r\n File \"/usr/local/Cellar/[email protected]/3.11.7/Frameworks/Python.framework/Versions/3.11/lib/python3.11/threading.py\", line 629 in wait\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/tqdm/_monitor.py\", line 60 in run\r\n File \"/usr/local/Cellar/[email protected]/3.11.7/Frameworks/Python.framework/Versions/3.11/lib/python3.11/threading.py\", line 1045 in _bootstrap_inner\r\n File \"/usr/local/Cellar/[email protected]/3.11.7/Frameworks/Python.framework/Versions/3.11/lib/python3.11/threading.py\", line 1002 in _bootstrap\r\n\r\nThread 0x00007ff845c66640 (most recent call first):\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/datasets/packaged_modules/json/json.py\", line 122 in _generate_tables\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/datasets/builder.py\", line 1995 in _prepare_split_single\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/datasets/builder.py\", line 1882 in _prepare_split\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/datasets/builder.py\", line 1122 in _download_and_prepare\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/datasets/builder.py\", line 1027 in download_and_prepare\r\n File \"/Users/scampion/src/test/venv_test/lib/python3.11/site-packages/datasets/load.py\", line 2609 in load_dataset\r\n File \"<string>\", line 1 in <module>\r\n\r\nExtension modules: numpy.core._multiarray_umath, numpy.core._multiarray_tests, numpy.linalg._umath_linalg, numpy.fft._pocketfft_internal, numpy.random._common, numpy.random.bit_generator, numpy.random._bounded_integers, numpy.random._mt19937, numpy.random.mtrand, numpy.random._philox, numpy.random._pcg64, numpy.random._sfc64, numpy.random._generator, pyarrow.lib, pyarrow._hdfsio, pandas._libs.tslibs.ccalendar, pandas._libs.tslibs.np_datetime, pandas._libs.tslibs.dtypes, pandas._libs.tslibs.base, pandas._libs.tslibs.nattype, pandas._libs.tslibs.timezones, pandas._libs.tslibs.fields, pandas._libs.tslibs.timedeltas, pandas._libs.tslibs.tzconversion, pandas._libs.tslibs.timestamps, pandas._libs.properties, pandas._libs.tslibs.offsets, pandas._libs.tslibs.strptime, pandas._libs.tslibs.parsing, pandas._libs.tslibs.conversion, pandas._libs.tslibs.period, pandas._libs.tslibs.vectorized, pandas._libs.ops_dispatch, pandas._libs.missing, pandas._libs.hashtable, pandas._libs.algos, pandas._libs.interval, pandas._libs.lib, pyarrow._compute, pandas._libs.ops, pandas._libs.hashing, pandas._libs.arrays, pandas._libs.tslib, pandas._libs.sparse, pandas._libs.internals, pandas._libs.indexing, pandas._libs.index, pandas._libs.writers, pandas._libs.join, pandas._libs.window.aggregations, pandas._libs.window.indexers, pandas._libs.reshape, pandas._libs.groupby, pandas._libs.json, pandas._libs.parsers, pandas._libs.testing, charset_normalizer.md, yaml._yaml, pyarrow._parquet, pyarrow._fs, pyarrow._hdfs, pyarrow._gcsfs, pyarrow._s3fs, multidict._multidict, yarl._quoting_c, aiohttp._helpers, aiohttp._http_writer, aiohttp._http_parser, aiohttp._websocket, frozenlist._frozenlist, xxhash._xxhash, pyarrow._json (total: 72)\r\n[1] 56678 segmentation fault python3 -q -X faulthandler -c\r\n/usr/local/Cellar/[email protected]/3.11.7/Frameworks/Python.framework/Versions/3.11/lib/python3.11/multiprocessing/resource_tracker.py:254: UserWarning: resource_tracker: There appear to be 1 leaked semaphore objects to clean up at shutdown\r\n warnings.warn('resource_tracker: There appear to be %d '\r\n(venv_test)", "The error comes from data where one line contains \"null\"" ]

[ "After investigation, I have found that when a local file is uploaded to the Hub, the new line character is no longer transformed to \"\\n\": on Windows machine now it is kept as \"\\r\\n\".\r\n\r\nAny idea why this changed?\r\nCC: @lhoestq " ]

[]

[]

[]

[]

[ "Version 2.18 works without problem.", "@Namangarg110 @mohsen-goodarzi The bug appears because the number of urls is less than 16 and the algorithm is meant to work on the previously created mode for a single url as stated on line 314: https://github.com/huggingface/datasets/blob/1bf8a46cc7b096d5c547ea3794f6a4b6c31ea762/src/datasets/download/download_manager.py#L314\r\n\r\nIn addition, previously `map_nested` function was supported without batching and it is meant to be the default performance. \r\n\r\nOne of the shortest walk-arounds would be changing the part of the manager with the current setting:\r\n```\r\n if len(url_or_urls) >= 16:\r\n download_func = partial(self._download_batched, download_config=download_config)\r\n else:\r\n download_func = partial(self._download_single, download_config=download_config)\r\n\r\n start_time = datetime.now()\r\n with stack_multiprocessing_download_progress_bars():\r\n downloaded_path_or_paths = map_nested(\r\n download_func,\r\n url_or_urls,\r\n map_tuple=True,\r\n num_proc=download_config.num_proc,\r\n desc=\"Downloading data files\",\r\n batched=True if len(url_or_urls) >= 16 else False,\r\n batch_size=-1,\r\n )\r\n```\r\n\r\nI would suggest to consider other datasets for similar issues and make a pull-request. ", "Thanks for reporting @Namangarg110 and thanks for the investigation @MilanaShhanukova.\r\n\r\nApparently, there is an issue with the download functionality.\r\nI am proposing a fix." ]

[ "Same issue here." ]

[ "This should help fixing this issue: https://github.com/huggingface/datasets/pull/6832", "I'm having a similar issue when using splices:\r\n<img width=\"947\" alt=\"image\" src=\"https://github.com/huggingface/datasets/assets/28941213/2153faac-e1fe-4b6d-a79b-30b2699407e8\">\r\n<img width=\"823\" alt=\"image\" src=\"https://github.com/huggingface/datasets/assets/28941213/80919eca-eb6c-407d-8070-52642fdcee54\">\r\n<img width=\"914\" alt=\"image\" src=\"https://github.com/huggingface/datasets/assets/28941213/5219c201-e22e-4536-acc3-a922677785ff\">\r\n\r\n\r\nIt seems to be downloading, loading, and generating splits using the entire dataset." ]

[ "In every iteration you load the full \"random_input\" column in memory, only then to access it's i-th element.\r\n\r\nYou can try using this instead\r\n\r\na,b=dataset[i]['random_input'],dataset[i]['random_output']" ]

[ "I encountered this same issue when loading a customized dataset for ORPO training, in which there were three columns and two of them were lists. \r\nI debugged and found that it might be caused by the type-infer mechanism and because in some chunks one of the columns is always an empty list ([]), it was regarded as ```list<item: null>```, however in some other chunk it was ```list<item: string>```. This triggered a TypeError running the function ```table_cast()```.\r\n\r\nI temporarily fixed this by re-dumping the file into a regular JSON format instead of lines of JSON dict. I didn't dig deeper for the lack of knowledge and programming ability but I do hope some developer of this repo will find and fix it." ]

[ "Thanks for reporting! I've opened a PR with a fix.", "Thanks, @mariosasko! Related question (although I guess this is a feature request): could we have some kind of exponential back-off for these retries? Here's my reasoning:\r\n- If a one-time accidental error happens, you should retry immediately and will succeed immediately.\r\n- If the Hub has a small outage on the order of minutes, you don't want to retry on the order of hours. \r\n- If the Hub has a prologned outage of several hours, we don't want to keep retrying on the order of minutes.\r\n\r\nThere actually already exists an implementation for (clipped) exponential backoff in the HuggingFace suite ([here](https://github.com/huggingface/huggingface_hub/blob/61b156a4f2e5fe1a492ed8712b26803e2122bde0/src/huggingface_hub/utils/_http.py#L306)), but I don't think it is used here.\r\n\r\nThe requirements are basically that you have an initial minimum waiting time and a maximum waiting time, and with each retry, the waiting time is doubled. We don't want to overload your servers with needless retries, especially when they're down :sweat_smile:", "Oh, I've just remembered that we added retries to the `HfFileSystem` in `huggingface_hub` 0.21.0 (see [this](https://github.com/huggingface/huggingface_hub/blob/61b156a4f2e5fe1a492ed8712b26803e2122bde0/src/huggingface_hub/hf_file_system.py#L703)), so I'll close the linked PR as we don't want to retry the retries :).\r\n\r\nI agree with the exponential backoff suggestion, so I'll open another PR.", "@mariosasko The call you linked indeed points to the implementation I linked in my previous comment, yes, but it has no configurability. Arguably, you want to have this hidden backoff under the hood that catches small network disturbances on the time scale of seconds -- perhaps even with hardcoded limits as is the case currently -- but you also still want to have a separate backoff on top of that with the configurability as suggested by @lhoestq in [the comment I linked](https://github.com/huggingface/datasets/issues/6172#issuecomment-1794876229).\r\n\r\nMy particular use-case is that I'm streaming a dataset while training on a university cluster with a very long scheduling queue. This means that when the backoff runs out of retries (which happens in under 30 seconds with the call you linked), I lose my spot on the cluster and have to queue for a whole day or more. Ideally, I should be able to specify that I want to retry for 2 to 3 hours but with more and more time between requests, so that I can smooth over hours-long outages without a setback of days.", "I also have my runs crash a surprising amount due to the dataloader crashing because of the hub, some way to address this would be nice." ]

[]

[ "Hi! I've opened a [PR](https://huggingface.co/datasets/GEM/wiki_auto_asset_turk/discussions/5) with a fix. While waiting for it to be merged, you can load the dataset from the PR branch with `datasets.load_dataset(\"GEM/wiki_auto_asset_turk\", revision=\"refs/pr/5\")`", "Thanks Mario. Still getting the same issue though with the suggested fix\r\n\r\n#cat gem_sari.py\r\nimport datasets\r\nprint (datasets.__version__)\r\ndataset =datasets.load_dataset(\"GEM/wiki_auto_asset_turk\", revision=\"refs/pr/5\")\r\n\r\nEnd up with \r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/load.py\", line 2582, in load_dataset\r\n builder_instance.download_and_prepare(\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/builder.py\", line 1005, in download_and_prepare\r\n self._download_and_prepare(\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/builder.py\", line 1767, in _download_and_prepare\r\n super()._download_and_prepare(\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/builder.py\", line 1100, in _download_and_prepare\r\n self._prepare_split(split_generator, **prepare_split_kwargs)\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/builder.py\", line 1565, in _prepare_split\r\n split_info = self.info.splits[split_generator.name]\r\n ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/splits.py\", line 532, in __getitem__\r\n instructions = make_file_instructions(\r\n ^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/arrow_reader.py\", line 121, in make_file_instructions\r\n info.name: filenames_for_dataset_split(\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"/Library/Frameworks/Python.framework/Versions/3.12/lib/python3.12/site-packages/datasets/naming.py\", line 72, in filenames_for_dataset_split\r\n prefix = os.path.join(path, prefix)\r\n ^^^^^^^^^^^^^^^^^^^^^^^^^^\r\n File \"<frozen posixpath>\", line 76, in join\r\nTypeError: expected str, bytes or os.PathLike object, not NoneType", "Hmm, that's weird. Maybe try deleting the cache with `!rm -rf ~/.cache/huggingface/datasets` and then re-download.", "Tried that a couple of time. It does download the data fresh but end up with same error. Is there a way to see if its using the right version ?", "You can check the version with `python -c \"import datasets; print(datasets.__version__)\"`", "the datasets version is 2.18. \r\n\r\nI wanted to see if the command datasets.load_dataset(\"GEM/wiki_auto_asset_turk\", revision=\"refs/pr/5\") is using the right revision (refs/pr/5). \r\n\r\n\r\n\r\n\r\n\r\n ", "Still have this problem", "The issue is fixed once the fixing PR has been merged and the dataset has been converted to Parquet.\r\n\r\nIf the problem persists on your side, you should update your `datasets` library:\r\n```shell\r\npip install -U datasets\r\n```\r\nAnd if you have already the latest version of `datasets`, then you need to delete the old version of this dataset in your cache:\r\n```shell\r\nrm -fr ~/.cache/huggingface/datasets/GEM___wiki_auto_asset_turk\r\nrm -fr ~/.cache/huggingface/modules/datasets_modules/datasets/GEM--wiki_auto_asset_turk\r\n```" ]

[]

[]

[ "There are 2 workarounds, tho:\r\n1. Download datasets from web and just load them locally\r\n2. Use metadata directly (temporal solution, since metadata can change)\r\n```\r\nimport datasets\r\nfrom datasets.data_files import DataFilesDict, DataFilesList\r\n\r\ndata_files_list = DataFilesList(\r\n [\r\n \"hf://datasets/allenai/c4@1588ec454efa1a09f29cd18ddd04fe05fc8653a2/en/c4-train.00000-of-01024.json.gz\"\r\n ],\r\n [(\"allenai/c4\", \"1588ec454efa1a09f29cd18ddd04fe05fc8653a2\")],\r\n)\r\ndata_files = DataFilesDict({\"train\": data_files_list})\r\nc4_dataset = datasets.load_dataset(\r\n path=\"allenai/c4\",\r\n data_files=data_files,\r\n split=\"train\",\r\n cache_dir=\"/datesets/cache\",\r\n download_mode=\"reuse_cache_if_exists\",\r\n token=False,\r\n)\r\n```\r\nSecond solution also shows where to find the bug. I suggest that the hashing functions should always use only original parameter `data_files`, and not the one they get after connecting to the server and creating `DataFilesDict`", "Hi! You need to set the `HF_DATASETS_OFFLINE` env variable to `1` to load cached datasets offline, as explained in the docs [here](https://huggingface.co/docs/datasets/v2.19.0/en/loading#offline).", "Just tested. It doesn't work, because of the exact problem I described above: hash of dataset config is different.\r\nThe only error difference is the reason why it cannot connect to HuggingFace (now it's 'offline mode is enabled')\r\n\r\n", "Met a pretty similar issue here, as I manually load the dataset into ~/.cache and try to let `load_dataset` detect it automatically, but it will always try reach hub even I set `HF_DATASETS_OFFLINE` to 1. Have you solved it? " ]

[ "Get same error on same datasets too.", "+1", "same error" ]

[]

[ "Similar issue in text process \r\n\r\n```python\r\n\r\ntokenizer=AutoTokenizer.from_pretrained(model_dir[args.model])\r\ntrain_dataset=datasets.load_from_disk(dataset_dir[args.dataset],keep_in_memory=True)['train']\r\ntrain_dataset=train_dataset.map(partial(dname2func[args.dataset],tokenizer=tokenizer),batched=True,num_proc =50,remove_columns=train_dataset.features.keys(),desc='tokenize',keep_in_memory=True)\r\n\r\n```\r\nAfter this train_dataset will be like\r\n```python\r\nDataset({\r\n features: ['input_ids', 'labels'],\r\n num_rows: 51760\r\n})\r\n```\r\nIn which input_ids and labels are both List[int]\r\nHowever, per iter on dataset cost 7.412479639053345s ……？\r\n```python\r\nfor j in tqdm(range(len(train_dataset)),desc='first stage'):\r\n input_id,label=train_dataset['input_ids'][j],train_dataset['labels'][j]\r\n\r\n``` ", "The transform currently replaces the numpy formatting.\r\n\r\nSo you're back to copying data to long python lists which is super slow.\r\n\r\nIt would be cool for the transform to not remove the formatting in this case, but this requires a few changes in the lib" ]

[]

[]

[]

[ "Hi ! Do you mean 2.18 ? Can you try to update `fsspec` and `huggingface_hub` ?\r\n\r\n```\r\npip install -U fsspec huggingface_hub\r\n```", "Yes I meant 2.18, and it works after updating `fsspec` and `huggingface_hub`. Thanks!" ]

[]

[]

[ "Hi ! You can try to reduce `writer_batch_size`. It corresponds to the number of samples that stay in RAM before being flushed to disk" ]

[ "Probably best to implement this as a CLI command?", "Thanks for your comment, @mariosasko. Or maybe both (in Python and as CLI command)? The Python command would be just the reverse of `push_to_hub`...\r\n\r\nI am working on a draft implementation, so we can discuss about the API and UX." ]

[]

[ "This seems to be hardcoded behavior in table.py `array_cast`.\r\n```python\r\nif (\r\n not allow_number_to_str\r\n and pa.types.is_string(pa_type)\r\n and (pa.types.is_floating(array.type) or pa.types.is_integer(array.type))\r\n ):\r\n raise TypeError(\r\n f\"Couldn't cast array of type {array.type} to {pa_type} since allow_number_to_str is set to {allow_number_to_str}\"\r\n )\r\n if pa.types.is_null(pa_type) and not pa.types.is_null(array.type):\r\n raise TypeError(f\"Couldn't cast array of type {array.type} to {pa_type}\")\r\n return array.cast(pa_type)\r\n```\r\nwhere floats and integers are not cast to string but booleans are.\r\nMaybe this should be extended to booleans?", "Thanks for reporting! @Modexus Do you want to open a PR with the suggested fix?", "I'll gladly create a PR but not sure what the behavior should be.\r\n\r\nShould a value returned from map be cast to the current feature?\r\nAt the moment this seems very inconsistent since `datetime `is also cast (this would only fix `boolean`) but nested structures are not.\r\n\r\n```python\r\ndset = Dataset.from_dict({\"a\": [\"Hello world!\"]})\r\ndset = dset.map(lambda x: {\"a\": date(2021, 1, 1)})\r\n# dset[0][\"a\"] == '2021-01-01'\r\n```\r\n```python\r\ndset = Dataset.from_dict({\"a\": [\"Hello world!\"]})\r\ndset = dset.map(lambda x: {\"a\": [True]})\r\n# dset[0][\"a\"] == [True]\r\n```\r\n\r\nIs there are reason to cast the value if the user doesn't specify it explicitly?\r\nSeems tricky that some things are cast and some are not.", "Indeed, it also makes sense to raise a `TypeError` for temporal and decimal types.\r\n\r\n> Is there are reason to cast the value if the user doesn't specify it explicitly?\r\n\r\nThis is how PyArrow's built-in `cast` behaves - it allows casting from primitive types to strings. Hence, we need `allow_number_to_str` to disallow such casts (e.g., in the [scenario](https://github.com/huggingface/datasets/blob/a3bc89d8bfd47c2a175c3ce16d92b7307cdeafd6/src/datasets/arrow_writer.py#L208) when we are \"trying a type\" to preserve the original type if there is a column in the output dataset with the same name as in the input one).\r\n\r\nPS: In the PR, we can introduce `allow_numeric_to_str` (for floats, integers, decimals, booleans) and `allow_temporal_to_str` (for dates, timestamps, ...) and deprecate `allow_number_to_str` to make it clear what each parameter does.", "Would just `allow_primitive_to_str` work?\r\nThis should include all `numeric`, `boolean `and `temporal`formats.\r\n\r\nNote that at least in the [ C++ implementation](https://arrow.apache.org/docs/cpp/api/utilities.html#_CPPv410is_numericRK8DataType) `numeric `seems to exclude `boolean`.\r\n[](https://arrow.apache.org/docs/cpp/api/utilities.html#_CPPv410is_numericRK8DataType)", "Indeed, `allow_primitive_to_str` sounds better.\r\n\r\nPS: PyArrow's `pa.types.is_primitive` returns `False` for decimal types, but I think is okay for us to treat decimals as primitive types (or we can have `allow_decimal_to_str` to be fully consistent with PyArrow)", "Fixed by:\r\n- #6811" ]

[ "Hi! I'll open a PR on the Hub to fix this, but please use the Hub's [Community tab](https://huggingface.co/datasets/nyanko7/danbooru2023/discussions) to report such issues in the future.", "I've opened a [PR](https://huggingface.co/datasets/nyanko7/danbooru2023/discussions/8) in the repo, so let's continue the discussion there" ]

[ "Hi !\r\n\r\nIt's possible to multiple files at once:\r\n\r\n```python\r\ndata_files = \"data/*.jsonl\"\r\n# Or pass a list of files\r\nlangs = ['ka-ml', 'br-sr', 'ka-pt', 'id-ko', ..., 'fi-ze_zh', 'he-kk', 'ka-tr']\r\ndata_files = [f\"data/{lang}.jsonl\" for lang in langs]\r\nds = load_dataset(\"loicmagne/open-subtitles-250-bitext-mining\", data_files=data_files, split=\"train\")\r\n```\r\n\r\nAlso maybe you can add a subset called \"all\" for people that want to load all the data without having to list all the languages ?\r\n\r\n```yaml\r\n - config_name: all\r\n data_files: data/*.jsonl\r\n```\r\n", "Thanks for your reply, it is indeed much faster, however the result is a dataset where all the subsets are \"merged\" together, the language pair is lost:\r\n```\r\nDatasetDict({\r\n train: Dataset({\r\n features: ['sentence1', 'sentence2'],\r\n num_rows: 247809\r\n })\r\n})\r\n```\r\nI guess I could add a 'lang' feature for each row in the dataset, is there a better way to do it ?", "Hi @lhoestq over at https://github.com/embeddings-benchmark/mteb/issues/530 we have started examining these issues and would love to make a PR for datasets if we believe there is a way to improve the speed. As I assume you have a better overview than me @lhoestq, would you be interested in a PR, and might you have an idea about where we would start working on it?\r\n\r\nWe see a speed comparison of \r\n1. 15 minutes (for ~20% of the languages) when loaded using a for loop\r\n2. 17 minutes using the your suggestion\r\n3. ~30 seconds when using @loicmagne \"merged\" method.\r\n\r\nWorth mentioning is that solution 2 looses the language information.", "Can you retry using `datasets` 2.19 ? We improved a lot the speed of downloading datasets with tons of small files.\r\n\r\n```\r\npip install -U datasets\r\n```\r\n\r\nNow this takes 17sec on my side instead of the 17min minutes @loicmagne mentioned :)\r\n\r\n```python\r\n>>> %time ds = load_dataset(\"loicmagne/open-subtitles-250-bitext-mining\", data_files=\"data/*.jsonl\")\r\nDownloading readme: 100%|█████████████████████████████████| 13.7k/13.7k [00:00<00:00, 5.47MB/s]\r\nResolving data files: 100%|█████████████████████████████████| 250/250 [00:00<00:00, 612.51it/s]\r\nDownloading data: 100%|██████████████████████████████████| 250/250 [00:12<00:00, 19.68files/s]\r\nGenerating train split: 247809 examples [00:00, 1057071.08 examples/s]\r\nCPU times: user 4.95 s, sys: 3.1 s, total: 8.05 s\r\nWall time: 17.4 s\r\n```", "> Can you retry using `datasets` 2.19 ? We improved a lot the speed of downloading datasets with tons of small files.\r\n> \r\n> ```\r\n> pip install -U datasets\r\n> ```\r\n> \r\n> Now this takes 17sec on my side instead of the 17min minutes @loicmagne mentioned :)\r\n> \r\n> ```python\r\n> >>> %time ds = load_dataset(\"loicmagne/open-subtitles-250-bitext-mining\", data_files=\"data/*.jsonl\")\r\n> Downloading readme: 100%|█████████████████████████████████| 13.7k/13.7k [00:00<00:00, 5.47MB/s]\r\n> Resolving data files: 100%|█████████████████████████████████| 250/250 [00:00<00:00, 612.51it/s]\r\n> Downloading data: 100%|██████████████████████████████████| 250/250 [00:12<00:00, 19.68files/s]\r\n> Generating train split: 247809 examples [00:00, 1057071.08 examples/s]\r\n> CPU times: user 4.95 s, sys: 3.1 s, total: 8.05 s\r\n> Wall time: 17.4 s\r\n> ```\r\n\r\nI was actually just noticing that, I bumped from 2.18 to 2.19 and got a massive speedup, amazing!\r\n\r\nAbout the fact that subset names are lost when loading all files at once, currently my solution is to add a 'lang' feature to each rows, convert to polars and use:\r\n\r\n```python\r\nds_split = ds.to_polars().group_by('lang')\r\n```\r\n\r\nIt's fast so I think it's an acceptable solution, but is there a better way to do it ?", "It's the fastest way I think :)\r\n\r\nAlternatively you can download the dataset repository locally using [huggingface_hub](https://huggingface.co/docs/huggingface_hub/guides/download) (either via CLI or in python) and load the subsets one by one locally using a for loop as you were doing before (just pass the directory path to load_dataset instead of the dataset_id). " ]

[ "Good catch! Feel free to open a PR with the suggested fix :).", "There is also the [`MockDownloadManager`](https://github.com/JonasLoos/datasets/blob/main/src/datasets/download/mock_download_manager.py#L33), which seems like it might get passed here too. However, to me, it doesn't really seem relevant to the users of the datasets library, so I would just ignore it. What do you think, @mariosasko?", "The API (`dummy_data` CLI command ) that uses the `MockDownloadManager` has been deprecated, so ignoring it sounds good!" ]

[ "Finally:\r\n- the initial issue seems it was temporary\r\n- there is a different issue now: https://github.com/huggingface/datasets/actions/runs/8627153993/job/23646584590?pr=6797\r\n```\r\nFAILED tests/test_load.py::ModuleFactoryTest::test_HubDatasetModuleFactoryWithParquetExport - datasets.utils._dataset_viewer.DatasetViewerError: No exported Parquet files available.\r\nFAILED tests/test_load.py::ModuleFactoryTest::test_HubDatasetModuleFactoryWithParquetExport_errors_on_wrong_sha - datasets.utils._dataset_viewer.DatasetViewerError: No exported Parquet files available.\r\nFAILED tests/test_load.py::test_load_dataset_builder_for_community_dataset_with_script - AssertionError: assert 'dataset_with_script' == 'parquet'\r\n \r\n - parquet\r\n + dataset_with_script\r\n```\r\n\r\nMaybe related to `hf-internal-testing/dataset_with_script` dataset: https://huggingface.co/datasets/hf-internal-testing/dataset_with_script", "This URL: https://datasets-server.huggingface.co/parquet?dataset=hf-internal-testing/dataset_with_script\r\nraises:\r\n> {\"error\":\"The dataset viewer doesn't support this dataset because it runs arbitrary python code. Please open a discussion in the discussion tab if you think this is an error and tag @lhoestq and @severo.\"}\r\n\r\nWas there a recent change on the Hub enforcing this behavior?", "OK, I just saw this PR:\r\n- https://github.com/huggingface/dataset-viewer/pull/2689\r\n\r\nOnce merged and deployed, it should fix the issue.", "Once the script-dataset has been allowed in the dataset-viewer, we should fix our test to make the CI pass.\r\n\r\nI am addressing this." ]

[]

[ "I realized I was passing a string column to this instead of float. Is it possible to add a warning or error to prevent users from falsely believing there's a bug?", "Hello!\r\n\r\nI agree that we could add some safeguards around the type of `ds[column]`. At least for FAISS, we need the column to be made of embeddings as FAISS doesn't perform the embeddings itself.\r\n\r\nI can propose a PR sometime this week.", "@Dref360 thanks for the initiative!" ]

[]

[ "Default `writer_batch_size `is set to 1000 (see [map](https://huggingface.co/docs/datasets/v2.16.1/en/package_reference/main_classes#datasets.Dataset.map)).\r\nThe \"tmp1335llua\" is probably the temp file it creates while writing to disk.\r\nMaybe try lowering the `writer_batch_size`.\r\n\r\nFor multi-processing you should probably pass the `processor `as an argument (with e.g. partial) to the function or create it inside so that the sub-processes have access to it and maybe add `if __name__ == \"__main__\"` (not sure that's necessary?).\r\n", "Hi @Modexus,\r\n\r\nThank you very much for the help! Yep after playing around with map, I managed to get the parallel processing to work by implementing it like you suggested.\r\n\r\nRegarding the temp files, it seems like the temp files just keep growing in size as the map continues. Eventually, once map finishes, the temp files are deleted, but they are instead saved as cache .arrow files. These cache files are absolutely gigantic (~ 30-50x the size of the initial dataset!).\r\n\r\nAfter playing around with the `prepare_dataset()` function above, it seems this issue is caused by the following line in the function, where the log-Mel spectrogram of the audio is calculated:\r\n\r\n`# compute log-Mel input features from input audio array\r\n batch[\"input_features\"] = processor.feature_extractor(audio[\"array\"], \r\n sampling_rate=audio[\"sampling_rate\"]).input_features[0]\r\n`\r\n\r\nWhen I remove this line, the final cache files are approximately the same size as the initial dataset.\r\n\r\nCan I check whether this is expected behavior with the whisper feature extractor? I cant imagine the spectrograms are that large!\r\n\r\nThank you so much for the help!", "I'm having a similar issue with the spectrographs taking up an incredibly large amount of space. (i.e. 100GB for 3GB of audio). Is this really normal behavior?", "Upon taking a look at the hex contents of the mapped dataset files I found that the overwhelming majority of the data contained within them was duplicated junk similar to this. I'm not very familiar with the inner workings of AI but I have to assume this is an inefficient way of storing data at best and a bug at worst.\r\n\r\n", "Same problem, dataset.map takes long time to process 12GB raw audio data and create 200GB cache file. Is there any method can run process(map) during train, instead current run \r\nonce and save cache file ? ", "Same issue here. Just trying to normalise image data for a 300MB dataset, ends up with an 11GB cache. The initial .map() call takes 80s over the 15000 images, but then simply iterating over the dataset takes almost 2 minutes. It should be doing no processing here! Something seems wrong.\r\nkeep_in_memory=True also offers no speedup.\r\nEDIT: Running the normalisation with set_transform (i.e. on the fly) iterates through the dataset in 18s. With no normalisation it takes around 14s. No reason for .map() to take 5 mins!", "@eufrizz How you handle this using set_transform?\r\nI have a really big dataset of size 1.2TB and i am going to use it for fine-tunning whisper model. if i use map for dataset_preparing function it will take over 20 days!!!", "> @eufrizz How you handle this using set_transform?\n> I have a really big dataset of size 1.2TB and i am going to use it for fine-tunning whisper model. if i use map for dataset_preparing function it will take over 20 days!!!\n\nJust give the preprocessing function you were using for map to set_transform. Just look at the set_transform documentation. If you're going to do lots of epochs you might be better off just saving the preprocessed data into a new dataset. " ]

[ "All data is saved in the arrow format on disk.\r\nIf you return a tensor it gets converted to arrow before saving to disk when using map.\r\n\r\nTo get a tensor when you access data elements you can use `dataset.set_format(\"pt\")`.\r\nNote that this just changes how the data is loaded, not how it is stored.", "> All data is saved in the arrow format on disk. If you return a tensor it gets converted to arrow before saving to disk when using map.\r\n> \r\n> To get a tensor when you access data elements you can use `dataset.set_format(\"pt\")`. Note that this just changes how the data is loaded, not how it is stored.\r\n\r\nThank you very much for your explanation, I understand what you mean now. So you're saying that when streaming=True, there's no need to convert it to the arrow format and save it to disk. But if we directly load all formats and then convert them into the arrow format after passing through the map function, it will convert torch.Tensor into a List. I see." ]

[ "From my current understanding, this timeout is only used when we need to get the results.\r\n\r\nOne of:\r\n1. All tasks are done\r\n2. One worker died\r\n\r\nYour function should work fine and it's definitely a bug if it doesn't.", "When one of the `map`'s worker processes crashes, the linked code re-raises an error from the crash and returns it to the caller.\r\n\r\nIf your question is how to limit the time of long-running tasks/worker processes, such functionality doesn't exist in `datasets` (yet), which means you need to implement it yourself.\r\n\r\nE.g., you can implement it using the built-in `signal` module like this:\r\n```python\r\nimport time\r\nimport signal\r\nfrom contextlib import contextmanager\r\n\r\nfrom datasets import Dataset\r\n\r\n\r\n@contextmanager\r\ndef max_exec_time(t):\r\n def raise_timeout_handler(signum, frame):\r\n raise TimeoutError\r\n \r\n orig_handler = signal.getsignal(signal.SIGALRM)\r\n signal.signal(signal.SIGALRM, raise_timeout_handler)\r\n try:\r\n signal.alarm(t)\r\n yield\r\n finally:\r\n signal.alarm(0)\r\n signal.signal(signal.SIGALRM, orig_handler)\r\n\r\n\r\ndef worker(example, rank):\r\n try:\r\n with max_exec_time(20): # 20 sec execution limit\r\n if rank % 2 == 0:\r\n time.sleep(50) # simulate a long-running task\r\n example[\"a\"] = 100\r\n except TimeoutError:\r\n example[\"a\"] = None # Or return empty batches here in the \"batched\" mode\r\n return example\r\n\r\ndata = Dataset.from_list([{\"a\": 1}, {\"a\": 2}])\r\ndata = data.map(worker, num_proc=2, with_rank=True)\r\nprint(data[0])\r\n```", "> From my current understanding, this timeout is only used when we need to get the results.\r\n> \r\n> One of:\r\n> \r\n> 1. All tasks are done\r\n> 2. One worker died\r\n> \r\n> Your function should work fine and it's definitely a bug if it doesn't.\r\n\r\nthanks for responding! can you reproduce the stuck with the above example code?", "> When one of the `map`'s worker processes crashes, the linked code re-raises an error from the crash and returns it to the caller.\r\n> \r\n> If your question is how to limit the time of long-running tasks/worker processes, such functionality doesn't exist in `datasets` (yet), which means you need to implement it yourself.\r\n> \r\n> E.g., you can implement it using the built-in `signal` module like this:\r\n> \r\n> ```python\r\n> import time\r\n> import signal\r\n> from contextlib import contextmanager\r\n> \r\n> from datasets import Dataset\r\n> \r\n> \r\n> @contextmanager\r\n> def max_exec_time(t):\r\n> def raise_timeout_handler(signum, frame):\r\n> raise TimeoutError\r\n> \r\n> orig_handler = signal.getsignal(signal.SIGALRM)\r\n> signal.signal(signal.SIGALRM, raise_timeout_handler)\r\n> try:\r\n> signal.alarm(t)\r\n> yield\r\n> finally:\r\n> signal.alarm(0)\r\n> signal.signal(signal.SIGALRM, orig_handler)\r\n> \r\n> \r\n> def worker(example, rank):\r\n> try:\r\n> with max_exec_time(20): # 20 sec execution limit\r\n> if rank % 2 == 0:\r\n> time.sleep(50) # simulate a long-running task\r\n> example[\"a\"] = 100\r\n> except TimeoutError:\r\n> example[\"a\"] = None # Or return empty batches here in the \"batched\" mode\r\n> return example\r\n> \r\n> data = Dataset.from_list([{\"a\": 1}, {\"a\": 2}])\r\n> data = data.map(worker, num_proc=2, with_rank=True)\r\n> print(data[0])\r\n> ```\r\n\r\nthanks for responding! However, I don't think we should use `signal` in the context of multiprocessing since sometimes it will crash one process and raise the following error\r\nhttps://github.com/huggingface/datasets/blob/c3ddb1ef00334a6f973679a51e783905fbc9ef0b/src/datasets/utils/py_utils.py#L664", "> thanks for responding! However, I don't think we should use signal in the context of multiprocessing since sometimes it will crash one process and raise the following error\r\n\r\nThe above code has `try/except` to catch the error from the handler. Or do you get an error other than `TimeoutError`?", "> > thanks for responding! However, I don't think we should use signal in the context of multiprocessing since sometimes it will crash one process and raise the following error\r\n> \r\n> The above code has `try/except` to catch the error from the handler. Or do you get an error other than `TimeoutError`?\r\n\r\nyup, it will raise the RuntimeError: https://github.com/huggingface/datasets/blob/c3ddb1ef00334a6f973679a51e783905fbc9ef0b/src/datasets/utils/py_utils.py#L667C19-L670C22\r\n\r\n```\r\n raise RuntimeError(\r\n \"One of the subprocesses has abruptly died during map operation.\"\r\n \"To debug the error, disable multiprocessing.\"\r\n )\r\n```" ]

[ "Hi! You can fix this by updating the `datasets` package with `pip install -U datasets` and restarting the notebook.\r\n", "Kaggle removed the problematic `datasets==2.1.0` pin last week, so I'm closing this issue (now it pre-installs the latest version)." ]

[ "This may be a solution that only changes `cast_storage` of `Image`.\r\nHowever, I'm not totally sure that the assumptions hold that are made about the `ListArray`.\r\n\r\n```python\r\nelif pa.types.is_list(storage.type):\r\n from .features import Array3DExtensionType\r\n\r\n def get_shapes(arr):\r\n shape = ()\r\n while isinstance(arr, pa.ListArray):\r\n len_curr = len(arr)\r\n arr = arr.flatten()\r\n len_new = len(arr)\r\n shape = shape + (len_new // len_curr,)\r\n return shape\r\n\r\n def get_dtypes(arr):\r\n dtype = storage.type\r\n while hasattr(dtype, \"value_type\"):\r\n dtype = dtype.value_type\r\n return dtype\r\n\r\n arrays = []\r\n for i, is_null in enumerate(storage.is_null()):\r\n if not is_null.as_py():\r\n storage_part = storage.take([i])\r\n shape = get_shapes(storage_part)\r\n dtype = get_dtypes(storage_part)\r\n\r\n extension_type = Array3DExtensionType(shape=shape, dtype=str(dtype))\r\n array = pa.ExtensionArray.from_storage(extension_type, storage_part)\r\n arrays.append(array.to_numpy().squeeze(0))\r\n else:\r\n arrays.append(None)\r\n\r\n bytes_array = pa.array(\r\n [encode_np_array(arr)[\"bytes\"] if arr is not None else None for arr in arrays],\r\n type=pa.binary(),\r\n )\r\n path_array = pa.array([None] * len(storage), type=pa.string())\r\n storage = pa.StructArray.from_arrays(\r\n [bytes_array, path_array], [\"bytes\", \"path\"], mask=bytes_array.is_null()\r\n )\r\n```\r\n(Edited): to handle nulls\r\n\r\nNotably this doesn't change anything about the passing through of data or other things, just in the `Image` class.\r\nSeems quite fast:\r\n```bash\r\nFri Apr 5 17:55:51 2024 restats\r\n\r\n 63818 function calls (61995 primitive calls) in 0.812 seconds\r\n\r\n Ordered by: cumulative time\r\n List reduced from 1051 to 20 due to restriction <20>\r\n\r\n ncalls tottime percall cumtime percall filename:lineno(function)\r\n 47/1 0.000 0.000 0.810 0.810 {built-in method builtins.exec}\r\n 2/1 0.000 0.000 0.810 0.810 <string>:1(<module>)\r\n 2/1 0.000 0.000 0.809 0.809 arrow_dataset.py:594(wrapper)\r\n 2/1 0.000 0.000 0.809 0.809 arrow_dataset.py:551(wrapper)\r\n 2/1 0.000 0.000 0.809 0.809 arrow_dataset.py:2916(map)\r\n 3 0.000 0.000 0.807 0.269 arrow_dataset.py:3277(_map_single)\r\n 1 0.000 0.000 0.760 0.760 arrow_writer.py:589(finalize)\r\n 1 0.000 0.000 0.760 0.760 arrow_writer.py:423(write_examples_on_file)\r\n 1 0.000 0.000 0.759 0.759 arrow_writer.py:527(write_batch)\r\n 1 0.001 0.001 0.754 0.754 arrow_writer.py:161(__arrow_array__)\r\n 2/1 0.000 0.000 0.719 0.719 table.py:1800(wrapper)\r\n 1 0.000 0.000 0.719 0.719 table.py:1950(cast_array_to_feature)\r\n 1 0.006 0.006 0.718 0.718 image.py:209(cast_storage)\r\n 1 0.000 0.000 0.451 0.451 image.py:361(encode_np_array)\r\n 1 0.000 0.000 0.444 0.444 image.py:343(image_to_bytes)\r\n 1 0.000 0.000 0.413 0.413 Image.py:2376(save)\r\n 1 0.000 0.000 0.413 0.413 PngImagePlugin.py:1233(_save)\r\n 1 0.000 0.000 0.413 0.413 ImageFile.py:517(_save)\r\n 1 0.000 0.000 0.413 0.413 ImageFile.py:545(_encode_tile)\r\n 397 0.409 0.001 0.409 0.001 {method 'encode' of 'ImagingEncoder' objects}\r\n```", "Also encounter this problem. Has been strugging with it for a long time...", "This actually applies to all arrays (numpy or tensors like in torch), not only from external files.\r\n```python\r\nimport numpy as np\r\nimport datasets\r\n\r\nds = datasets.Dataset.from_dict(\r\n {\"image\": [np.random.randint(0, 255, (2048, 2048, 3), dtype=np.uint8)]},\r\n features=datasets.Features({\"image\": datasets.Image(decode=True)}),\r\n)\r\nds.set_format(\"numpy\")\r\n\r\nds = ds.map(load_from_cache_file=False)\r\n```" ]

[ "Hello!\r\n\r\nThis is due to how pandas write numpy arrays to csv. [Source](https://stackoverflow.com/questions/54753179/to-csv-saves-np-array-as-string-instead-of-as-a-list)\r\nTo fix this, you can convert them to list yourselves.\r\n\r\n```python\r\ndf = ds.to_pandas()\r\ndf['int'] = df['int'].apply(lambda arr: list(arr))\r\ndf.to_csv(index=False, '../output/temp.csv')\r\n```\r\n\r\nI think it would be good if `datasets` would do the conversion itself, but it's a breaking change and I would wait for the greenlight from someone from HF." ]

[ "Hi! `metadata.jsonl` (or `metadata.csv`) is the only allowed name for the `imagefolder`'s metadata files.", "@mariosasko hey i tried with metadata.jsonl also and it still doesn't get the right columns", "@mariosasko it says metadata.csv not found\r\n<img width=\"1150\" alt=\"image\" src=\"https://github.com/huggingface/datasets/assets/81643693/3754980c-6185-4413-88fa-b499bcdd4195\">\r\n\r\ndataset = load_dataset('/dataset',metadata.csv) \r\n\r\n| workspace\r\n|| source code\r\n| dataset\r\n| |-- images\r\n| |-- metadata.csv\r\n| |-- metadata.jsonl\r\n| |-- padded_images\r\n\r\nExample of metadata.jsonl file\r\n{\"caption\": \"a drawing depicts a full shot of a black t-shirt with a triangular pattern on the front there is a white label on the left side of the triangle\", \"image\": \"images/212734.png\", \"gaussian_padded_image\": \"padded_images/p_212734.png\"}\r\n{\"caption\": \"an eye-level full shot of a large elephant and a baby elephant standing in a watering hole on the left side is a small elephant with its head turned to the right of dry land, trees, and bushes\", \"image\": \"images/212735.png\", \"gaussian_padded_image\": \"padded_images/p_212735.png\"}\r\n", "Loading more than one image per row with `imagefolder` is not supported currently. You can subscribe to https://github.com/huggingface/datasets/issues/5760 to see when it will be.\r\n\r\nInstead, you can load the dataset with `Dataset.from_generator`:\r\n```python\r\nimport json\r\nfrom datasets import Dataset, Value, Image, Features\r\n\r\ndef gen():\r\n with open(\"./dataset/metadata.jsonl\") as f:\r\n for line in f:\r\n line = json.loads(line)\r\n yield {\"caption\": line[\"caption\"], \"image\": os.path.join(\"./dataset\", line[\"image\"], \"gaussian_padded_image\": os.path.join(\"./dataset\", line[\"gaussian_padded_image\"]))}\r\n\r\nfeatures = Features({\"caption\": Value(\"string\"), \"image\": Image(), \"gaussian_padded_image\": Image()})\r\ndataset = Dataset.from_generator(gen, features=features)\r\n```\r\n(E.g., if you want to share this dataset on the Hub, you can call `dataset.push_to_hub(...)` afterward)", "hi Thanks for sharing this, Actually I was trying with a webdataset format of the data as well and it did'nt work. Could you share how i can create Dataset object from webdataset format of this data?" ]