Datasets:

---

mingyujeon

/

github-issues

like 0

[]

[]

[ "@lhoestq " ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7417). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7416). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "Hi ! if it's an option I'd suggest to have one sequence per row instead.\n\nOtherwise you'd have to make your own save/load mechanism", "Saving one sequence per row is very difficult and heavy and makes all the optimizations pointless. How would a custom save/load mechanism look like?" ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7414). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "Yes this is correct and it works with huggingface datasets as well ! Feel free to include an example here: https://github.com/huggingface/datasets/blob/main/docs/source/video_dataset.mdx" ]

[]

[]

[]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7411). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "Thanks for the fix! We have been affected by this a lot when we try to use LLM Foundry with larger multimodal ICL datasets. ", "@lorabit110 are you able to test it out for your case as well? Would be great to get a second validation that it actually fixes the issue. Thanks!" ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7410). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7409). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7408). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[ "@lhoestq", "there is no per-module maintainer and the list is me alone nowadays ^^'", "@lhoestq \nOh... I feel for you. \nWhat are your criteria for choosing a core maintainer? \nIt seems like it's too much work for you to manage all this code by yourself.\n\nAlso, if you don't mind, can you check this PR for me?\n#7368 I'd like this to be added as soon as possible because I need it." ]

[]

[]

[ "Many python packages out there, including `huggingface_hub`, do load the environment variables on import.\nYou should `load_dotenv()` before importing the libraries.\n\nFor example you can move all you imports inside your `main()` function" ]

[]

[ { "login": "lhoestq", "id": 42851186, "node_id": "MDQ6VXNlcjQyODUxMTg2", "avatar_url": "https://avatars.githubusercontent.com/u/42851186?v=4", "gravatar_id": "", "url": "https://api.github.com/users/lhoestq", "html_url": "https://github.com/lhoestq", "followers_url": "https://api.github.com/users/lhoestq/followers", "following_url": "https://api.github.com/users/lhoestq/following{/other_user}", "gists_url": "https://api.github.com/users/lhoestq/gists{/gist_id}", "starred_url": "https://api.github.com/users/lhoestq/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/lhoestq/subscriptions", "organizations_url": "https://api.github.com/users/lhoestq/orgs", "repos_url": "https://api.github.com/users/lhoestq/repos", "events_url": "https://api.github.com/users/lhoestq/events{/privacy}", "received_events_url": "https://api.github.com/users/lhoestq/received_events", "type": "User", "user_view_type": "public", "site_admin": false } ]

[ "Thanks for reporting, I'll fix the regression today", "I just released `datasets` 3.3.1 with a fix, let me know if it's good now :)", "@lhoestq it fixed the issue.\n\nThis was (very) fast, thank you very much!" ]

[]

[]

[]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7401). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "Hi ! the `desc` parameter is only available for Dataset / DatasetDict for the progress bar of `map()``\n\nSince IterableDataset only runs the map functions when you iterate over the dataset, there is no progress bar and `desc` is useless. We could still add the argument for parity but it wouldn't be used for anything", "I think you should add it. It doesn't hurt. The reason I ran into it was because I re-wrote a pipeline to use either a stream or a fully loaded dataset. Of course I can simply remove it but it is nice to have on the memory loaded dataset. " ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7398). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "Hi ! feel free to uplad the CSV on https://huggingface.co/datasets :)\r\n\r\nwe don't store the datasets' data in this github repository" ]

[]

[]

[ "I transferred to the `datasets` repository. Is there any retry mechanism in `datasets` @lhoestq ?\n\nAnother solution @hotchpotch if you want to get your dataset pushed to the Hub in a robust way is to save it to a local folder first and then use `huggingface-cli upload-large-folder` (see https://huggingface.co/docs/huggingface_hub/guides/upload#upload-a-large-folder). It has better retry mechanism in case of failure.", "There is no retry mechanism for `api.preupload_lfs_files` in `push_to_hub()` but we can definitely add one here\n\nhttps://github.com/huggingface/datasets/blob/de062f0552a810c52077543c1169c38c1f0c53fc/src/datasets/arrow_dataset.py#L5372", "@Wauplin \n\nThank you! I believe that to use load_dataset() to read data from Hugging Face, we need to first save the markdown metadata and parquet files in our local filesystem, then upload them using upload-large-folder. If you know how to do this, could you please let me know?\n\n", "@lhoestq \n\nI see, so adding a retry mechanism there would solve it. If I continue to have issues, I'll consider implementing that kind of solution." ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7396). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7395). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7393). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[]

[]

[ "See also <https://discuss.huggingface.co/t/datasetdict-push-to-hub-failing-with-payload-to-large/140083/8>\n" ]

[]

[]

[]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[]

[]

[]

[ "You can actually track a running sum in map() or filter() :)\n\n```python\nnum_filtered = 0\n\ndef f(x):\n global num_filtered\n condition = len(x[\"text\"]) < 1000\n if not condition:\n num_filtered += 1\n return condition\n\nds = ds.filter(f)\nprint(num_filtered)\n```\n\nand if you want to use multiprocessing, make sure to use a variable that is shared across processes\n\n\n```python\nfrom multiprocess import Manager\n\nmanager = Manager()\nnum_filtered = manager.Value('i', 0)\n\ndef f(x):\n global num_filtered\n condition = len(x[\"text\"]) < 1000\n if not condition:\n num_filtered.value += 1\n return condition\n\nds = ds.filter(f, num_proc=4)\nprint(num_filtered.value)\n```\n\nPS: `datasets` uses `multiprocess` instead of the `multiprocessing` package to support lambda functions in map() and filter()", "Oh that's great to know!\n\nI guess this value would not be exactly synced with the batch in cases of pre-fetch and shuffle buffers and so on, but that's probably fine. Thanks!" ]

[]

[]

[ "You can probably copy the dataset in your HF account and rename the files (without having to download them to your disk). Or alternatively feel free to open a Pull Request to this dataset with the renamed file", "Thank you, that will help me work around this problem" ]

[]

[]

[ "You mean based on a condition that has to be checked on-the-fly during training ? Otherwise if you know in advance after how many samples you need to change the sampling you can simply concatenate the two mixes", "Yes, like during training, if one data sample's prediction is consistently wrong, its sampling weight gets higher and higher, and if one data sample's prediction is already correct, then we rarely sample it", "it's not possible to use `interleave_datasets()` and modify the probabilities while iterating on the dataset at the moment, so you'd have to implement your own `IterableDataset` to implement this logic" ]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[ "On second thought, probably not a good idea." ]

[]

[]

[ "@lhoestq Hi again~ Just circling back on this\r\nWondering if there’s anything I can do to help move this forward. 🤗 \r\nThanks!" ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7384). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "example of what you can do with it:\r\n\r\n```python\r\nimport aiohttp\r\nfrom huggingface_hub import get_token\r\n\r\nfrom datasets import Dataset\r\n\r\n\r\nAPI_URL = \"https://api-inference.huggingface.co/models/microsoft/Phi-3-mini-4k-instruct/v1/chat/completions\"\r\nPROMPT = \"What is this text mainly about ? Here is the text:\\n\\n```\\n{Problem}\\n```\\n\\nReply in one or two words.\"\r\n\r\nasync def query(example):\r\n headers = {\"Authorization\": f\"Bearer {get_token()}\", \"Content-Type\": \"application/json\"}\r\n json = {\"messages\": [{\"role\": \"user\", \"content\": PROMPT.format(Problem=example[\"Problem\"])}], \"max_tokens\": 20, \"seed\": 42}\r\n async with aiohttp.ClientSession() as session, session.post(API_URL, headers=headers, json=json) as response:\r\n output = await response.json()\r\n return {\"output\": output[\"choices\"][0][\"message\"][\"content\"]}\r\n\r\nds = Dataset.from_dict({\"Problem\": [\"1 + 1\"] * 10})\r\nds = ds.map(query)\r\nprint(ds[0])\r\n# {'Problem': '1 + 1', 'output': 'Arithmetic'}\r\n```" ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7382). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update." ]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[ "I'd be in favor of that ! I saw many people implementing their own iterables that wrap a dataset just to iterate on a single column, that would make things more practical.\n\nKinda related: https://github.com/huggingface/datasets/issues/5847", "(For anyone's information, I'm going on vacation for the next 3 weeks, so the work is postponed. If anyone can implement this feature within the next 4 weeks, go ahead :) )" ]

[]

[]

[]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[ "Hi ! This sounds reasonable to me, feel free to open a PR :)" ]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[ "Hi ! Unfortunately the Sparse Tensor structure in Arrow is not part of the Arrow format (yes it's confusing...), so it's not possible to use it in `datasets`. It's a separate structure that doesn't correspond to any type or extension type in Arrow.\n\nThe Arrow community recently added an extension type for fixed shape tensors at https://arrow.apache.org/docs/format/CanonicalExtensions.html#fixed-shape-tensor, it should be possible to contribute an extension type for sparse tensors as well." ]

[]

[]

[]

[]

[]

[ "Make sure you have installed a recent version of `soundfile`" ]

[]

[]

[]

[]

[]

[ "\n\n\n\nAdding a img from memray\nhttps://gist.github.com/sam-hey/00c958f13fb0f7b54d17197fe353002f" ]

[]

[]

[]

[]

[]

[ "EDIT: seems to be all good now. I'll add a comment if the error happens again within the next 48 hours. If it doesn't, I'll just close the topic." ]

[]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_7370). All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.", "merging this and will make some docs and communications around using polars for optimizing data processing :)" ]

[]

[]

[ "I'd prefer even more verbose errors; like `\"file123.mp3\" is referenced in metadata.csv, but not found in the data directory '/path/to/audiofolder' ! (and 100+ more missing files)` Or something along those lines." ]

[]

[]

[ "Can you check this out, @lhoestq?", "cc @lhoestq @albertvillanova ", "@lhoestq\r\n", "@lhoestq\r\n", "@lhoestq" ]

[]

[]

[]

[]

[]

[]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" } ]

[]

[ "Looks like a [similar feature request](https://github.com/huggingface/huggingface_hub/issues/1198) was made to the HF Hub team. Is handling this at the Hub level more appropriate?\r\n\r\n(As an aside, I've gotten the [HTTP-based solution](https://github.com/huggingface/huggingface_hub/issues/1198#issuecomment-1905774983) proposed in that forum to work for simple cases.)", "yes it's more for https://github.com/huggingface/huggingface_hub cc @hanouticelina ", "yes i think @Wauplin's comment on that thread is still what we recommend" ]

[]

[]

[ "what's your `pip show Pillow` output", "same issue.. my pip show Pillow output as below:\n\n```\nName: pillow\nVersion: 11.1.0\nSummary: Python Imaging Library (Fork)\nHome-page: https://python-pillow.github.io/\nAuthor: \nAuthor-email: \"Jeffrey A. Clark\" <[email protected]>\nLicense: MIT-CMU\nLocation: [/opt/homebrew/lib/python3.10/site-packages](https://file+.vscode-resource.vscode-cdn.net/opt/homebrew/lib/python3.10/site-packages)\nRequires: \nRequired-by:\n```", "I encountered the same problem on Ubuntu system, my pip show Pillow output as below:\n\n```\nName: pillow\nVersion: 10.4.0\nSummary: Python Imaging Library (Fork)\nHome-page: https://python-pillow.org/\nAuthor: \nAuthor-email: \"Jeffrey A. Clark\" <[[email protected]](mailto:[email protected])>\nLicense: HPND\nLocation: /home/shunying/.local/lib/python3.8/site-packages\nRequires: \nRequired-by: \n```\n\nWell, solved this by specifying the pip version to my conda virtual environment :)" ]

[]

[]

[ "I got the same error and was able to resolve it by upgrading from 2.15.0 to 3.2.0.", "> I got the same error and was able to resolve it by upgrading from 2.15.0 to 3.2.0.\r\n\r\nWhat is needed is upgrading `huggingface-hub==0.27.1`. `datasets` does not appear to have anything to do with the error. The upgrade is a workaround, if the workaround works for your use case. Otherwise, this issue breaks all existing Python clients not using some minimum version of `huggingface-hub`. ", "Correct, this has to do with `huggingface_hub`, not `datasets`. Some old versions of `huggingface_hub` are unfortunately not robust to recent changes on HF. Updating `huggingface_hub` fixes the issue :)\r\n\r\nClosing this issue since it's not directly related to `datasets`" ]

[]

[]

[]