Datasets:

---

amannagrawall002

/

github-issues

like 0

[ "Hi @Darshan2104, thanks for reporting.\r\n\r\nPlease note that at Hugging Face we do not host the data of this dataset, but just a loading script pointing to the host of the data owners.\r\n\r\nApparently the data owners changed their data host server. After googling it, I found their new website at: https://lil.nlp.cornell.edu/newsroom/index.html\r\n- Download page: https://lil.nlp.cornell.edu/newsroom/download/index.html\r\n\r\nI'm fixing the link in our Datasets library.", "@albertvillanova Thanks for the solution and link you made my day!" ]

[ "Hi @jvanz, thank you for your proposal.\r\n\r\nIn fact, we are aware that it is very common the problem you mention. Because of that, we are currently working in implementing a new version of wikipedia on the Hub, with all data preprocessed (no need to use Apache Beam), from where you will be able to use `data_files` to load only a specific subset of the data files.\r\n\r\nSee:\r\n- #3401 " ]

[ "Thanks for reporting, @anna-kay. We are fixing it.", "@albertvillanova Thank you for the fast response! However I am still getting the same error:\r\n\r\nDownloading: 2.23kB [00:00, ?B/s]\r\nTraceback (most recent call last):\r\n File \"C:\\Users\\Anna\\PycharmProjects\\summarization\\main.py\", line 17, in <module>\r\n dataset = load_dataset('reddit_tifu', 'long')\r\n File \"C:\\Users\\Anna\\Desktop\\summarization\\summarization_env\\lib\\site-packages\\datasets\\load.py\", line 1702, in load_dataset\r\n builder_instance.download_and_prepare(\r\n File \"C:\\Users\\Anna\\Desktop\\summarization\\summarization_env\\lib\\site-packages\\datasets\\builder.py\", line 594, in download_and_prepare\r\n self._download_and_prepare(\r\n File \"C:\\Users\\Anna\\Desktop\\summarization\\summarization_env\\lib\\site-packages\\datasets\\builder.py\", line 665, in _download_and_prepare\r\n verify_checksums(\r\n File \"C:\\Users\\Anna\\Desktop\\summarization\\summarization_env\\lib\\site-packages\\datasets\\utils\\info_utils.py\", line 40, in verify_checksums\r\n raise NonMatchingChecksumError(error_msg + str(bad_urls))\r\ndatasets.utils.info_utils.NonMatchingChecksumError: Checksums didn't match for dataset source files:\r\n['https://drive.google.com/uc?export=download&id=1ffWfITKFMJeqjT8loC8aiCLRNJpc_XnF']\r\n\r\nI have cleaned the cache/huggingface/datasets & cache/huggingface/modules files and also tried on another machine with a fresh installation of trasnformers & datasets. \r\nThe reddit_tifu.py that gets downloaded still has the previous url on line 51, _URL = \"https://drive.google.com/uc?export=download&id=1ffWfITKFMJeqjT8loC8aiCLRNJpc_XnF\" ", "Hi @anna-kay, I'm sorry I didn't clearly explain the details to you:\r\n- the error has been fixed in our `master` branch on GitHub: https://github.com/huggingface/datasets/commit/8ae21bf6a77175dc803ce2f1b93d18b8fbf45586\r\n- the fix will not be accessible to users in PyPI until our next release of the `datasets` library\r\n - our latest release (version 1.18.3) was made 23 days ago: https://github.com/huggingface/datasets/releases/tag/1.18.3\r\n- in the meantime, you can get the fix if you install datasets from our GitHub `master` branch:\r\n ```\r\n pip install git+https://github.com/huggingface/datasets#egg=datasets\r\n ```", "@albertvillanova Ok great, makes sence. Thank you very much for the explanation!" ]

[ "Thanks for reporting, @kolk.\r\n\r\nWe are fixing it. " ]

[ "Hi ! Indeed `map` is dropping the index right now, because one can create a dataset with more or fewer rows using `map` (and therefore the index might not be relevant anymore)\r\n\r\nI guess we could check the resulting dataset length, and if the user hasn't changed the dataset size we could keep the index, what do you think ?", "doing `.add_column(\"x\",x_data)` also removes the index. the new column might be irrelevant to the index so I don't think it should drop. \r\n\r\nMinimal example\r\n\r\n```python\r\nfrom datasets import load_dataset\r\nimport numpy as np\r\n\r\ndata=load_dataset(\"ceyda/cats_vs_dogs_sample\") #just a test dataset\r\ndata=data[\"train\"]\r\nembd_data=data.map(lambda x: {\"emb\":np.random.uniform(-1,0,50).astype(np.float32)})\r\nembd_data.add_faiss_index(column=\"emb\")\r\nprint(embd_data.list_indexes())\r\nembd_data=embd_data.add_column(\"x\",[0]*data.num_rows)\r\nprint(embd_data.list_indexes())\r\n```", "I agree `add_column` shouldn't drop the index indeed ! Is it something you'd like to contribute ? I think it's just a matter of copying the `self._indexes` dictionary to the output dataset" ]

[]

[ "Hi @jvanz, thanks for reporting.\r\n\r\nPlease note that Wikimedia website does not longer host Wikipedia dumps for so old dates.\r\n\r\nFor a list of accessible dump dates of `pt` Wikipedia, please see: https://dumps.wikimedia.org/ptwiki/\r\n\r\nYou can load for example `20220220` `pt` Wikipedia:\r\n```python\r\ndataset = load_dataset(\"wikipedia\", language=\"pt\", date=\"20220220\", beam_runner=\"DirectRunner\")\r\n```", "> ```python\r\n> dataset = load_dataset(\"wikipedia\", language=\"pt\", date=\"20220220\", beam_runner=\"DirectRunner\")\r\n> ```\r\n\r\nThank you! I did not know that I can do this. I was following the example in the error message when I do not define which language dataset I'm trying to download.\r\n\r\nI've tried something similar changing the date in the `load_dataset` call that I've shared in the bug description. Obviously, it did not work. I need to read the docs more carefully next time. My bad!\r\n\r\nThanks again and sorry for the noise.\r\n\r\n" ]

[ "Hi @Dref360, thanks a lot for your proposal.\r\n\r\nIt totally makes sense to have more flexibility when class encoding, I agree.\r\n\r\nYou could even further customize the class encoding by passing an instance of `ClassLabel` itself (instead of replicating `ClassLabel` instantiation arguments as `Dataset.class_encode_column` arguments).\r\n\r\nAnd the latter made me think of `Dataset.cast_column`...\r\n\r\nMaybe better to have some others' opinions @lhoestq @mariosasko ", "Hi @Dref360! You can use [`Dataset.align_labels_with_mapping`](https://huggingface.co/docs/datasets/master/package_reference/main_classes.html#datasets.Dataset.align_labels_with_mapping) after `Dataset.class_encode_column` to assign a different mapping of labels to ids.\r\n\r\n@albertvillanova I'd like to avoid adding more complexity to the API where it's not (absolutely) needed, so I don't think introducing a new param in `Dataset.class_encode_column` is a good idea.\r\n\r\n", "I wasn't aware that it existed thank you for the link.\n\nClosing then! " ]

[ "Hi @Muhtasham you should take a look at https://huggingface.co/blog/data-measurements-tool and accompanying demo app at https://huggingface.co/spaces/huggingface/data-measurements-tool\r\n\r\nWe would be interested in your feedback. cc @meg-huggingface @sashavor @yjernite " ]

[ "Hi @kingabzpro.\r\n\r\nI think you need to create a loading script that creates the dataset from the CSV file and the image paths.\r\n\r\nAs example, you could have a look at the Food-101 dataset: https://huggingface.co/datasets/food101\r\n- Loading script: https://huggingface.co/datasets/food101/blob/main/food101.py\r\n\r\nOnce the loading script is created, the viewer will show a previsualization of your dataset. ", "@albertvillanova I don't think this is the issue. I have created another dataset with similar files and format and it works. https://huggingface.co/datasets/kingabzpro/savtadepth-flags-V2", "Yes, you are right, that was not the issue.\r\n\r\nJust take into account that sometimes the viewer can take some time until it shows the preview of the dataset.\r\nAfter some time, yours is finally properly shown: https://huggingface.co/datasets/kingabzpro/savtadepth-flags", "The problem was resolved by deleted the dataset and creating new one with similar name and then clicking on flag button.", "I think if you make manual changes to dataset the whole system breaks. " ]

[ "We usually find issues with files hosted at Google Drive...\r\n\r\nIn this case we download the Google Drive Virus scan warning instead of the data file.", "Fixed: https://huggingface.co/datasets/head_qa/viewer/en/train. Thanks\r\n\r\n<img width=\"1551\" alt=\"Capture d’écran 2022-02-28 à 15 29 04\" src=\"https://user-images.githubusercontent.com/1676121/156000224-fd3f62c6-8b54-4df1-8911-bdcb0bac3f1a.png\">\r\n" ]

[ "Hi! If I'm not mistaken, this behavior is intentional, but I agree it could be more intuitive.\r\n\r\n@albertvillanova Do you remember why you decided not to decode columns in the `Audio` feature PR when `input_columns` is not `None`? IMO we should decode those columns, and we don't even have to use lazy structures here because the user explicitly requires them in the map transform. \r\n\r\ncc @lhoestq for visibility", "I think I excluded to decorate the function when `input_columns` were passed as a quick fix for some non-passing tests: \r\n- https://github.com/huggingface/datasets/pull/2324/commits/9d7c3e8fa53e23ec636859b4407eeec904b1b3f9\r\n\r\nThat PR was quite complex and I decided to focus on the main feature requests, leaving refinements for subsequent PRs.\r\n\r\nNote that when `input_columns` are passed, the signature of the function is effectively changed, while the decorated function expects an item (whether an example or a batch) as first arg (which is not the case when passing `input_columns`.\r\n\r\nI agree we should consider supporting the case when `input_columns` are passed." ]

[ "Thanks for reporting. The dataset viewer depends on some backend treatments, and for now, they might take some hours to get processed. We're working on improving it.", "It has finally been processed. Thanks for the patience.", "Thanks for the info @severo !" ]

[ "Fixed on master (see https://github.com/huggingface/datasets/pull/3719).", "Awesome, I did not find that one! Thanks." ]

[ "Related to: https://github.com/huggingface/datasets/issues/3444", "Cool ! `filter` will be very useful. There can be a filter that you can apply on a streaming dataset:\r\n```python\r\nload_dataset(..., streaming=True).filter(lambda x: x[\"lang\"] == \"sw\")\r\n```\r\n\r\nOtherwise if you want to apply a filter on the source files that are going to be used for streaming, the logic has to be impIemented directly in the dataset script, or if there's no dataset script this can be done with pattern matching\r\n```python\r\nload_dataset(..., lang=\"sw\") # if the dataset script supports this parameter\r\nload_dataset(..., data_files=\"data/lang=sw/*\") # if there's no dataset script, but only data files\r\n```\r\n\r\n--------------\r\n\r\nHere are also some additional ideas of API to convert from iterable to map-style dataset:\r\n```python\r\non_disk_dataset = streaming_dataset.to_disk()\r\non_disk_dataset = streaming_dataset.to_disk(path=\"path/to/my/dataset/dir\")\r\n\r\nin_memory_dataset = streaming_dataset.take(100).to_memory() # to experiment without having to write files\r\n```\r\n--------------\r\n\r\nFinally regarding `push_to_hub`, we can replace `batch_size` by `shard_size` (same API as for on-disk datasets). The default is 500MB per file\r\n\r\nLet me know what you think !", "Regarding conversion, I'd also ask for some kind of equivalent to `save_to_disk` for an `IterableDataset`.\r\n\r\nSimilarly to the streaming to hub idea, my use case would be to define a sequence of dataset transforms via `.map()`, using an `IterableDataset` as the input (so processing could start without doing whole download up-front), but streaming the resultant processed dataset just to disk.", "That makes sense @athewsey , thanks for the suggestion :)\r\n\r\nMaybe instead of the `to_disk` we could simply have `save_to_disk` instead:\r\n```python\r\nstreaming_dataset.save_to_disk(\"path/to/my/dataset/dir\")\r\non_disk_dataset = load_from_disk(\"path/to/my/dataset/dir\")\r\n\r\nin_memory_dataset = Dataset.from_list(list(streaming_dataset.take(100))) # to experiment without having to write files\r\n```", "Any updates on this?", "So far are implemented: `IterableDataset.filter()` and `Dataset.to_iterable_dataset()`.\r\n\r\nStill missing: `IterableDataset.push_to_hub()` - though there is a hack to write on disk and then push to hub using\r\n\r\n```python\r\nds_on_disk = Dataset.from_generator(streaming_ds.__iter__) # stream to disk\r\nds_on_disk.push_to_hub(...)\r\n```" ]

[ "Thnaks for reporting @jaketae. We are fixing it. " ]

[]

[]

[ "Thanks for reporting, @abhi-mosaic (related to #3655).\r\n\r\nPlease note that `xml.etree.ElementTree.parse` already supports streaming:\r\n- #3476\r\n\r\nNo need to refactor to use `open`/`xopen`. Is is enough with importing the package `as ET` (instead of `as etree`)." ]

[ "Note that we might change the heuristic and create a different config per file, at least in that case.", "Hi @severo, thanks for reporting.\r\n\r\nYes, this happens because when non-streaming, a cast of all data is done in order to \"concatenate\" it all into a single dataset (thus the error), while this casting is not done while yielding item by item in streaming mode.\r\n\r\nMaybe in streaming mode we should keep the schema (inferred from the first item) and throw an exception if a subsequent item does not conform to the inferred schema?", "Why do we want to concatenate the files? Is it the expected behavior for most datasets that lack a script and dataset info?", "These files are two different dataset configurations since they don't share the same schema.\r\n\r\nIMO the streaming mode should fail in this case, as @albertvillanova said.\r\n\r\nThere is one challenge though: inferring the schema from the first example is not robust enough in the general case - especially if some fields are nullable. I guess we can at least make sure that no new columns are added", "OK. So, if we make the streaming also fail, the dataset https://huggingface.co/datasets/huggingface/transformers-metadata will never be [viewable](https://github.com/huggingface/datasets-preview-backend/issues/144) (be it using streaming or fallback to downloading the files), right?\r\n", "Yes, until we have a way for the user to specify explicitly that those two files are different configurations.\r\n\r\nWe can maybe have some rule to detect this automatically, maybe checking the first line of each file ? That would mean that for dataset of 10,000+ files we would have to verify every single one of them just to know if there is one ore more configurations, so I'm not sure if this is a good idea", "i think requiring the user to specify that those two files are different configurations is in that case perfectly reasonable.\r\n\r\n(Maybe at some point we could however detect this type of case and prompt them to define a config mapping etc)", "OK, so, before closing the issue, what do you think should be done?\r\n\r\n> Maybe in streaming mode we should keep the schema (inferred from the first item) and throw an exception if a subsequent item does not conform to the inferred schema?\r\n\r\nor nothing?", "We should at least raise an error if a new sample has column names that are missing, or if it has extra columns. No need to check for the type for now.\r\n\r\nI'm in favor of having an error especially because we want to avoid silent issues as much as possible - i.e. when something goes wrong (when schemas don't match or some data are missing) and no errors/warnings are raised.\r\n\r\nConsistency between streaming and non-streaming is also important." ]

[ "> using command line git-lfs - [...] 300MB/s!\r\n\r\nwhich server location did you upload from?", "From GCP region `us-central1-a`.", "The most surprising part to me is the saving time. Wondering if it could be due to compression (`ParquetWriter` uses SNAPPY compression by default; it can be turned off with `to_parquet(..., compression=None)`). ", "+1 to what @mariosasko mentioned. Also, @lvwerra did you parallelize `to_parquet` using similar approach in #2747? (we used multiprocessing at the shard level). I'm working on a similar PR to add multi_proc in `to_parquet` which might give you further speed up. \r\nStas benchmarked his approach and mine in this [gist](https://gist.github.com/stas00/dc1597a1e245c5915cfeefa0eee6902c) for `lama` dataset when we were working on adding multi_proc support for `to_json`.", "@mariosasko I did not turn it off but I can try the next time - I have to run the pipeline again, anyway. \r\n\r\n@bhavitvyamalik Yes, I also sharded the dataset and used multiprocessing to save each shard. I'll have a closer look at your approach, too.", "Is there a way to read from the cache files directly as a dataset in its own" ]

[]

[ "Thanks for reporting @byw2.\r\nWe are fixing it.\r\nIn the meantime, you can load the dataset by passing `ignore_verifications=True`:\r\n ```python\r\ndataset = load_dataset(\"multi_news\", ignore_verifications=True)" ]

[ "Hi @kg-nlp, thanks for reporting.\r\n\r\nThat is weird... I guess we would need some sample data file where this behavior appears to reproduce the bug for further investigation... ", "ok, I found the reason why that two results are not same.\r\nthere is /u2029 in the text, the datasets will split sentence according to the /u2029,but when I use open function will not do that .\r\nso I want to know which function shell do that\r\nthanks" ]

[ "duplicate of https://github.com/huggingface/datasets/issues/2300" ]

[]

[ "Hi @aasem, thanks for reporting.\r\n\r\nPlease note that currently Commom Voice is hosted on our Hub as a community dataset by the Mozilla Foundation. See all Common Voice versions here: https://huggingface.co/mozilla-foundation\r\n\r\nMaybe we should add an explaining note in our \"legacy\" Common Voice canonical script? What do you think @lhoestq @mariosasko ?", "Thank you, @albertvillanova, for the quick response. I am not sure about the exact flow but I guess adding the following lines under the `_Languages` dictionary definition in [common_voice.py](https://github.com/huggingface/datasets/blob/master/datasets/common_voice/common_voice.py) might resolve the issue. I guess the dataset is recently made available so the file needs updating.\r\n\r\n```\r\n\"ur\": {\r\n \"Language\": \"Urdu\",\r\n \"Date\": \"2022-01-19\",\r\n \"Size\": \"68 MB\",\r\n \"Version\": \"ur_3h_2022-01-19\",\r\n \"Validated_Hr_Total\": 1,\r\n \"Overall_Hr_Total\": 3,\r\n \"Number_Of_Voice\": 48,\r\n },\r\n```\r\n", "@aasem for compliance reasons, we are no longer updating the `common_voice.py` script.\r\n\r\nWe agreed with Mozilla Foundation to use their community datasets instead, which will ask you to accept their terms of use:\r\n```\r\nYou need to share your contact information to access this dataset.\r\n\r\nThis repository is publicly accessible, but you have to register to access its content — don't worry, it's just one click!\r\n\r\nBy clicking on “Access repository” below, you accept that your contact information (email address and username) can be shared with the repository authors. This will let the authors get in touch for instance if some parts of the repository's contents need to be taken down for licensing reasons.\r\n\r\nBy clicking on “Access repository” below, you also agree to not attempt to determine the identity of speakers in the Common Voice dataset.\r\n\r\nYou will immediately be granted access to the contents of the dataset. \r\n```\r\n\r\nIn order to use e.g. their Common Voice dataset version 8.0, please:\r\n- First visit their dataset page: https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0\r\n- Accept their term of use by clicking \"Access repository\"\r\n- You can then load their dataset with:\r\n ```python\r\n load_dataset(\"mozilla-foundation/common_voice_8_0\", \"ur\", split=\"train+validation\")\r\n ```", "@albertvillanova \r\n>Maybe we should add an explaining note in our \"legacy\" Common Voice canonical script?\r\n\r\nYes, I agree we should have a deprecation notice in the canonical script to redirect users to the new script.", "@albertvillanova, \r\nI now get the following error after downloading my access token from the huggingface and passing it to `load_dataset` call:\r\n\r\n`AttributeError: 'DownloadManager' object has no attribute 'download_config'`\r\n\r\nAny quick pointer on how it might be resolved?", "@aasem What version of `datasets` are you using? We renamed that attribute from `_download_config` to `download_conig` fairly recently, so updating to the newest version should resolve the issue:\r\n```\r\npip install -U datasets\r\n```", "Thanks a lot, @mariosasko. That completely resolved the issue. " ]

[ "Hi @Tudyx, \r\n\r\nPlease note that in Python, the boolean NOT operator (`not`) has lower precedence than comparison operators (`<=`, `<`), thus the expression you mention is equivalent to:\r\n```python\r\n not (-1 <= example_data < self.num_classes)\r\n```\r\n\r\nAlso note that as expected, the exception is raised if:\r\n- `example_data < -1`\r\n- or `example_data >= self.num_classes`\r\n\r\nThe raise of the exception is expected when `example_data` equals 4 and `self.num_classes` equals 4 too." ]

[ "Hi @rentruewang, thansk for reporting and for your PR!!! We should definitely support this. ", "@albertvillanova Great! :)" ]

[ "Looks like I solved my problems ..." ]

[ "Hi ! Note that it does `block_size *= 2` until `block_size > len(batch)`, so it doesn't loop indefinitely. What do you mean by \"get stuck indefinitely\" then ? Is this the actual call to `paj.read_json` that hangs ?\r\n\r\n> increasing the `chunksize` argument decreases the chance of getting stuck\r\n\r\nCould you share the values of chunksize that you're using to observe this ? And maybe the order of magnitude of number of bytes per line of JSON ?", "To clarify, I don't think it loops indefinitely but the `paj.read_json` gets stuck after the first try. That's why I think it could be an issue with a lock somewhere. \r\n\r\nUsing `load_dataset(..., chunksize=40<<20)` worked without errors.", "@lhoestq I encountered another related issue. I use load_dataset() for my json data and set_transform() for preprocessing. But it hangs at the end of the epoch if `dataloader_num_workers>=1`. It appears to be working fine with num_worker=0, but it's slow.\r\n```\r\ntrain_dataset = datasets.load_dataset(\"json\", \r\n data_files=corpus_jsonl_path,\r\n keep_in_memory=False,\r\n cache_dir=model_args.cache_dir,\r\n streaming=False)\r\ntrain_dataset.set_transform(psg_parse_fn)\r\n```\r\n", "I couldn't I think your problem is unrelated to this issue @memray\r\nIndeed this issue discusses a bug when doing `load_dataset`, while your case has to do with the dataloader in a multiprocessing setup. Can you open a new issue and provide more details (share your env and what psg_parse_fn does) ?", "I also encountered a similar issue when loading a 190GB dataset of jsonl files (255 files with less than 1Gb) where it got stuck for over 20h at tables generation (fig below), increasing the `chunksize` with `load_dataset(..., chunksize=40<<20)` fixed the issue\r\n\r\n<img width=\"560\" alt=\"image\" src=\"https://user-images.githubusercontent.com/44069155/195605603-548a106e-7ad3-4269-8cdd-2ad3e975bf16.png\">\r\n", "> @lhoestq I encountered another related issue. I use load_dataset() for my json data and set_transform() for preprocessing. But it hangs at the end of the epoch if `dataloader_num_workers>=1`. It appears to be working fine with num_worker=0, but it's slow.\r\n> \r\n> ```\r\n> train_dataset = datasets.load_dataset(\"json\", \r\n> data_files=corpus_jsonl_path,\r\n> keep_in_memory=False,\r\n> cache_dir=model_args.cache_dir,\r\n> streaming=False)\r\n> train_dataset.set_transform(psg_parse_fn)\r\n> ```\r\n\r\nIn case people also get this problem, I found a way to fix it by adding `persistent_workers=True` when initializing DataLoader, like:\r\n`train_loader = DataLoader(\r\n train_dataset,\r\n batch_size=self._train_batch_size,\r\n sampler=train_sampler,\r\n collate_fn=data_collator,\r\n num_workers=self.args.dataloader_num_workers,\r\n persistent_workers=True\r\n )`\r\n\r\nThe error was `CUDA error: initialization error Exception raised from insert_events at ../c10/cuda/CUDACachingAllocator.cpp:1266` after the 1st epoch, I guess it's because the data_loader worker is killed after each epoch and the data supply is cut off. This error only occurs when num_workers>1.\r\n\r\n\r\n", "I can confirm the issue using datasets (2.12.0) with the following code and Accelerate (0.20.3) env:\r\n\r\n````\r\ntrainDataloader = DataLoader(trainSplit, batch_size=args.train_batch_size, shuffle=True)\r\nevalDataloader = DataLoader(validSplit, batch_size=args.valid_batch_size) // Here is where it gets stuck.\r\n````\r\n````\r\n- `Accelerate` version: 0.20.3\r\n- Platform: Linux-5.4.0-150-generic-x86_64-with-glibc2.29\r\n- Python version: 3.8.10\r\n- Numpy version: 1.24.3\r\n- PyTorch version (GPU?): 2.0.1+cu117 (True)\r\n- PyTorch XPU available: False\r\n- System RAM: 503.28 GB\r\n- GPU type: Tesla V100-SXM2-32GB\r\n- `Accelerate` default config:\r\n\t- compute_environment: LOCAL_MACHINE\r\n\t- distributed_type: MULTI_GPU\r\n\t- mixed_precision: fp16\r\n\t- use_cpu: False\r\n\t- num_processes: 2\r\n\t- machine_rank: 0\r\n\t- num_machines: 1\r\n\t- gpu_ids: 0,1\r\n\t- rdzv_backend: static\r\n\t- same_network: True\r\n\t- main_training_function: main\r\n\t- downcast_bf16: no\r\n\t- tpu_use_cluster: False\r\n\t- tpu_use_sudo: False\r\n\t- tpu_env: []\r\n````\r\n\r\nNotable that with Accelerate configured for one GPU only, **it doesn't get stuck.** \r\n\r\nThe suggestion made by @memray worked in my case. This is how it was applied: \r\n````\r\ntrainDataloader = DataLoader(trainSplit, batch_size=args.train_batch_size, shuffle=True, num_workers=2, persistent_workers=True)\r\nevalDataloader = DataLoader(validSplit, batch_size=args.valid_batch_size, num_workers=2, persistent_workers=True)\r\n````\r\n", "I think your issue is related to `accelerate`, feel free to open an issue there: https://github.com/huggingface/accelerate/issues\r\n\r\n`Dataset` objects generally work fine with the torch DataLoader, idk what `accelerate` does that could make it get stuck." ]

[ "Hi! Currently, we compute the final index as `index % len(dset)`. I agree this behavior is somewhat unexpected and that it would be more appropriate to raise an error instead (this is what `df.iloc` in Pandas does, for instance).\r\n\r\n@albertvillanova @lhoestq wdyt?", "I agree. I think `index % len(dset)` was used to support negative indices.\r\n\r\nI think this needs to be fixed in `datasets.formatting.formatting._check_valid_index_key` if I'm not mistaken" ]

[ "Hi @ankitk2109,\r\n\r\nHave you tried passing the split name with the keyword `split=`? See e.g. an example in our Quick Start docs: https://huggingface.co/docs/datasets/quickstart.html#load-the-dataset-and-model\r\n```python\r\n ds = load_dataset(\"big_patent\", \"d\", split=\"validation\")", "Hi @albertvillanova,\r\n\r\nThanks for your response.\r\n\r\nYes, I tried the `split='validation'` as well. But getting the same issue. ", "I'm sorry, but I can't reproduce your problem:\r\n```python\r\nIn [5]: ds = load_dataset(\"big_patent\", \"d\", split=\"validation\")\r\nDownloading and preparing dataset big_patent/d (download: 6.01 GiB, generated: 169.61 MiB, post-processed: Unknown size, total: 6.17 GiB) to .../.cache/big_patent/d/1.0.0/bdefa7c0b39fba8bba1c6331b70b738e30d63c8ad4567f983ce315a5fef6131c...\r\nDownloading data: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 6.45G/6.45G [27:36<00:00, 3.89MB/s]\r\nExtracting data files: 100%|█████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [03:18<00:00, 66.08s/it]\r\nDataset big_patent downloaded and prepared to .../.cache/big_patent/d/1.0.0/bdefa7c0b39fba8bba1c6331b70b738e30d63c8ad4567f983ce315a5fef6131c. Subsequent calls will reuse this data. \r\n\r\nIn [6]: ds\r\nOut[6]: \r\nDataset({\r\n features: ['description', 'abstract'],\r\n num_rows: 565\r\n})\r\n", "Maybe you had a connection issue while downloading the file and this was corrupted?\r\nOur cache system uses the file you downloaded first time.\r\nIf so, you could try forcing redownload of the file with:\r\n```python\r\nds = load_dataset(\"big_patent\", \"d\", split=\"validation\", download_mode=\"force_redownload\")", "I am able to download the dataset with ``` download_mode=\"force_redownload\"```. As you mentioned it was an issue with the cached version which was failed earlier due to a network issue. I am closing the issue now, once again thank you." ]

[ "Hi @adrianeboyd, thanks for reporting.\r\n\r\nThere is indeed a bug in that community dataset:\r\nLine:\r\n```python\r\nmetadata_and_text_files = list(zip(metadata_files, text_files))\r\n``` \r\nshould be replaced with\r\n```python\r\nmetadata_and_text_files = list(zip(sorted(metadata_files), sorted(text_files)))\r\n```\r\n\r\nI am going to contact their owners (https://huggingface.co/oscar-corpus) in order to inform them about the bug.\r\n\r\nI keep you informed.", "That fix is part of it, but it's clearly not the only issue.\r\n\r\nI also already contacted the OSCAR creators, but I reported it here because it looked like huggingface members were the main authors in the git history. Is there a better place to have reported this?", "Hello,\r\n\r\nWe've had an issue that could be linked to this one here: https://github.com/oscar-corpus/corpus/issues/15.\r\n\r\nI have been spot checking the source (`.txt`/`.jsonl`) files for a while, and have not found issues, especially in the start/end of corpora (but I conceed that more integration testing would be necessary on our side).\r\n\r\nThe text and metadata files are designed to be used in sync (with `lang_part_n.txt` and `lang_meta_part_n.jsonl` working together), while staying independent from part to part, so that anyone could randomly choose a part and work with it.\r\n\r\nThe fix @albertvillanova proposed should fix the problem, as the parts will be in sync again.\r\n\r\nLet me know if you need help or more details, I'd be glad to help!", "I'm happy to move the discussion to the other repo!\r\n\r\nMerely sorting the files only **maybe** fixes the processing of the first part. If the first part contains non-unix newlines, it will still be misaligned/truncated, and all the following parts will be truncated with incorrect text offsets and metadata due the offset and newline bugs.", "Fixed:\r\n- https://huggingface.co/datasets/oscar-corpus/OSCAR-2109/commit/3cd7e95aa1799b73c5ea8afc3989635f3e19b86b", "Hi @Uinelj, This is a total noobs question but how can I integrate that bugfix into my code? I reinstalled the datasets library this time from source. Should that have fixed the issue? I am still facing the misalignment issue. Do I need to download the dataset from scratch?", "Hi, I re-downloaded the dataset and still have the problem. See: https://github.com/oscar-corpus/corpus/issues/18", "Sorry @norakassner for the late reply.\r\n\r\nThere are indeed several issues creating the misalignment, as @adrianeboyd cleverly pointed out:\r\n- https://huggingface.co/datasets/oscar-corpus/OSCAR-2109/commit/3cd7e95aa1799b73c5ea8afc3989635f3e19b86b fixed one of them\r\n- but there are still others to be fixed", "Normally, the issues should be fixed now:\r\n- Fix offset initialization for each file: https://huggingface.co/datasets/oscar-corpus/OSCAR-2109/commit/1ad9b7bfe00798a9258a923b887bb1c8d732b833\r\n- Disable default universal newline support: https://huggingface.co/datasets/oscar-corpus/OSCAR-2109/commit/0c2f307d3167f03632f502af361ac6c3c393f510\r\n\r\nFeel free to reopen if you find additional misalignments/truncations.\r\n\r\nCC: @adrianeboyd @norakassner @Uinelj ", "Thanks for the updates!\r\n\r\nThe purist in me would still like to have the rstrip not strip additional characters from the original text (unicode whitespace mainly in practice, I think), but the differences are extremely small in practice and it doesn't actually matter for my current task:\r\n\r\n```python\r\ntext = \"\".join([text_f.readline() for _ in range(meta[\"nb_sentences\"])]).rstrip(\"\\n\")\r\n```" ]

[ "\r\nMy datasets version", "\r\n", "Hi! Some of our metrics require additional dependencies to work. In your case, simply installing the `seqeval` package with `pip install seqeval` should resolve the issue.", "> Hi! Some of our metrics require additional dependencies to work. In your case, simply installing the `seqeval` package with `pip install seqeval` should resolve the issue.\r\nI installed seqeval, but still reported the same error. That's too bad.\r\n", "> > Hi! Some of our metrics require additional dependencies to work. In your case, simply installing the `seqeval` package with `pip install seqeval` should resolve the issue.\r\n> > I installed seqeval, but still reported the same error. That's too bad.\r\n\r\nSame issue here. What should I do to fix this error? Please help! Thank you.", "I tried to install **seqeval** package through anaconda instead of pip:\r\n`conda install -c conda-forge seqeval`\r\nIt worked for me!", "I can run it through the following steps:\r\n\r\nThank you for answering for me!", "just change the file name seqeval.py to myseqeval.py", "Metrics are deprecated in `datasets` and `evaluate` should be used instead: https://github.com/huggingface/evaluate" ]

[ "Hi! `load_dataset` is intended to be used to load a canonical dataset (`wikipedia`), a packaged dataset (`csv`, `json`, ...) or a dataset hosted on the Hub. For local datasets saved with `save_to_disk(\"path/to/dataset\")`, use `load_from_disk(\"path/to/dataset\")`.", "Maybe we should raise an informative error message in this case...", "How should we load a locally self-gathered dataset then?" ]

[ "Hi @Zaker237, thanks for reporting.\r\n\r\nThis is weird: the error you get is only thrown if the installed pyarrow version is less than 3.0.0.\r\n\r\nCould you please check that you install pyarrow in the same Python virtual environment where you installed datasets?\r\n\r\nFrom the Python command line (or terminal) where you get the error, please type:\r\n```\r\nimport pyarrow\r\nprint(pyarrow.__version__)\r\nimport datasets\r\nprint(datasets.__version__)\r\n``` ", "hi @albertvillanova i try yesterday to create a new python environement with python 7 and try it on the environement and it worked. so i think that the error was not the package but may be jupyter notebook on conda. still yet i'm not yet sure but it worked in an environment created with venv", "OK, thanks @Zaker237 for your feedback.\r\n\r\nI close this issue then. Please, feel free to reopen it if the problem arises again." ]

[ "cc @Rocketknight1 ", "You are correct that `to_tf_dataset` only handles numerical columns right now, yes, though this is a limitation we might remove in future! The main reason we do this is that our models mostly do not include the tokenizer as a model layer, because it's very difficult to compile some of them in TF. So the \"normal\" Huggingface workflow is to first tokenize your dataset, and then pass tokenized tensors to the model.\r\n\r\nFor your use case, would you prefer to pass strings to the model, and use some text processing layers instead of the built-in tokenizers?", "Also tagging @gante just so he's aware, but I can handle this one!", "Thanks for the quick follow-up to my issue.\r\n\r\nFor my use-case, instead of the built-in tokenizers I wanted to use the `TextVectorization` layer to map from strings to integers. To achieve this, I came up with the following solution:\r\n\r\n```\r\nfrom datasets import load_dataset\r\nfrom transformers import DefaultDataCollator\r\nimport tensorflow as tf\r\nimport string\r\nimport re\r\nfrom tensorflow.keras.layers.experimental.preprocessing import TextVectorization\r\n\r\n#some hyper-parameters for the text-to-integer mapping\r\nmax_features = 20000\r\nembedding_dim = 128\r\nsequence_length = 210\r\n\r\ndata_collator = DefaultDataCollator(return_tensors=\"tf\")\r\ndataset = load_dataset(\"sst\", \"default\")\r\n\r\n#adapt the vectorization layer on train data only\r\nvectorize_layer.adapt(dataset[\"train\"].to_dict(batched=False)[\"sentence\"])\r\n\r\ndef prepare_features(text, label):\r\n text = tf.expand_dims(text, -1)\r\n return {\"vectorized_text\": vectorize_layer(text)[0], \"label\": tf.expand_dims(label, axis=-1)}\r\n\r\nencoded_dataset = dataset.map(lambda example: prepare_features(example[\"sentence\"], example[\"label\"]), batched=False)\r\n\r\n\r\ndef custom_standardization(input_data):\r\n lowercase = tf.strings.lower(input_data)\r\n return tf.strings.regex_replace(\r\n lowercase, f\"[{re.escape(string.punctuation)}]\", \"\"\r\n )\r\n\r\nvectorize_layer = TextVectorization(\r\n standardize=custom_standardization,\r\n max_tokens=max_features,\r\n output_mode=\"int\",\r\n output_sequence_length=sequence_length,\r\n)\r\n\r\ntrain_dataset = encoded_dataset[\"train\"].to_tf_dataset(columns=['vectorized_text'], label_cols=[\"label\"],\r\n shuffle=True, batch_size=1, collate_fn=data_collator).unbatch()\r\n#similar for the other sub-sets\r\n\r\n```\r\n\r\nSince the strings would have been mapped to integers or floats at some point, it's no drawback that this mapping is done early in the process. \r\n\r\nFor the future, however, it'd be more convenient to get the string data, since I am also inspecting the dataset (longest sentence, shortest sentence), which is more challenging when working with integer or float. For now, this can be done by calling `to_dict`.", "> For the future, however, it'd be more convenient to get the string data, since I am also inspecting the dataset (longest sentence, shortest sentence), which is more challenging when working with integer or float.\r\n\r\nYes, I agree, so let's keep this issue open.", "Going to close this now - methods like `to_tf_dataset` and `prepare_tf_dataset` now support string data, and have done for a while! If anyone sees this and is encountering issues with string data in those methods, please file a new issue!" ]

[ "Thanks for reporting, @SBrandeis! Some additional feature types that don't behave as expected when flattened: `Audio`, `Image` and `TranslationVariableLanguages`" ]

[ "For reference:\r\nhttps://github.com/huggingface/transformers/issues/15514\r\nhttps://github.com/huggingface/huggingface_hub/issues/650", "Hi ! For information one can set the environment variable `HF_ENDPOINT` (default is `https://huggingface.co`) if they want to use a private hub.\r\n\r\nWe may need to coordinate with the other libraries to have a consistent way of changing the hub endpoint", "Yes, I tested it successfully this morning. Thanks." ]

[ "Seems like a regression from https://github.com/huggingface/datasets/pull/2843\r\n\r\nOr maybe it's an issue with the hosting. I don't think so, though, because https://www.dropbox.com/s/aox84z90nyyuikz/discovery.zip seems to work as expected\r\n\r\n", "Hi @severo, thanks for reporting.\r\n\r\nSome servers do not support HTTP range requests, and those are required to stream some file formats (like ZIP in this case).\r\n\r\nLet me try to propose a workaround. " ]

[ "It looks like this is because of this behavior in pyarrow:\r\n```python\r\nimport pyarrow as pa\r\n\r\narr = pa.array([None, [0]])\r\nreconstructed_arr = pa.ListArray.from_arrays(arr.offsets, arr.values)\r\nprint(reconstructed_arr.to_pylist())\r\n# [[], [0]]\r\n```\r\n\r\nIt seems that `arr.offsets` can reconstruct the array properly, but an offsets array with null values can:\r\n```python\r\nfixed_offsets = pa.array([None, 0, 1])\r\nfixed_arr = pa.ListArray.from_arrays(fixed_offsets, arr.values)\r\nprint(fixed_arr.to_pylist())\r\n# [None, [0]]\r\n\r\nprint(arr.offsets.to_pylist())\r\n# [0, 0, 1]\r\nprint(fixed_offsets.to_pylist())\r\n# [None, 0, 1]\r\n```\r\nEDIT: this is because `arr.offsets` is not enough to reconstruct the array, we also need the validity bitmap", "The offsets don't have nulls because they don't include the validity bitmap from `arr.buffers()[0]`, which is used to say which values are null and which values are non-null.\r\n\r\nThough the validity bitmap also seems to be wrong:\r\n```python\r\nbin(int(arr.buffers()[0].hex(), 16))\r\n# '0b10'\r\n# it should be 0b110 - 1 corresponds to non-null and 0 corresponds to null, if you take the bits in reverse order\r\n```\r\n\r\nSo apparently I can't even create the fixed offsets array using this.\r\n\r\nIf I understand correctly it's always missing the 1 on the left, so I can add it manually as a hack to fix the issue until this is fixed in pyarrow EDIT: actually it may be more complicated than that\r\n\r\nEDIT2: actuall it's right, it corresponds to the validity bitmap of the array of logical length 2. So if we use the offsets array, the values array, and this validity bitmap it should be possible to reconstruct the array properly", "I created an issue on Apache Arrow's JIRA: https://issues.apache.org/jira/browse/ARROW-15837", "And another one: https://issues.apache.org/jira/browse/ARROW-15839", "FYI the behavior is the same with:\r\n- `datasets` version: 1.18.3\r\n- Platform: Linux-5.8.0-50-generic-x86_64-with-debian-bullseye-sid\r\n- Python version: 3.7.11\r\n- PyArrow version: 6.0.1\r\n\r\n\r\nbut not with:\r\n- `datasets` version: 1.8.0\r\n- Platform: Linux-4.18.0-305.40.2.el8_4.x86_64-x86_64-with-redhat-8.4-Ootpa\r\n- Python version: 3.7.11\r\n- PyArrow version: 3.0.0\r\n\r\ni.e. it outputs:\r\n```py\r\n0 [None, [0]]\r\n1 [None, [0]]\r\n2 [None, [0]]\r\n3 [None, [0]]\r\n```\r\n", "Thanks for the insights @PaulLerner !\r\n\r\nI found a way to workaround this issue for the code example presented in this issue.\r\n\r\nNote that empty lists will still appear when you explicitly `cast` a list of lists that contain None values like [None, [0]] to a new feature type (e.g. to change the integer precision). In this case it will show a warning that it happened. If you don't cast anything, then the None values will be kept as expected.\r\n\r\nLet me know what you think !", "Hi! I feel like I’m missing something in your answer, *what* is the workaround? Is it fixed in some `datasets` version?", "`pa.ListArray.from_arrays` returns empty lists instead of None values. The workaround I added inside `datasets` simply consists in not using `pa.ListArray.from_arrays` :)\r\n\r\nOnce this PR [here ](https://github.com/huggingface/datasets/pull/4282)is merged, we'll release a new version of `datasets` that currectly returns the None values in the case described in this issue\r\n\r\nEDIT: released :) but let's keep this issue open because it might happen again if users change the integer precision for example" ]

[ "@mariosasko Can I take this up?", "This dataset is now available here: https://huggingface.co/datasets/deepmind/code_contests." ]

[ "Yes, we decided to replace the encoded label with the corresponding label when possible in the dataset viewer. But\r\n1. maybe it's the wrong default\r\n2. we could find a way to show both (with a switch, or showing both ie. `0 (neutral)`).\r\n", "Hi @severo,\r\n\r\nThanks for clarifying. \r\n\r\nI think this default is a bit counterintuitive for the user. However, this is a personal opinion that might not be general. I think it is nice to have the actual (non-encoded) labels in the viewer. On the other hand, it would be nice to match what the user sees with what they get when they download a dataset. I don't know - I can see the difficulty of choosing a default :)\r\nMaybe having non-encoded labels as a default can be useful?\r\n\r\nAnyway, I think the issue has been addressed. Thanks a lot for your super-quick answer!\r\n\r\n ", "Thanks for the 👍 in https://github.com/huggingface/datasets/issues/3673#issuecomment-1029008349 @mariosasko @gary149 @pietrolesci, but as I proposed various solutions, it's not clear to me which you prefer. Could you write your preferences as a comment?\r\n\r\n_(note for myself: one idea per comment in the future)_", "As I am working with seq2seq, I prefer having the label in string form rather than numeric. So the viewer is fine and the underlying dataset should be \"decoded\" (from int to str). In this way, the user does not have to search for a mapping `int -> original name` (even though is trivial to find, I reckon). Also, encoding labels is rather easy.\r\n\r\nI hope this is useful", "I like the idea of \"0 (neutral)\". The label name can even be greyed to make it clear that it's not part of the actual item in the dataset, it's just the meaning.", "I like @lhoestq's idea of having grayed-out labels.", "Proposals by @gary149. Which one do you prefer? Please vote with the thumbs\r\n\r\n- 👍 \r\n\r\n \r\n\r\n- 👎 \r\n\r\n \r\n\r\n", "I like Option 1 better as it shows clearly what the user is downloading", "Thanks! ", "It's [live](https://huggingface.co/datasets/glue/viewer/cola/train):\r\n\r\n<img width=\"1126\" alt=\"Capture d’écran 2022-02-14 à 10 26 03\" src=\"https://user-images.githubusercontent.com/1676121/153836716-25f6205b-96af-42d8-880a-7c09cb24c420.png\">\r\n\r\nThanks all for the help to improve the UI!", "Love it ! thanks :)" ]

[]

[ "Hi ! I wasn't able to reproduce the error, are you still experiencing this ? I tried calling `cast_column` on a string column containing paths.\r\n\r\nIf you manage to share a reproducible code example that would be perfect", "Hi,\r\n\r\nI think my team mate got this solved. Clolsing it for now and will reopen if I experience this again.\r\nThanks :) ", "Hi @R4ZZ3,\r\n\r\nIf it is not too much of a bother, can you please help me how to resolve this error? I am exactly getting the same error where I am going as per the documentation guideline:\r\n\r\n`my_audio_dataset = my_audio_dataset.cast_column(\"audio_paths\", Audio())`\r\n\r\nwhere `\"audio_paths\"` is a dataset column (feature) having strings of absolute paths to mp3 files of the dataset.\r\n\r\n", "I was having the same issue with this code:\r\n\r\n```\r\ndataset = dataset.map(\r\n lambda batch: {\"full_path\" : os.path.join(self.data_path, batch[\"path\"])},\r\n num_procs = 4\r\n)\r\nmy_audio_dataset = dataset.cast_column(\"full_path\", Audio(sampling_rate=16_000))\r\n```\r\n\r\nRemoving the \"num_procs\" argument fixed it somehow.\r\nUsing a mac with m1 chip", "Hi @Hubert-Bonisseur, I think this will be fixed by https://github.com/huggingface/datasets/pull/4614" ]

[ "Having talked to @lhoestq, I see that this feature is no longer supported. \r\n\r\nI really don't think this was a good idea. It is a major breaking change and one for which we don't even have a working solution at the moment, which is bad for PyTorch as we don't want to force people to have `datasets` decode audio files automatically, but **really** bad for Tensorflow and Flax where we **currently cannot** even use `datasets` to load `.mp3` files - e.g. `common_voice` doesn't work anymore in a TF training script. Note this worked perfectly fine before making the change (think it was done [here](https://github.com/huggingface/datasets/pull/3290) no?)\r\n\r\nIMO, it's really important to think about a solution here and I strongly favor to make a difference here between loading a dataset in streaming mode and in non-streaming mode, so that in non-streaming mode the actual downloaded file is displayed. It's really crucial for people to be able to analyse the original files IMO when the dataset is not downloaded in streaming mode. \r\n\r\nThere are the following reasons why it is paramount to have access to the **original** audio file in my opinion (in non-streaming mode):\r\n- There are a wide variety of different libraries to load audio data with varying support on different platforms. For me it was quite clear that there is simply to single good library to load audio files for all platforms - so we have to leave the option to the user to decide which loading to use.\r\n- We had support for audio datasets a long time before streaming audio was possible. There were quite some versions where we advertised **everywhere** to load the audio from the path name (and there are many places where we still do even though it's not possible anymore). To give some examples:\r\n - Official example of TF Wav2Vec2: https://github.com/huggingface/transformers/blob/f427e750490b486944cc9be3c99834ad5cf78b57/src/transformers/models/wav2vec2/modeling_tf_wav2vec2.py#L1423 Wav2Vec2 is as important for speech as BERT is for NLP - so it's **very** important. The official example currently doesn't work and we don't even have a workaround for it for MP3 files at the moment. Same goes for Flax.\r\n - The most downloaded non-nlp checkpoint: https://huggingface.co/facebook/wav2vec2-base-960h#usage has a usage example which doesn't work anymore with the current datasets implementation. I'll update this now, but we have >1000 wav2vec2 checkpoints on the Hub and we can't update all the model cards.\r\n => This is a big breaking change with no current solution. For `transformers` breaking changes are one of the biggest complaints.\r\n- Similar to this we also shouldn't assume that there is only one resampling method for Audio. I think it's good to have one offered automatically by `datasets`, but we have to leave the user the freedom to choose her/his own resampling as well. Resampling can take very different filtering windows and other parameters which are currently somewhat hardcoded in `datasets`, which users might very well want to change.\r\n\r\n\r\n=> IMO, it's a **very** big priority to again have the correct absolute path in non-streaming mode. The other solution of providing a path-like object derived from the bytes stocked in the `.array` file is not nearly as user-friendly, but better than nothing. ", "Agree that we need to have access to the original sound files. Few days ago I was looking for these original files because I suspected there is bug in the audio resampling (confirmed in https://github.com/huggingface/datasets/issues/3662) and I want to do my own resampling to workaround the bug, which is now not possible anymore due to the unavailability of the original files.", "@patrickvonplaten \r\n> The other solution of providing a path-like object derived from the bytes stocked in the .array file is not nearly as user-friendly, but better than nothing\r\n\r\nJust to clarify, here you describe the approach that uses the `Audio.decode` attribute to access the underlying bytes?\r\n\r\n> The official example currently doesn't work and we don't even have a workaround for it for MP3 files at the moment\r\n\r\nI'd assume this is because we use `sox_io` as a backend for decoding. However, soon we should be able to use `soundfile`, which supports path-like objects, for MP3 (https://github.com/huggingface/datasets/pull/3667#issuecomment-1030090627).\r\n\r\nYour concern is reasonable, but there are situations where we can only serve bytes (see https://github.com/huggingface/datasets/pull/3685 for instance). IMO it makes sense to fix the affected datasets for now, but I don't think we should care too much whether we rely on local paths or bytes after soundfile adds support for MP3 as long as our examples work (shouldn't be too hard to update the `map_to_array` functions) and we properly document how to access the underlying path/bytes for custom decoding (via `ds.cast_column(\"audio\", Audio(decode=False))`).\r\n", "Related to this discussion: in https://github.com/huggingface/datasets/pull/3664#issuecomment-1031866858 I propose how we could change `iter_archive` to work for streaming and also return local paths (as it used too !). I'd love your opinions on this", "> @patrickvonplaten\r\n> \r\n> > The other solution of providing a path-like object derived from the bytes stocked in the .array file is not nearly as user-friendly, but better than nothing\r\n> \r\n> Just to clarify, here you describe the approach that uses the `Audio.decode` attribute to access the underlying bytes?\r\n\r\nYes! \r\n\r\n> \r\n> > The official example currently doesn't work and we don't even have a workaround for it for MP3 files at the moment\r\n> \r\n> I'd assume this is because we use `sox_io` as a backend for decoding. However, soon we should be able to use `soundfile`, which supports path-like objects, for MP3 ([#3667 (comment)](https://github.com/huggingface/datasets/pull/3667#issuecomment-1030090627)). \r\n> Your concern is reasonable, but there are situations where we can only serve bytes (see #3685 for instance). IMO it makes sense to fix the affected datasets for now, but I don't think we should care too much whether we rely on local paths or bytes after soundfile adds support for MP3 as long as our examples work (shouldn't be too hard to update the `map_to_array` functions) and we properly document how to access the underlying path/bytes for custom decoding (via `ds.cast_column(\"audio\", Audio(decode=False))`).\r\n\r\nYes this might be, but I highly doubt that `soundfile` is the go-to library for audio then. @anton-l and I have tried out a bunch of different audio loading libraries (`soundfile`, `librosa`, `torchaudio`, pure `ffmpeg`, `audioread`, ...). One thing that was pretty clear to me is that there is just no \"de-facto standard\" library and they all have pros and cons. None of the libraries really supports \"batch\"-ed audio loading. Some depend on PyTorch. `torchaudio` is 100x faster (really!) than `librosa's` fallback on MP3. `torchaudio` often has problems with multi-proessing, ... Also we should keep in mind that resampling is similarly not as simple as reading a text file. It's a pretty complex signal processing transform and people very well might want to use special filters, etc...at the moment we just hard-code `torchaudio's` or `librosa's` default filter when doing resampling.\r\n\r\n=> All this to say that we **should definitely** care about whether we rely on local paths or bytes IMO. We don't want to loose all users that are forced to use `datasets` decoding or resampling or have to built a very much not intuitive way of loading bytes into a numpy array. It's much more intuitive to be able to inspect a local file. I feel pretty strongly about this and am happy to also jump on a call. Keeping libraries flexible and lean as well as exposing internals is very important IMO (this philosophy has worked quite well so far with Transformers).\r\n\r\n", "Thanks a lot for the very detailed explanation. Now everything makes much more sense.", "From https://github.com/huggingface/datasets/pull/3736 the Common Voice dataset now gives access to the local audio files as before", "I understand the argument that it is bad to have a breaking change. How to deal with the introduction of breaking changes is a topic of its own and not sure how you want to deal with that (or is the policy this is never allowed, and there must be a `load_dataset_v2` or so if you really want to introduce a breaking change?).\r\n\r\nRegardless of whether it is a breaking change, however, I don't see the other arguments.\r\n\r\n> but **really** bad for Tensorflow and Flax where we **currently cannot** even use `datasets` to load `.mp3` files\r\n\r\nI don't exactly understand this. Why not?\r\n\r\nWhy does the HF dataset on-the-fly decoding mechanism not work? Why is it anyway specific to PyTorch or TensorFlow? Isn't this independent?\r\n\r\nBut even if you just provide the raw bytes to TF, on TF you could just use sth like `tfio.audio.decode_mp3` or `tf.audio.decode_ogg` or `tfio.audio.decode_flac`?\r\n\r\n> There are the following reasons why it is paramount to have access to the original audio file in my opinion ...\r\n\r\nI don't really understand the arguments (despite that it maybe breaks existing code). You anyway have the original audio files but it is just embedded in the dataset? I don't really know about any library which cannot also load the audio from memory (i.e. from the dataset).\r\n\r\nBtw, on librosa being slow for decoding audio files, I saw that as well, so we have this comment RETURNN:\r\n\r\n> Don't use librosa.load which internally uses audioread which would use Gstreamer as a backend which has multiple issues:\r\n> https://github.com/beetbox/audioread/issues/62\r\n> https://github.com/beetbox/audioread/issues/63\r\n> Instead, use PySoundFile (soundfile), which is also faster. See here for discussions:\r\n> https://github.com/beetbox/audioread/issues/64\r\n> https://github.com/librosa/librosa/issues/681\r\n\r\nResampling is also a separate aspect, which is also less straightforward and with different compromises between speed and quality. So there the different tradeoffs and different implementations can make a difference.\r\n\r\nHowever, I don't see how this is related to the question whether there should be the raw bytes inside the dataset or as separate local files.\r\n", "Thanks for your comments here @albertz - cool to get your input! \r\n\r\nAnswering a bit here between the lines:\r\n\r\n> I understand the argument that it is bad to have a breaking change. How to deal with the introduction of breaking changes is a topic of its own and not sure how you want to deal with that (or is the policy this is never allowed, and there must be a `load_dataset_v2` or so if you really want to introduce a breaking change?).\r\n> \r\n> Regardless of whether it is a breaking change, however, I don't see the other arguments.\r\n> \r\n> > but **really** bad for Tensorflow and Flax where we **currently cannot** even use `datasets` to load `.mp3` files\r\n> \r\n> I don't exactly understand this. Why not?\r\n\r\n> Why does the HF dataset on-the-fly decoding mechanism not work? Why is it anyway specific to PyTorch or TensorFlow? Isn't this independent?\r\n\r\nThe problem with decoding on the fly is that we currently rely on `torchaudio` for this now which relies on `torch` which is not necessarily something people would like to install when using `tensorflow` or `flax`. Therefore we cannot just rely on people using the decoding on the fly method. We just didn't find a library that is ML framework independent and fast enough for all formats. `torchaudio` is currently in our opinion by far the best here.\r\n\r\nSo for TF and Flax it's important that users can load audio files or bytes they way the want to - this might become less important if we find (or make) a good library with few dependencies that is fast for all kinds of platforms / use cases.\r\n\r\n\r\nNow the question is whether it's better to store audio data as a path to a file or as raw bytes I guess.\\\r\nMy main arguments for storing the audio data as a path to a file is pretty much all about users experience - I don't really expect our users to understand the inner workings of datasets:\r\n\r\n- 1. It's not straightforward to know which function to use to decode it - not all `load_audio(...)` or `read_audio(...)` work on raw bytes. E.g. Looking at https://pytorch.org/audio/stable/torchaudio.html?highlight=load#torchaudio.load one would not see directly how to load raw bytes . There are also some functions of other libraries which only work on files which would require the user to save the bytes as a file first before being able to load it.\r\n- 2. It's difficult to see which format the bytes are coming from (mp3, ogg, ...) - guess this could be remedied by adding the format to each sample though\r\n- 3. It is a bit scary IMO to see raw bytes for users. Overall, I think it's better to leave the data in it's raw form as this way it's much easier for people to play around with the audio files, less need to read docs because people don't worry about what happened to the audio files (are the bytes already resampled?)\r\n\r\nBut the argument that the audio should be loadable directly from memory is good - haven't thought about this too much. \r\nI guess it's still very much possible for the user to do this:\r\n\r\n```python\r\ndef save_as_bytes:\r\n batch[\"bytes\"] = read_in_bytes_from_file(batch[\"file\"])\\\r\n os.remove(batch[\"file\"])\r\n\r\nds = ds.map(save_as_bytes)\r\n\r\nds.save_to_disk(...)\r\n```\r\n\r\nGuess the question is more a bit about what should be the default case?", "> The problem with decoding on the fly is that we currently rely on `torchaudio` for this now which relies on `torch` which is not necessarily something people would like to install when using `tensorflow` or `flax`. Therefore we cannot just rely on people using the decoding on the fly method. We just didn't find a library that is ML framework independent and fast enough for all formats. `torchaudio` is currently in our opinion by far the best here.\r\n\r\nBut how is this relevant for this issue here? I thought this issue here is about having the (correct) path in the dataset or having raw bytes in the dataset.\r\n\r\nHow did TF users use it at all then? Or they just do not use on-the-fly decoding? I did not even notice this problem (maybe because I had `torchaudio` installed). But what do they use instead?\r\n\r\nBut as I outlined before, they could just use `tfio.audio.decode_flac` and co, where it would be more natural if you already provide the raw bytes.\r\n\r\n> Looking at https://pytorch.org/audio/stable/torchaudio.html?highlight=load#torchaudio.load one would not see directly how to load raw bytes\r\n\r\nI was not really familiar with `torchaudio`. It seems that they really don't provide an easy/direct API to operate on raw bytes. Which is very strange and unfortunate because as far as I can see, all the underlying backend libraries (e.g. soundfile) easily allow that. So I would say that this is the fault of `torchaudio` then. But despite, if you anyway use `torchaudio` with `soundfile` backend, why not just use `soundfile` directly. It's very simple to use and crossplatform.\r\n\r\nBut ok, now we are just discussing how to handle the on-the-fly decoding. I still think this is a separate issue and having raw bytes in the dataset instead of local files should just be fine as well.\r\n\r\n\r\n> It is a bit scary IMO to see raw bytes for users.\r\n\r\nI think nobody who writes code is scared by seeing the raw bytes content of a binary file. :)\r\n\r\n\r\n> I guess it's still very much possible for the user to do this:\r\n> \r\n> ```python\r\n> def save_as_bytes:\r\n> batch[\"bytes\"] = read_in_bytes_from_file(batch[\"file\"])\\\r\n> os.remove(batch[\"file\"])\r\n> \r\n> ds = ds.map(save_as_bytes)\r\n> \r\n> ds.save_to_disk(...)\r\n> ```\r\n\r\nIn https://github.com/huggingface/datasets/pull/4184#issuecomment-1105191639, you said/proposed that this `map` is not needed anymore and `save_to_disk` could do it automatically (maybe via some option)?\r\n\r\n> Guess the question is more a bit about what should be the default case?\r\n\r\nYea this is up to you. I'm happy as long as we can get it the way we want easily and this is a well supported use case. :)\r\n", "> In https://github.com/huggingface/datasets/pull/4184#issuecomment-1105191639, you said/proposed that this map is not needed anymore and save_to_disk could do it automatically (maybe via some option)?\r\n\r\nYes! Should be super easy now see discussion here: https://github.com/rwth-i6/i6_core/issues/257#issuecomment-1105494468\r\n\r\nThanks for the super useful input :-)", "Despite the comments that this has been fixed, I am finding the exact same problem is occurring again (with datasets version 2.3.2)", "> Despite the comments that this has been fixed, I am finding the exact same problem is occurring again (with datasets version 2.3.2)\r\n\r\nIt appears downgrading to torchaudio 0.11.0 fixed this problem.", "@DCNemesis, sorry which problem exactly is occuring again? Also cc @lhoestq @polinaeterna here", "@patrickvonplaten @lhoestq @polinaeterna I was unable to load audio from Common Voice using 🤗 with the current version of torchaudio, but downgrading to torchaudio 0.11.0 fixed it. This is probably more of a torch problem than a Hugging Face problem.", "@DCNemesis that's interesting, could you please share the error message if you still can access it? ", "@polinaeterna I believe it is the same exact error as above. It occurs on other .mp3 sources as well, but the problem is with torchaudio > 0.11.0. I've created a short colab notebook that reproduces the error, and the fix here: https://colab.research.google.com/drive/18wsuwdHwBPN3JkcnhEtk8MUYqF9swuWZ?usp=sharing", "Hi @DCNemesis,\r\n\r\nYour issue was slightly different from the original one in this issue page. Yours seems related to a change in the backend used by `torchaudio` (`ffmpeg` instead of `sox`). Refer to the issue page here:\r\n- #4776\r\n\r\nNormally, it should be circumvented with the patch made by @polinaeterna in:\r\n- #4923", "I think the original issue reported here was already fixed by:\r\n- #3736\r\n\r\nOtherwise, feel free to reopen." ]

[ "Thanks @lhoestq for finding the reason of incorrect resampling. This issue affects all languages which have sound files with different sampling rates such as Turkish and Luganda.", "@cahya-wirawan - do you know how many languages have different sampling rates in Common Voice? I'm quite surprised to see this for multiple languages actually", "@cahya-wirawan, I can reproduce the problem for Common Voice 7 for Turkish. Here a script you can use:\r\n\r\n\r\n```python\r\n#!/usr/bin/env python3\r\nfrom datasets import load_dataset\r\nimport torchaudio\r\nfrom io import BytesIO\r\nfrom datasets import Audio\r\nfrom collections import Counter\r\nimport sys\r\n\r\nds_name = str(sys.argv[1])\r\nlang = str(sys.argv[2])\r\n\r\nds = load_dataset(ds_name, lang, split=\"train\", use_auth_token=True)\r\nds = ds.cast_column(\"audio\", Audio(decode=False))\r\n\r\nall_sampling_rates = []\r\n\r\n\r\ndef print_sampling_rate(x):\r\n x, sr = torchaudio.load(BytesIO(x[\"audio\"][\"bytes\"]), format=\"mp3\")\r\n all_sampling_rates.append(sr)\r\n\r\nds.map(print_sampling_rate)\r\n\r\n\r\nprint(Counter(all_sampling_rates))\r\n```\r\n\r\ncan be run with:\r\n\r\n```bash\r\npython run.py mozilla-foundation/common_voice_7_0 tr\r\n```\r\n\r\nFor CV 6.1 all samples seem to have the same audio", "It actually shows that many more samples are in 32kHz format than it 48kHz which is unexpected. Thanks a lot for flagging! Will contact Common Voice about this as well", "I only checked the CV 7.0 for Turkish, Luganda and Indonesian, they have audio files with difference sampling rates, and all of them are affected by this issue. Percentage of incorrect resampling as follow, Turkish: 9.1%, Luganda: 88.2% and Indonesian: 64.1%.\r\nI checked it using the original CV files. I check the original sampling rates and the length of audio array of each files and compare it with the length of audio array (and the sampling rate which is always 48kHz) from mozilla-foundation/common_voice_7_0 datasets. if the length of audio array from dataset is not equal to 48kHz/original sampling rate * length of audio array of the original audio file then it is affected,", "Ok wow, thanks a lot for checking this - you've found a pretty big bug :sweat_smile: It seems like **a lot** more datasets are actually affected than I original thought. We'll try to solve this as soon as possible and make an announcement tomorrow." ]

[ "Hi @thomas-happify, please note that the preview may take some time before rendering the data.\r\n\r\nI've seen it is already working.\r\n\r\nI close this issue. Please feel free to reopen it if the problem arises again." ]

[ "The error is now:\r\n\r\n```\r\nStatus code: 400\r\nException: Status400Error\r\nMessage: this dataset is not supported for now.\r\n```\r\n\r\nWe've disabled the dataset viewer for several big datasets like this one. We hope being able to reenable it soon.", "The list of splits cannot be obtained. cc @huggingface/datasets ", "```\r\nError code: SplitsNamesError\r\nException: SplitsNotFoundError\r\nMessage: The split names could not be parsed from the dataset config.\r\nTraceback: Traceback (most recent call last):\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/inspect.py\", line 354, in get_dataset_config_info\r\n for split_generator in builder._split_generators(\r\n File \"/tmp/modules-cache/datasets_modules/datasets/bigscience--P3/12c0badfecad4564ecb8a6f81b5d0559656f269f08b13c59c93283f3a84134ba/P3.py\", line 154, in _split_generators\r\n data_dir = dl_manager.download_and_extract(_URLs)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/download/streaming_download_manager.py\", line 944, in download_and_extract\r\n return self.extract(self.download(url_or_urls))\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/download/streaming_download_manager.py\", line 907, in extract\r\n urlpaths = map_nested(self._extract, path_or_paths, map_tuple=True)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 393, in map_nested\r\n mapped = [\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 394, in <listcomp>\r\n _single_map_nested((function, obj, types, None, True, None))\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 346, in _single_map_nested\r\n return {k: _single_map_nested((function, v, types, None, True, None)) for k, v in pbar}\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 346, in <dictcomp>\r\n return {k: _single_map_nested((function, v, types, None, True, None)) for k, v in pbar}\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 346, in _single_map_nested\r\n return {k: _single_map_nested((function, v, types, None, True, None)) for k, v in pbar}\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 346, in <dictcomp>\r\n return {k: _single_map_nested((function, v, types, None, True, None)) for k, v in pbar}\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 330, in _single_map_nested\r\n return function(data_struct)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/download/streaming_download_manager.py\", line 912, in _extract\r\n protocol = _get_extraction_protocol(urlpath, use_auth_token=self.download_config.use_auth_token)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/download/streaming_download_manager.py\", line 402, in _get_extraction_protocol\r\n return _get_extraction_protocol_with_magic_number(f)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/download/streaming_download_manager.py\", line 367, in _get_extraction_protocol_with_magic_number\r\n magic_number = f.read(MAGIC_NUMBER_MAX_LENGTH)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/implementations/http.py\", line 574, in read\r\n return super().read(length)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/spec.py\", line 1575, in read\r\n out = self.cache._fetch(self.loc, self.loc + length)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/caching.py\", line 377, in _fetch\r\n self.cache = self.fetcher(start, bend)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/asyn.py\", line 111, in wrapper\r\n return sync(self.loop, func, *args, **kwargs)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/asyn.py\", line 96, in sync\r\n raise return_result\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/asyn.py\", line 53, in _runner\r\n result[0] = await coro\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/fsspec/implementations/http.py\", line 616, in async_fetch_range\r\n out = await r.read()\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/aiohttp/client_reqrep.py\", line 1036, in read\r\n self._body = await self.content.read()\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/aiohttp/streams.py\", line 375, in read\r\n block = await self.readany()\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/aiohttp/streams.py\", line 397, in readany\r\n await self._wait(\"readany\")\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/aiohttp/streams.py\", line 304, in _wait\r\n await waiter\r\n aiohttp.client_exceptions.ClientPayloadError: Response payload is not completed\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/responses/splits.py\", line 75, in get_splits_response\r\n split_full_names = get_dataset_split_full_names(dataset, hf_token)\r\n File \"/src/services/worker/src/worker/responses/splits.py\", line 35, in get_dataset_split_full_names\r\n return [\r\n File \"/src/services/worker/src/worker/responses/splits.py\", line 38, in <listcomp>\r\n for split in get_dataset_split_names(dataset, config, use_auth_token=hf_token)\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/inspect.py\", line 404, in get_dataset_split_names\r\n info = get_dataset_config_info(\r\n File \"/src/services/worker/.venv/lib/python3.9/site-packages/datasets/inspect.py\", line 359, in get_dataset_config_info\r\n raise SplitsNotFoundError(\"The split names could not be parsed from the dataset config.\") from err\r\n datasets.inspect.SplitsNotFoundError: The split names could not be parsed from the dataset config.\r\n```", "Closing in favor of https://huggingface.co/datasets/bigscience/P3/discussions/6 and https://github.com/huggingface/datasets-server/issues/1689" ]

[ "Hi @RensDimmendaal, \r\n\r\nI'm sorry but I can't reproduce your bug:\r\n```python\r\nIn [1]: from datasets import load_dataset\r\n ...: ds = load_dataset(\"subjqa\", \"electronics\")\r\nDownloading builder script: 9.15kB [00:00, 4.10MB/s] \r\nDownloading metadata: 17.7kB [00:00, 8.51MB/s] \r\nDownloading and preparing dataset subjqa/electronics (download: 10.86 MiB, generated: 3.01 MiB, post-processed: Unknown size, total: 13.86 MiB) to .../.cache/huggingface/datasets/subjqa/electronics/1.1.0/e5588f9298ff2d70686a00cc377e4bdccf4e32287459e3c6baf2dc5ab57fe7fd...\r\nDownloading data: 11.4MB [00:03, 3.50MB/s]\r\nDataset subjqa downloaded and prepared to .../.cache/huggingface/datasets/subjqa/electronics/1.1.0/e5588f9298ff2d70686a00cc377e4bdccf4e32287459e3c6baf2dc5ab57fe7fd. Subsequent calls will reuse this data.\r\n100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 3/3 [00:00<00:00, 605.09it/s]\r\n\r\nIn [2]: ds\r\nOut[2]: \r\nDatasetDict({\r\n train: Dataset({\r\n features: ['domain', 'nn_mod', 'nn_asp', 'query_mod', 'query_asp', 'q_reviews_id', 'question_subj_level', 'ques_subj_score', 'is_ques_subjective', 'review_id', 'id', 'title', 'context', 'question', 'answers'],\r\n num_rows: 1295\r\n })\r\n test: Dataset({\r\n features: ['domain', 'nn_mod', 'nn_asp', 'query_mod', 'query_asp', 'q_reviews_id', 'question_subj_level', 'ques_subj_score', 'is_ques_subjective', 'review_id', 'id', 'title', 'context', 'question', 'answers'],\r\n num_rows: 358\r\n })\r\n validation: Dataset({\r\n features: ['domain', 'nn_mod', 'nn_asp', 'query_mod', 'query_asp', 'q_reviews_id', 'question_subj_level', 'ques_subj_score', 'is_ques_subjective', 'review_id', 'id', 'title', 'context', 'question', 'answers'],\r\n num_rows: 255\r\n })\r\n})\r\n```\r\n\r\nCould you please try again and see if the problem persists?\r\n\r\nIf that is the case, you can circumvent the issue by passing `ignore_verifications`:\r\n```python\r\nds = load_dataset(\"subjqa\", \"electronics\", ignore_verifications=True)", "Thanks checking!\r\n\r\nYou're totally right. I don't know what's changed, but I'm glad it's working now!\r\n\r\n" ]

[ "Hi @abhi-mosaic, thanks for reporting.\r\n\r\nI'm looking at it... ", "also hitting this issue", "Hey @albertvillanova, sorry to reopen this... I can confirm that on `master` branch the dataset is downloadable now but it is still broken in streaming mode:\r\n\r\n```python\r\n >>> import datasets\r\n >>> pubmed_train = datasets.load_dataset('pubmed', split='train', streaming=True)\r\n >>> next(iter(pubmed_train))\r\n```\r\n```\r\n No such file or directory: 'gzip://pubmed22n0001.xml::ftp://ftp.ncbi.nlm.nih.gov/pubmed/baseline/pubmed22n0001.xml.gz'\r\n```\r\n", "Hi @abhi-mosaic, would you mind opening another issue for this new problem?\r\n\r\nFirst issue (already solved) was a ConnectionError due to the yearly update release of PubMed: we fixed it by updating the URLs from year 2021 to year 2022.\r\n\r\nHowever this is another problem: to make pubmed streamable. Please note that NOT all our datastes are streamable: we are making streamable more and more of them... but this is an on-going process...\r\n\r\nThanks.", "@albertvillanova \r\nWhen I tried below codes, I got the similar error\r\n\r\n```\r\n\r\ndataset=load_dataset(\"pubmed\",split=\"train\")\r\n\r\nCouldn't reach ftp://ftp.ncbi.nlm.nih.gov/pubmed/baseline/pubmed21n0601.xml.gz\r\n```", "@y-rok you need to update `datasets`:\r\n```shell\r\npip install -U datasets\r\n```" ]

[]

[]

[]

[ "Hi ! At the moment you can use `to_pandas()` to get a pandas DataFrame that supports `group_by` operations (make sure your dataset fits in memory though)\r\n\r\nWe use Arrow as a back-end for `datasets` and it doesn't have native group by (see https://github.com/apache/arrow/issues/2189) unfortunately.\r\n\r\nI just drafted what it could look like to have `group_by` in `datasets`:\r\n```python\r\nfrom datasets import concatenate_datasets\r\n\r\ndef group_by(d, col, join): \r\n \"\"\"from: https://github.com/huggingface/datasets/issues/3644\"\"\"\r\n # Get the indices of each group\r\n groups = {key: [] for key in d.unique(col)} \r\n def create_groups_indices(key, i): \r\n groups[key].append(i) \r\n d.map(create_groups_indices, with_indices=True, input_columns=col) \r\n # Get one dataset object per group\r\n groups = {key: d.select(indices) for key, indices in groups.items()} \r\n # Apply join function\r\n groups = {\r\n key: dataset_group.map(join, batched=True, batch_size=len(dataset_group), remove_columns=d.column_names)\r\n for key, dataset_group in groups.items()\r\n } \r\n # Return concatenation of all the joined groups\r\n return concatenate_datasets(groups.values())\r\n```\r\n\r\nexample of usage:\r\n```python\r\n\r\ndef join(batch): \r\n # take the batch of all the examples of a group, and return a batch with one aggregated example\r\n # (we could aggregate examples into several rows instead of one, if you want)\r\n return {\"total\": [batch[\"i\"]]} \r\n\r\nd = Dataset.from_dict({\r\n \"i\": [i for i in range(50)],\r\n \"group_key\": [i % 4 for i in range(50)],\r\n})\r\nprint(group_by(d, \"group_key\", join))\r\n# total\r\n# 0 [0, 4, 8, 12, 16, 20, 24, 28, 32, 36, 40, 44, 48]\r\n# 1 [1, 5, 9, 13, 17, 21, 25, 29, 33, 37, 41, 45, 49]\r\n# 2 [2, 6, 10, 14, 18, 22, 26, 30, 34, 38, 42, 46]\r\n# 3 [3, 7, 11, 15, 19, 23, 27, 31, 35, 39, 43, 47]\r\n```\r\n\r\nLet me know if that helps !\r\n\r\ncc @albertvillanova @mariosasko for visibility", "@lhoestq As of PyArrow 7.0.0, `pa.Table` has the [`group_by` method](https://arrow.apache.org/docs/python/generated/pyarrow.Table.html#pyarrow.Table.group_by), so we should also consider using that function for grouping. ", "Any update on this?", "You can use https://github.com/mariosasko/datasets_sql by @mariosasko to go group by operations using SQL queries", "Hi, I have a similar issue as OP but the suggested solutions do not work for my case. Basically, I process documents through a model to extract the last_hidden_state, using the \"map\" method on a Dataset object, but would like to average the result over a categorical column at the end (i.e. groupby this column).\r\n- A to_pandas() saturates the memory, although it gives me the desired result through a .groupby().apply(np.mean, axis=0) on a smaller use-case,\r\n- The solution posted on Feb 4 is much too slow,\r\n- datasets_sql seems to not like the fact that I'm averaging np.arrays.\r\nSo I'm kinda out of \"non brute force\" options... Any help appreciated", "> Hi, I have a similar issue as OP but the suggested solutions do not work for my case. Basically, I process documents through a model to extract the last_hidden_state, using the \"map\" method on a Dataset object, but would like to average the result over a categorical column at the end (i.e. groupby this column).\r\n \r\nIf you haven't yet, you could explore using [Polars](https://www.pola.rs/) for this. It's a new DataFrame library written in Rust with Python bindings. It is Pandas like it in many ways ,but does have some biggish differences in syntax/approach so it's definitely not a drop-in replacement. \r\n\r\nPolar's also uses Arrow as a backend but also supports out-of-memory operations; in this case, it's probably easiest to write out your dataset to parquet and then use the polar's `scan_parquet` method (this will lazily read from the parquet file). The thing you get back from that is a `LazyDataFrame` i.e. nothing is loaded into memory until you specify a query and call a `collect` method. \r\n\r\nExample below of doing a groupby on a dataset which definitely wouldn't fit into memory on my machine:\r\n\r\n```\r\nfrom datasets import load_dataset\r\nimport polars as pl\r\n\r\nds = load_dataset(\"blbooks\")\r\nds['train'].to_parquet(\"test.parquet\")\r\ndf = pl.scan_parquet(\"test.parquet\")\r\ndf.groupby('date').agg([pl.count()]).collect()\r\n```\r\n\r\n>datasets_sql seems to not like the fact that I'm averaging np.arrays.\r\n\r\nI am not certain how Polars will handle this either. It does have NumPy support (https://pola-rs.github.io/polars-book/user-guide/howcani/interop/numpy.html) but I assume Polars will need to have at least enough memory in each group you want to average over so you may still end up needing more memory depending on the size of your dataset/groups. \r\n\r\n\r\n", "Hi @davanstrien , thanks a lot, I didn't know about this library and the answer works! I need to try it on the full dataset now, but I'm hopeful. Here's what my code looks like:\r\n```\r\nlist_size = 768\r\ndf.groupby(\"date\").agg(\r\n pl.concat_list(\r\n [\r\n pl.col(\"hidden_state\")\r\n .arr.slice(n, 1)\r\n .arr.first()\r\n .mean()\r\n for n in range(0, list_size)\r\n ]\r\n ).collect()\r\n```\r\n\r\nFor some reasons, the following code was giving me a \"mean() got unexpected argument 'axis'\":\r\n```\r\ndf2 = df.groupby('date').agg(\r\n pl.col(\"hidden_state\").map(np.mean).alias(\"average_hidden_state\")\r\n).collect()\r\n\r\n```\r\n\r\nEDIT: The solution works on my large dataset, the memory does not crash and the time is reasonable, thanks a lot again!", "@jeremylhour glad this worked for you :) ", "I find this functionality missing in my workflow as well and the workarounds with SQL and Polars unsatisfying. Since PyArrow has exposed this functionality, I hope this soon makes it into a release. (:", "Any update on this feature? ", "We added a proper Polars integration at #3334 if it can help:\r\n```python\r\n>>> from datasets import load_dataset\r\n>>> ds = load_dataset(\"TheBritishLibrary/blbooks\", \"1700_1799\", split=\"train\")\r\n>>> ds.to_polars().groupby('date').len()\r\n┌─────────────────────┬──────┐\r\n│ date ┆ len │\r\n│ --- ┆ --- │\r\n│ datetime[ms] ┆ u32 │\r\n╞═════════════════════╪══════╡\r\n│ 1796-01-01 00:00:00 ┆ 5831 │\r\n│ 1775-01-01 00:00:00 ┆ 4697 │\r\n│ 1749-01-01 00:00:00 ┆ 1118 │\r\n│ 1740-01-01 00:00:00 ┆ 713 │\r\n│ 1714-01-01 00:00:00 ┆ 865 │\r\n│ … ┆ … │\r\n│ 1795-01-01 00:00:00 ┆ 5930 │\r\n│ 1754-01-01 00:00:00 ┆ 1373 │\r\n│ 1780-01-01 00:00:00 ┆ 1970 │\r\n│ 1734-01-01 00:00:00 ┆ 1047 │\r\n│ 1719-01-01 00:00:00 ┆ 1235 │\r\n└─────────────────────┴──────┘\r\n```\r\n" ]

[ "Closed and moved to transformers." ]

[ "Hi @Dhanachandra, \r\n\r\nWe have tests for all our metrics and they work as expected: under the hood, we use scikit-learn implementations.\r\n\r\nMaybe the cause is somewhere else. For example:\r\n- Is it a binary or a multiclass or a multilabel classification? Default computation of these metrics is for binary classification; if you would like multiclass or multilabel, you should pass the corresponding parameters; see their documentation (e.g.: https://scikit-learn.org/stable/modules/generated/sklearn.metrics.precision_score.html) or code below:\r\n\r\nhttps://huggingface.co/docs/datasets/using_metrics.html#computing-the-metric-scores\r\n\r\n```python\r\nIn [1]: from datasets import load_metric\r\n\r\nIn [2]: precision = load_metric(\"precision\")\r\n\r\nIn [3]: print(precision.inputs_description)\r\n\r\nArgs:\r\n predictions: Predicted labels, as returned by a model.\r\n references: Ground truth labels.\r\n labels: The set of labels to include when average != 'binary', and\r\n their order if average is None. Labels present in the data can\r\n be excluded, for example to calculate a multiclass average ignoring\r\n a majority negative class, while labels not present in the data will\r\n result in 0 components in a macro average. For multilabel targets,\r\n labels are column indices. By default, all labels in y_true and\r\n y_pred are used in sorted order.\r\n average: This parameter is required for multiclass/multilabel targets.\r\n If None, the scores for each class are returned. Otherwise, this\r\n determines the type of averaging performed on the data:\r\n binary: Only report results for the class specified by pos_label.\r\n This is applicable only if targets (y_{true,pred}) are binary.\r\n micro: Calculate metrics globally by counting the total true positives,\r\n false negatives and false positives.\r\n macro: Calculate metrics for each label, and find their unweighted mean.\r\n This does not take label imbalance into account.\r\n weighted: Calculate metrics for each label, and find their average\r\n weighted by support (the number of true instances for each label).\r\n This alters ‘macro’ to account for label imbalance; it can result\r\n in an F-score that is not between precision and recall.\r\n samples: Calculate metrics for each instance, and find their average\r\n (only meaningful for multilabel classification).\r\n sample_weight: Sample weights.\r\n\r\nReturns:\r\n precision: Precision score.\r\n\r\nExamples:\r\n\r\n >>> precision_metric = datasets.load_metric(\"precision\")\r\n >>> results = precision_metric.compute(references=[0, 1], predictions=[0, 1])\r\n >>> print(results)\r\n {'precision': 1.0}\r\n\r\n >>> predictions = [0, 2, 1, 0, 0, 1]\r\n >>> references = [0, 1, 2, 0, 1, 2]\r\n >>> results = precision_metric.compute(predictions=predictions, references=references, average='macro')\r\n >>> print(results)\r\n {'precision': 0.2222222222222222}\r\n >>> results = precision_metric.compute(predictions=predictions, references=references, average='micro')\r\n >>> print(results)\r\n {'precision': 0.3333333333333333}\r\n >>> results = precision_metric.compute(predictions=predictions, references=references, average='weighted')\r\n >>> print(results)\r\n {'precision': 0.2222222222222222}\r\n >>> results = precision_metric.compute(predictions=predictions, references=references, average=None)\r\n >>> print(results)\r\n {'precision': array([0.66666667, 0. , 0. ])}\r\n```\r\n" ]

[ "This issue was original reported at https://github.com/huggingface/transformers/issues/14931 and It seems like this issue also occur with other AutoClass like AutoFeatureExtractor.", "Thanks for moving the issue here !\r\n\r\nI wasn't able to reproduce the issue on my env (the hashes stay the same):\r\n```\r\n- `transformers` version: 1.15.0\r\n- `tokenizers` version: 0.10.3\r\n- `datasets` version: 1.18.1\r\n- `dill` version: 0.3.4\r\n- Platform: Linux-4.19.0-18-cloud-amd64-x86_64-with-debian-10.11\r\n- Python version: 3.7.10\r\n- PyArrow version: 6.0.1\r\n```\r\nHowever I was able to reproduce it on Google Colab (the hashes end up different):\r\n```\r\n- `transformers` version: 1.15.0\r\n- `tokenizers` version: 0.10.3\r\n- `datasets` version: 1.18.1\r\n- `dill` version: 0.3.4\r\n- Platform: Linux-5.4.144+-x86_64-with-Ubuntu-18.04-bionic\r\n- Python version: 3.7.12\r\n- PyArrow version: 3.0.0\r\n```\r\nI'll investigate why it doesn't work properly on Google Colab :)", "I found the issue: the tokenizer has something inside it that changes.\r\n\r\nBefore the call, `tokenizer._tokenizer.truncation` is None, and after the call it changes to this for some reason:\r\n```\r\n{'max_length': 512, 'strategy': 'longest_first', 'stride': 0}\r\n```\r\n\r\nDoes anybody know why calling the tokenizer would change its state this way ? cc @Narsil @SaulLu maybe ?", "`tokenizer.encode(..)` does not accept argument like max_length, strategy or stride.\r\n\r\nIn `tokenizers` you have to modify the tokenizer state by setting various `TruncationParams` (and/or `PaddingParams`).\r\nHowever, since this is modifying the state, you need to mutably borrow the tokenizer (a rust concept). The key principle is that there can ever be only 1 mutable borrow at a time during the span of the tokenizer lifecycle.\r\n\r\nBecause of this, if `transformers` blindly set `TruncationParams` and `PaddingParams` on every call, it would cause the tokenizer to crash (or make the various threads accessing it hang, which is not necessarily better).\r\n\r\nIn order to avoid that, we decided to handle it this way : https://github.com/huggingface/transformers/pull/12550 . \r\n\r\nWhich should explain the state of the tokenizer being modified (hence its hash).\r\n\r\nNow for a temporary solution, simply encoding once with the tokenizer should give it it's proper hash (since by default the tokenizer doesn't have this state, looks at the first encoding call, and creates it).\r\n\r\nWe could try and set these 2 dicts at initialization time, but it wouldn't work if a user modified the tokenizer state later\r\n```python\r\ntokenizer = AutoTokenizer.from_pretrained(..)\r\ntokenizer.truncation_side = \"left\"\r\n# Now we have a difference between `tokenizer._tokenizer.truncation` and `tokenizer.truncation_side`\r\n```\r\nIf we wanted to fix it correctly it would mean mapping every assignation to it's proper location on `tokenizer.{padding/truncation}`\r\n\r\nI think it's important to note that we cannot guarantee a tokenizer' hash remains the same if *any* of those parameters are modified through the `.map` function.\r\n\r\nEdit: Another option would be to override the default __hash__ function, but I don't know if there's a sound implementation that could fit.", "Thanks a lot for the explanation !\r\nI think if we set these 2 dicts at initialization time it would be amazing already\r\n\r\nShall we open an issue in `transformers` to ask for these dictionaries to be set when the tokenizer is instantiated ?\r\n\r\n> Edit: Another option would be to override the default hash function, but I don't know if there's a sound implementation that could fit.\r\n\r\nIn `datasets` we can easily have custom hashing for objects of the other HF libraries if we want. For example we ignore the cache some tokenizers have. However in this specific case it touches parameters that may change the behavior of the tokenizer itself. I'm not sure the logic that determines how a tokenizer behaves should be in `datasets`", "A hack we could have in the `datasets` lib would be to call the tokenizer before hashing it in order to set all its parameters correctly - but it sounds a lot like a hack and I'm not sure this can work in the long run", "Fully agree with everything you said. \r\n\r\nI think the best course of action is creating an issue in `transformers`. I can start the work on this.\r\nI think the code changes are fairly simple. Making a sound test + not breaking other stuff might be different :D", "It should be noted that this problem also occurs in other AutoClasses, such as AutoFeatureExtractor, so I don't think handling it in Datasets is a long-term practice either.", "> I think the best course of action is creating an issue in `transformers`. I can start the work on this.\r\n\r\n@Narsil Hi, I reopen this issue in `transformers` https://github.com/huggingface/transformers/issues/14931", "Here is @Narsil comment from https://github.com/huggingface/transformers/issues/14931#issuecomment-1074981569\r\n> # TL;DR\r\n> Call the function once on a dummy example beforehand will fix it.\r\n> \r\n> ```python\r\n> tokenizer(\"Some\", \"test\", truncation=True)\r\n> ```\r\n> \r\n> # Long answer\r\n> If I remember the last status, it's hard doing anything, since the call itself\r\n> \r\n> ```python\r\n> tokenizer(example[\"sentence1\"], example[\"sentence2\"], truncation=True)\r\n> ```\r\n> \r\n> will modify the tokenizer. It's the `truncation=True` that modifies the tokenizer to put it into truncation mode if you will. Calling the tokenizer once with that argument would fix the cache.\r\n> \r\n> Finding a fix that :\r\n> \r\n> * Doesn't imply a huge chunk of work on `tokenizers` (with potential loss of performance, and breaking backward compatibility)\r\n> * Doesn't imply `datasets` running a first pass of the loop\r\n> * Doesn't imply `datasets` looking at the map function itself\r\n> * Uses a sound `hash` for this object in `datasets`.\r\n> \r\n> is IIRC impossible for this use case.\r\n> \r\n> I can explain a bit more why the first option is not desirable.\r\n> \r\n> In order to \"fix\" this for tokenizers, we would need to make `tokenizer(..)` purely without side effects. This means that the \"options\" of tokenization (like `truncation` and `padding` at least) would have\r\n", "For me this workaround only works if I don't pass the `num_proc=X` argument to `datasets.map`", "Is there an easy solution for setting both num_proc and padding/truncation for fast tokenizer or caching just not a thing in this case? " ]

[ "Hi @lewtun!\r\n \r\nThis one was tricky to debug. Initially, I tought there is a bug in the recently-added (by @lhoestq ) `cast_array_to_feature` function because `git bisect` points to the https://github.com/huggingface/datasets/commit/6ca96c707502e0689f9b58d94f46d871fa5a3c9c commit. Then, I noticed that the feature tpye of the `dialogue` field is `list`, which explains why you didn't get an error in earlier versions. Is there a specific reason why you use `list` instead of `Sequence` in the script? Maybe to avoid turning list of dicts to dicts of lists as it's done by `Sequence` for compatibility with TFDS or for performance reasons? If the field was `Sequence`, you would get an error in `encode_nested_example` because **the scripts yields some additional (nested) columns which are not specified in the `features` dictionary**. Previously, these additional columns would've been ignored by PyArrow (1), but now we have a check for them (2).\r\n(1) See PyArrow behavior:\r\n```\r\n>>> pa.array([{\"a\": 2, \"b\": 3}], type=pa.struct({\"a\": pa.int32()})) # pyarrow ignores the extra column\r\n-- is_valid: all not null\r\n-- child 0 type: int32\r\n [\r\n 2\r\n ]\r\n ```\r\n\r\n(2) Check:\r\nhttps://github.com/huggingface/datasets/blob/4c417d52def6e20359ca16c6723e0a2855e5c3fd/src/datasets/table.py#L1059\r\n\r\nThe fix is very simple: just add the missing columns to the _EMPTY_BELIEF_STATE list:\r\n```python\r\n_EMPTY_BELIEF_STATE.extend(['通用-产品类别', '火车-舱位档次', '通用-系列', '通用-价格区间', '通用-品牌'])\r\n```", "Hey @mariosasko, thank you so much for figuring this one out - it certainly looks like a tricky bug 😱 ! I don't think there's a specific reason to use `list` instead of `Sequence` with the script, but I'll let the dataset creators know to see if your suggestion is acceptable.\r\n\r\nThank you again!", "Thanks, this was indeed the fix! Would it make sense to produce a more informative error message in such cases? \r\n\r\nThe issue can be closed. \r\n\r\n" ]

[ "I'm not sure if this is expected behavior.\r\n\r\nAm I supposed to work with a copy of the dataset, i.e. `shuffled_dataset = data.shuffle(seed=None)`?\r\n\r\n```diff\r\nimport datasets\r\n\r\n# Some toy example\r\ndata = datasets.Dataset.from_dict(\r\n {\"feature\": [1, 2, 3, 4, 5], \"label\": [\"a\", \"b\", \"c\", \"d\", \"e\"]}\r\n)\r\n\r\n+shuffled_data = data.shuffle(seed=None)\r\n\r\n# Doesn't work as expected\r\nprint(\"Shuffle dataset\")\r\nfor _ in range(3):\r\n+ shuffled_data = shuffled_data.shuffle(seed=None)\r\n+ print(shuffled_data[:])\r\n- print(data.shuffle(seed=None)[:])\r\n\r\n# This seems to work with pandas\r\nprint(\"\\nShuffle via pandas\")\r\nfor _ in range(3):\r\n df = data.to_pandas().sample(frac=1.0)\r\n print(datasets.Dataset.from_pandas(df, preserve_index=False)[:])\r\n\r\n```\r\n\r\nor provide a `generator` instead?\r\n\r\n```diff\r\nimport datasets\r\n+from numpy.random import default_rng\r\n\r\n# Some toy example\r\ndata = datasets.Dataset.from_dict(\r\n {\"feature\": [1, 2, 3, 4, 5], \"label\": [\"a\", \"b\", \"c\", \"d\", \"e\"]}\r\n)\r\n\r\n+rng = default_rng()\r\n\r\n# Doesn't work as expected\r\nprint(\"Shuffle dataset\")\r\nfor _ in range(3):\r\n+ print(data.shuffle(generator=rng)[:])\r\n- print(data.shuffle(seed=None)[:])\r\n\r\n# This seems to work with pandas\r\nprint(\"\\nShuffle via pandas\")\r\nfor _ in range(3):\r\n df = data.to_pandas().sample(frac=1.0)\r\n print(datasets.Dataset.from_pandas(df, preserve_index=False)[:])\r\n\r\n```", "Hi! Thanks for reporting! Yes, this is not expected behavior. I've opened a PR with the fix." ]

[ "Hi @AnzorGozalishvili,\r\n\r\nMaybe their site was temporarily down, but it seems to work fine now.\r\n\r\nCould you please try again and confirm if the problem persists? ", "Hi @albertvillanova \r\nI checked and it works. \r\nIt seems that it was really temporarily down.\r\nThanks!" ]

[ "Hi @pichljan, thanks for reporting.\r\n\r\nThis should be fixed. I'm looking at it. " ]

[ "Thanks for reporting, @StevenTang1998.\r\n\r\nI'm fixing it. " ]

[]

[]

[ "A question to the datasets maintainers: is there a policy about how the set of allowed metadata fields is maintained and expanded?\r\n\r\nMetadata are very important, but defining the standard is always a struggle between allowing exhaustivity without being too complex. Archivists have Dublin Core, open data has https://frictionlessdata.io/, geo has ISO 19139 and INSPIRE, etc. and it's always a mess! I'm not sure we want to dig too much into it, but I'm curious to know if there has been some work on the metadata standard.", "> Metadata are very important, but defining the standard is always a struggle between allowing exhaustivity without being too complex. Archivists have Dublin Core, open data has [frictionlessdata.io](https://frictionlessdata.io/), geo has ISO 19139 and INSPIRE, etc. and it's always a mess! I'm not sure we want to dig too much into it, but I'm curious to know if there has been some work on the metadata standard.\r\n\r\n\r\nI thought this is a potential issue with adding this field since it might be hard to define what is general enough to be useful for most data vs what becomes very domain-specific. Potentially adding one extra field leads to more and more fields in the future. \r\n\r\nAnother issue is that there are some metadata standards around data i.e. [datacite](https://schema.datacite.org/meta/kernel-4.4/), but not many aimed explicitly at ML data afaik. Some of the discussions around metadata for ML are also more focused on versioning/managing data in production environments. My thinking is that here, some reference to the time of production would also often be tracked/relevant, i.e. for triggering model training, so having this information available in the hub would also help address this use case. ", "Adding a relevant paper related to this topic: [TimeLMs: Diachronic Language Models from Twitter](https://arxiv.org/abs/2202.03829)\r\n\r\n", "Related: https://github.com/huggingface/datasets/issues/3877", "Also related: the [Data Catalog Vocabulary - DCAT](https://www.w3.org/TR/vocab-dcat/) standard will be discussed in a new Working Group at the W3C: https://www.w3.org/2022/06/dx-wg-charter.html" ]

[]

[ "Hi! We use `tqdm` to display progress bars, so I suggest you open this issue in their repo.", "It depends on how you use `tqdm`, no? \r\n\r\nDoesn't this library import via; \r\n\r\n```\r\nfrom tqdm.notebook import tqdm\r\n```", "Hi! Sorry for the late reply. We import `tqdm` as `from tqdm.auto import tqdm`, which should be equal to `from tqdm.notebook import tqdm` in Jupyter.", "Any objection if I make a PR that checks if the widgets library is installed beforehand? " ]

[ "Hi ! I think you should avoid calling `timit_train['audio']`. Indeed by doing so you're **loading all the audio column in memory**. This is problematic in your case because the TIMIT dataset is huge.\r\n\r\nIf you want to access the audio data of some samples, you should do this instead `timit_train[:10][\"train\"]` for example.\r\n\r\nOther than that, I'm not sure why you get a `TypeError: string indices must be integers`, do you have a code snippet that reproduces the issue that you can share here ?", "I get the same error when I try to do `timit_train[0]` or really any indexing into the whole thing. \r\n\r\nReally, that IS the code snippet that reproduces the issue. If you index into other fields like 'file' or whatever, it works. As soon as one of the fields you're looking into is 'audio', you get that issue. It's a weird issue and I suspect it's Sagemaker/environment related, maybe the mix of libraries and dependencies are not good. \r\n\r\n\r\nExample code snippet with issue. \r\n```python\r\nfrom datasets import load_dataset\r\n\r\ntimit_train = load_dataset('timit_asr', split='train')\r\nprint(timit_train[0])\r\n```", "Ok I see ! From the error you got, it looks like the `value` encoded in the arrow file of the TIMIT dataset you loaded is a string instead of a dictionary with keys \"path\" and \"bytes\" but we don't support this since 1.18\r\n\r\nCan you try regenerating the dataset with `load_dataset('timit_asr', download_mode=\"force_redownload\")` please ? I think it should fix the issue." ]

[ "@albertvillanova let me know if there is anything I can do to help with this. I had a quick look at the code again and though I could try the following changes:\r\n- use `download` instead of `download_and_extract`\r\nhttps://github.com/huggingface/datasets/blob/d3d339fb86d378f4cb3c5d1de423315c07a466c6/datasets/bnl_newspapers/bnl_newspapers.py#L136\r\n- swith to using `iter_archive` to loop through downloaded data to replace\r\nhttps://github.com/huggingface/datasets/blob/d3d339fb86d378f4cb3c5d1de423315c07a466c6/datasets/bnl_newspapers/bnl_newspapers.py#L159\r\n\r\nLet me know if it's useful for me to try and make those changes. ", "Thanks @davanstrien.\r\n\r\nI have already been working on it so that it can be used in the BigScience workshop.\r\n\r\nI agree that the `rglob()` is not efficient in this case.\r\n\r\nI tried different solutions without success:\r\n- `iter_archive` cannot be used in this case because it does not support ZIP files yet\r\n\r\nFinally I have used `iter_files()`.", "I see this is fixed now 🙂. I also picked up a few other tips from your redactors so hopefully my next attempts will support streaming from the start. " ]

[ "Yes. The jobs queue is full right now, following an upgrade... Back to normality in the next hours hopefully. I'll look at your datasets to be sure the dataset viewer works as expected on them.", "Should have been fixed now." ]

[ "Hi! Thanks for reporting! I've opened a PR with the fix." ]

[ "It is solved now" ]

[ "Hey @ck37 \r\n\r\nYou can always use a custom metric as explained [in this guide from HF](https://huggingface.co/docs/datasets/master/loading_metrics.html#using-a-custom-metric-script).\r\n\r\nIf this issue needs to be contributed to (for enhancing the metric API) I think [this link](https://scikit-learn.org/stable/modules/generated/sklearn.metrics.mean_absolute_error.html) would be helpful for the `MAE` metric.", "You can use a local metric script just by providing its path instead of the usual shortcut name ", "#self-assign I have starting working on this issue to enhance the metric API." ]

[ "Hi ! We just released a new version of `datasets` that should fix this.\r\n\r\nI tested resampling and using save/load_from_disk afterwards and it seems to be fixed now", "Hi @lhoestq, \r\n\r\nJust tested the latest datasets version, and confirming that this is fixed for me. \r\n\r\nThanks!", "Also, just an FYI, data that I had saved (with save_to_disk) previously from common voice using datasets==1.17.0 now give the error below when loading (with load_from disk) using datasets==1.18.0. \r\n\r\nHowever, when starting fresh using load_dataset, then doing the resampling, the save/load_from disk worked fine. \r\n\r\n```\r\n---------------------------------------------------------------------------\r\nValueError Traceback (most recent call last)\r\n<timed exec> in <module>\r\n\r\n/opt/conda/lib/python3.7/site-packages/datasets/load.py in load_from_disk(dataset_path, fs, keep_in_memory)\r\n 1747 return Dataset.load_from_disk(dataset_path, fs, keep_in_memory=keep_in_memory)\r\n 1748 elif fs.isfile(Path(dest_dataset_path, config.DATASETDICT_JSON_FILENAME).as_posix()):\r\n-> 1749 return DatasetDict.load_from_disk(dataset_path, fs, keep_in_memory=keep_in_memory)\r\n 1750 else:\r\n 1751 raise FileNotFoundError(\r\n\r\n/opt/conda/lib/python3.7/site-packages/datasets/dataset_dict.py in load_from_disk(dataset_dict_path, fs, keep_in_memory)\r\n 769 else Path(dest_dataset_dict_path, k).as_posix()\r\n 770 )\r\n--> 771 dataset_dict[k] = Dataset.load_from_disk(dataset_dict_split_path, fs, keep_in_memory=keep_in_memory)\r\n 772 return dataset_dict\r\n 773 \r\n\r\n/opt/conda/lib/python3.7/site-packages/datasets/arrow_dataset.py in load_from_disk(dataset_path, fs, keep_in_memory)\r\n 1118 info=dataset_info,\r\n 1119 split=split,\r\n-> 1120 fingerprint=state[\"_fingerprint\"],\r\n 1121 )\r\n 1122 \r\n\r\n/opt/conda/lib/python3.7/site-packages/datasets/arrow_dataset.py in __init__(self, arrow_table, info, split, indices_table, fingerprint)\r\n 655 if self.info.features.type != inferred_features.type:\r\n 656 raise ValueError(\r\n--> 657 f\"External features info don't match the dataset:\\nGot\\n{self.info.features}\\nwith type\\n{self.info.features.type}\\n\\nbut expected something like\\n{inferred_features}\\nwith type\\n{inferred_features.type}\"\r\n 658 )\r\n 659 \r\n\r\nValueError: External features info don't match the dataset:\r\nGot\r\n{'accent': Value(dtype='string', id=None), 'age': Value(dtype='string', id=None), 'audio': Audio(sampling_rate=48000, mono=True, id=None), 'client_id': Value(dtype='string', id=None), 'down_votes': Value(dtype='int64', id=None), 'gender': Value(dtype='string', id=None), 'locale': Value(dtype='string', id=None), 'path': Value(dtype='string', id=None), 'segment': Value(dtype='string', id=None), 'sentence': Value(dtype='string', id=None), 'up_votes': Value(dtype='int64', id=None)}\r\nwith type\r\nstruct<accent: string, age: string, audio: struct<bytes: binary, path: string>, client_id: string, down_votes: int64, gender: string, locale: string, path: string, segment: string, sentence: string, up_votes: int64>\r\n\r\nbut expected something like\r\n{'accent': Value(dtype='string', id=None), 'age': Value(dtype='string', id=None), 'audio': {'path': Value(dtype='string', id=None), 'bytes': Value(dtype='binary', id=None)}, 'client_id': Value(dtype='string', id=None), 'down_votes': Value(dtype='int64', id=None), 'gender': Value(dtype='string', id=None), 'locale': Value(dtype='string', id=None), 'path': Value(dtype='string', id=None), 'segment': Value(dtype='string', id=None), 'sentence': Value(dtype='string', id=None), 'up_votes': Value(dtype='int64', id=None)}\r\nwith type\r\nstruct<accent: string, age: string, audio: struct<path: string, bytes: binary>, client_id: string, down_votes: int64, gender: string, locale: string, path: string, segment: string, sentence: string, up_votes: int64> \r\n```" ]

[ "Sure, it's on the roadmap.", "Closing in favor of https://github.com/huggingface/datasets-server/issues/39." ]

[ "similar #3611 " ]

[ "i suspect a markdown parsing error, @severo do you want to take a quick look at it when you have some time?", "# Problem\r\nThe issue seems to coming from the front matter of the README\r\n```---\r\nannotations_creators:\r\n- no-annotation\r\nlanguage_creators:\r\n- machine-generated\r\nlanguages:\r\n- 'ca'\r\n- 'de'\r\nlicenses:\r\n- cc-by-4.0\r\nmultilinguality:\r\n- translation\r\npretty_name: Catalan-German aligned corpora to train NMT systems.\r\nsize_categories:\r\n- \"1M<n<10M\" \r\nsource_datasets:\r\n- extended|tilde_model\r\ntask_categories:\r\n- machine-translation\r\ntask_ids:\r\n- machine-translation\r\n---\r\n``` \r\n# Solution\r\nThe fix is to correctly style the README as explained [here](https://huggingface.co/docs/datasets/v1.12.0/dataset_card.html). I have also correctly parsed the font matter as shown below:\r\n```\r\n---\r\nannotations_creators: []\r\nlanguage_creators: [machine-generated]\r\nlanguages: ['ca', 'de']\r\nlicenses: []\r\nmultilinguality:\r\n- multilingual\r\npretty_name: 'Catalan-German aligned corpora to train NMT systems.'\r\nsize_categories: \r\n- 1M<n<10M\r\nsource_datasets: ['extended|tilde_model']\r\ntask_categories: ['machine-translation']\r\ntask_ids: ['machine-translation']\r\n---\r\n```\r\nYou can find the README for a sample dataset [here](https://huggingface.co/datasets/ritwikraha/Test)", "Thank you. It finally worked implementing your changes and leaving a white line between title and text in the description.", "Thanks, if this solves your issue, can you please close it?" ]

[ "Hi! The `cd` command in Jupyer/Colab needs to start with `%`, so this should work:\r\n```\r\n!git clone https://github.com/huggingface/datasets.git\r\n%cd datasets\r\n!pip install -e \".[streaming]\"\r\n```", "thanks @mariosasko i had the same mistake and your solution is what was needed" ]

[ "Hi! Thanks for reporting! The issue with `cast_column` should be fixed by #3575 and after we merge that PR I'll start working on the `push_to_hub` support for the `Image`/`Audio` feature.", "> Hi! Thanks for reporting! The issue with `cast_column` should be fixed by #3575 and after we merge that PR I'll start working on the `push_to_hub` support for the `Image`/`Audio` feature.\r\n\r\nThanks, I'll keep an eye out for #3575 getting merged. I managed to use `push_to_hub` sucesfully with images when they were loaded via `map` - something like `ds.map(lambda example: {\"img\": load_image_function(example['fname']})`, this only pushed the images to the hub if the `load_image_function` return a PIL Image without the filename attribute though. I guess this might often be the prefered behaviour though. \r\n", "Hi ! We merged the PR and did a release of `datasets` that includes the changes. Can you try updating `datasets` and try again ?", "> Hi ! We merged the PR and did a release of `datasets` that includes the changes. Can you try updating `datasets` and try again ?\r\n\r\nThanks for checking. There is no longer an error when calling `select` but it appears the cast value isn't preserved. Before `select`\r\n\r\n```python\r\ndataset.features\r\n{'url': Image(id=None)}\r\n```\r\n\r\nafter select:\r\n```\r\n{'url': Value(dtype='string', id=None)}\r\n```\r\n\r\nUpdated Colab example [here](https://colab.research.google.com/gist/davanstrien/4e88f55a3675c279b5c2f64299ae5c6f/potential_casting_bug.ipynb) ", "Hmmm, if I re-run your google colab I'm getting the right type at the end:\r\n```\r\nsample.features\r\n# {'url': Image(id=None)}\r\n```", "Appolgies - I've just run again and also got this output. I have also sucesfully used the `push_to_hub` method. I think this is fixed now so will close this issue. ", "Fixed in #3575 " ]

[]

[]

[ "This seems related to https://github.com/huggingface/datasets/issues/3505." ]

[]

[ "Hello! I have recently been involved with a medical image segmentation project myself and was going through the `The Medical Segmentation Decathlon Dataset` as well. \r\nI haven't yet had experience adding datasets to this repository yet but would love to get started. Should I take this issue?\r\nIf yes, I've got two questions -\r\n1. There are 10 different datasets available, so are all datasets to be added in a single PR, or one at a time? \r\n2. Since it's a competition, masks for the test-set are not available. How is that to be tackled? Sorry if it's a silly question, I have recently started exploring `datasets`.", "Hi! Sure, feel free to take this issue. You can self-assign the issue by commenting `#self-assign`.\r\n\r\nTo answer your questions:\r\n1. It makes the most sense to add each one as a separate config, so one dataset script with 10 configs in a single PR.\r\n2. Just set masks in the test set to `None`.\r\n\r\nNote that the images/masks in this dataset are in NIfTI format, which our `Image` feature currently doesn't support, so I think it's best to yield the paths to the images/masks in the script and add a preprocessing section to the card where we explain how to load/process the images/masks with `nibabel` (I can help with that). \r\n\r\n", "> Note that the images/masks in this dataset are in NIfTI format, which our `Image` feature currently doesn't support, so I think it's best to yield the paths to the images/masks in the script and add a preprocessing section to the card where we explain how to load/process the images/masks with `nibabel` (I can help with that).\r\n\r\nGotcha, thanks. Will start working on the issue and let you know in case of any doubt.", "#self-assign", "This is great! There is a first model on the HUb that uses this dataset! https://huggingface.co/MONAI/example_spleen_segmentation" ]

[ "I came to open the same issue.", "Thanks for reporting !\r\n\r\nI pushed a temporary fix on `master` that uses an URL from a previous commit to access the dataset for now, until we have a better solution", "I changed the URL again to use another host, the fix is available on `master` and we'll probably do a new release of `datasets` tomorrow.\r\n\r\nIn the meantime, feel free to do `load_dataset(..., revision=\"master\")` to use the fixed script", "We just released a new version of `datasets` with a working URL. Feel free to update `datasets` and try again :)", "Hello! Unfortunately, this URL does not work for me. \r\nCould you please tell me how I can solve the problem?\r\n\r\n`>>> from datasets import load_dataset\r\n>>> dataset = load_dataset(\"conll2003\")\r\nDownloading and preparing dataset conll2003/conll2003 (download: 4.63 MiB, generated: 9.78 MiB, post-processed: Unknown size, total: 14.41 MiB) to /home/dafedo/.cache/huggingface/datasets/conll2003/conll2003/1.0.0/40e7cb6bcc374f7c349c83acd1e9352a4f09474eb691f64f364ee62eb65d0ca6...\r\nTraceback (most recent call last):\r\n File \"<stdin>\", line 1, in <module>\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/load.py\", line 745, in load_dataset\r\n builder_instance.download_and_prepare(\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/builder.py\", line 574, in download_and_prepare\r\n self._download_and_prepare(\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/builder.py\", line 630, in _download_and_prepare\r\n split_generators = self._split_generators(dl_manager, **split_generators_kwargs)\r\n File \"/home/dafedo/.cache/huggingface/modules/datasets_modules/datasets/conll2003/40e7cb6bcc374f7c349c83acd1e9352a4f09474eb691f64f364ee62eb65d0ca6/conll2003.py\", line 196, in _split_generators\r\n downloaded_files = dl_manager.download_and_extract(urls_to_download)\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/download_manager.py\", line 287, in download_and_extract\r\n return self.extract(self.download(url_or_urls))\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/download_manager.py\", line 195, in download\r\n downloaded_path_or_paths = map_nested(\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/py_utils.py\", line 203, in map_nested\r\n mapped = [\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/py_utils.py\", line 204, in <listcomp>\r\n _single_map_nested((function, obj, types, None, True)) for obj in tqdm(iterable, disable=disable_tqdm)\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/py_utils.py\", line 142, in _single_map_nested\r\n return function(data_struct)\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/download_manager.py\", line 218, in _download\r\n return cached_path(url_or_filename, download_config=download_config)\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/file_utils.py\", line 281, in cached_path\r\n output_path = get_from_cache(\r\n File \"/home/dafedo/efficient-task-transfer/venv/lib/python3.8/site-packages/datasets/utils/file_utils.py\", line 621, in get_from_cache\r\n raise FileNotFoundError(\"Couldn't find file at {}\".format(url))\r\nFileNotFoundError: Couldn't find file at https://github.com/davidsbatista/NER-datasets/raw/master/CONLL2003/train.txt\r\n`\r\n\r\nI receive the same error when I run \"itrain run_configs/conll2003.json\" from https://github.com/adapter-hub/efficient-task-transfer\r\n\r\nThank you very much in advance!\r\n\r\nRegards, \r\nDaria\r\n", "Can you try updating `datasets` and try again ?\r\n```\r\npip install -U datasets\r\n```", "@lhoestq Thank you very much for your answer! \r\n\r\nIt works this way, but for my research I need datasets==1.6.3 or closest to it because otherwise the other package would not work as it is built on this version.\r\nDo you have any other suggestion? I would really appreciate it. Maybe which version of the datasets is without hard-coded link but closest to 1.6.3\r\n", "No problem, I have solved it. \r\nThank you anyway.", "Out of curiosity, which package has the `datasets==1.6.3` requirement ?" ]

[ "Hi ! Currently it only works with local paths, file-like objects are not supported yet" ]

[ "+1, here's the error I got: \r\n\r\n```\r\n>>> from datasets import load_dataset\r\n>>>\r\n>>> load_dataset(\"wiki_bio\")\r\nDownloading: 7.58kB [00:00, 4.42MB/s]\r\nDownloading: 2.71kB [00:00, 1.30MB/s]\r\nUsing custom data configuration default\r\nDownloading and preparing dataset wiki_bio/default (download: 318.53 MiB, generated: 736.94 MiB, post-processed: Unknown size, total: 1.03 GiB) to /home/jxm3/.cache/huggingface/datasets/wiki_bio/default/1.1.0/5293ce565954ba965dada626f1e79684e98172d950371d266bf3caaf87e911c9...\r\nTraceback (most recent call last):\r\n File \"<stdin>\", line 1, in <module>\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/load.py\", line 1694, in load_dataset\r\n builder_instance.download_and_prepare(\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/builder.py\", line 595, in download_and_prepare\r\n self._download_and_prepare(\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/builder.py\", line 662, in _download_and_prepare\r\n split_generators = self._split_generators(dl_manager, **split_generators_kwargs)\r\n File \"/home/jxm3/.cache/huggingface/modules/datasets_modules/datasets/wiki_bio/5293ce565954ba965dada626f1e79684e98172d950371d266bf3caaf87e911c9/wiki_bio.py\", line 125, in _split_generators\r\n data_dir = dl_manager.download_and_extract(my_urls)\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/utils/download_manager.py\", line 308, in download_and_extract\r\n return self.extract(self.download(url_or_urls))\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/utils/download_manager.py\", line 196, in download\r\n downloaded_path_or_paths = map_nested(\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/utils/py_utils.py\", line 251, in map_nested\r\n return function(data_struct)\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/utils/download_manager.py\", line 217, in _download\r\n return cached_path(url_or_filename, download_config=download_config)\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/utils/file_utils.py\", line 298, in cached_path\r\n output_path = get_from_cache(\r\n File \"/home/jxm3/.conda/envs/torch/lib/python3.9/site-packages/datasets/utils/file_utils.py\", line 612, in get_from_cache\r\n raise FileNotFoundError(f\"Couldn't find file at {url}\")\r\nFileNotFoundError: Couldn't find file at https://drive.google.com/uc?export=download&id=1L7aoUXzHPzyzQ0ns4ApBbYepsjFOtXil\r\n>>>\r\n```\r\n", "@alejandrocros and @lhoestq - you added the wiki_bio dataset in #1173. It doesn't work anymore. Can you take a look at this?", "And if something is wrong with Google Drive, you could try to download (and collate and unzip) from here: https://github.com/DavidGrangier/wikipedia-biography-dataset", "Hi ! Thanks for reporting. I've downloaded the data and concatenated them into a zip file available here: https://huggingface.co/datasets/wiki_bio/tree/main/data\r\n\r\nI guess we can update the dataset script to use this zip file now :)" ]

[ "Hi ! We did a release of `datasets` today that may fix this issue. Can you try updating `datasets` and trying again ?\r\n\r\nEDIT: the issue is still there actually\r\n\r\nI think we can fix that by storing the Features in the parquet schema metadata, and then reload them when loading the parquet file", "This info is stored in the Parquet schema metadata as of https://github.com/huggingface/datasets/pull/5516" ]

[]

[ "@sahoodib, thanks for reporting.\r\n\r\nIndeed, none of the data links appearing in the IndicGLUE website are working, e.g.: https://storage.googleapis.com/ai4bharat-public-indic-nlp-corpora/evaluations/soham-articles.tar.gz\r\n```\r\n<Error>\r\n<Code>UserProjectAccountProblem</Code>\r\n<Message>User project billing account not in good standing.</Message>\r\n<Details>\r\nThe billing account for the owning project is disabled in state delinquent\r\n</Details>\r\n</Error>\r\n```\r\n\r\nWe have contacted the data owners to inform them about their issue and ask them if they plan to fix it.", "Yesterday I resent a reminder email with more AI4Bharat-related people in the loop.\r\n\r\nI also opened an issue in their repos:\r\n- https://github.com/AI4Bharat/indicnlp_corpus/issues/14\r\n- https://github.com/AI4Bharat/ai4bharat.org/issues/71", "We have received a reply from the authors reporting they have updated the URLs of their data files and opened a PR. See:\r\n- #4978 " ]

[ "Hi @fabianslife, thanks for reporting.\r\n\r\nI think you were using an old version of `datasets` because this bug was already fixed in version `1.13.0` (13 Oct 2021):\r\n- Fix: 55fd140a63b8f03a0e72985647e498f1fc799d3f\r\n- PR: #3046\r\n- Issue: #2969 \r\n\r\nPlease, feel free to update the library: `pip install -U datasets`." ]

[ "Hi! That makes sense. Sure, feel free to open a PR! Just a small suggestion: let's make `preserve_index` a parameter of `Dataset.from_pandas` (which we then pass to `InMemoryTable.from_pandas`) with `None` as a default value to not have this as a breaking change. " ]

[ "The host of this copy of the dataset (https://the-eye.eu) is down and has been down for a good amount of time ([potentially months](https://www.reddit.com/r/Roms/comments/q82s15/theeye_downdied/))\r\n\r\nFinding this dataset is a little esoteric, as the original authors took down the official BookCorpus dataset some time ago.\r\n\r\nThere are community-created versions of BookCorpus, such as the files hosted in the link below.\r\nhttps://battle.shawwn.com/sdb/bookcorpus/\r\n\r\nAnd more discussion here:\r\nhttps://github.com/soskek/bookcorpus\r\n\r\nDo we want to remove this dataset entirely? There's a fair argument for this, given that the official BookCorpus dataset was taken down by the authors. If not, perhaps can open a PR with the link to the community-created tar above and updated dataset description.", "Hi! The `bookcorpusopen` dataset is not working for the same reason as explained in this comment: https://github.com/huggingface/datasets/issues/3504#issuecomment-1004564980", "Hi @HUIYINXUE, it should work now that the data owners created a mirror server with all data, and we updated the URL in our library." ]

[ "Hi @mariosasko，Just search randomly and I found this issue~ I'm the tech lead of Milvus and we are looking forward to integrate milvus together with huggingface datasets.\r\n\r\nAny suggestion on how we could start?\r\n", "Feel free to assign to me and we probably need some guide on it", "@mariosasko any updates my man?\r\n", "Hi! For starters, I suggest you take a look at this file: https://github.com/huggingface/datasets/blob/master/src/datasets/search.py, which contains all the code for Faiss/ElasticSearch support. We could set up a Slack channel for additional guidance. Let me know what you prefer.", "> Hi! For starters, I suggest you take a look at this file: https://github.com/huggingface/datasets/blob/master/src/datasets/search.py, which contains all the code for Faiss/ElasticSearch support. We could set up a Slack channel for additional guidance. Let me know what you prefer.\r\n\r\nSure, we take a look and do some research" ]

[ "Hi, we've just merged the PR with the fix. The fixed version of the dataset can be downloaded as follows:\r\n```python\r\nimport datasets\r\ndset = datasets.load_dataset(\"tweet_qa\", revision=\"master\")\r\n```" ]

[ "Hi! I can't reproduce this error in Colab, but I'm assuming you are using Amazon SageMaker Studio Notebooks (you mention the `conda_pytorch_p36` kernel), so maybe @philschmid knows more about what might be causing this issue? ", "Hey @mariosasko. Yes, I am using **Amazon SageMaker Studio Jupyter Labs**. However, I no longer need this notebook; but it would be nice to have this problem solved for others. So don't stress too much if you two can't reproduce error.", "Hey @danielbellhv, \r\n\r\nThis issue might be related to Studio probably not having an up to date `botocore` and `boto3` version. I ran into this as well a while back. My workaround was \r\n```python\r\n# using older dataset due to incompatibility of sagemaker notebook & aws-cli with > s3fs and fsspec to >= 2021.10\r\n!pip install \"datasets==1.13\" --upgrade\r\n```\r\n\r\nIn `datasets` we use the latest `s3fs` and `fsspec` but aws-cli and notebook is not supporting this. You could also update the `aws-cli` and associated packages to get the latest `datasets` version\r\n" ]

[ "A quick fix for now is doing this:\r\n\r\n```python\r\nX_train = np.stack(dataset[\"train\"][\"image\"])[..., None]", "This error also propagates to jax and is even trickier to fix, since `.with_format(type='jax')` will use numpy conversion internally (and fail). For a three line failure:\r\n\r\n```python\r\ndataset = datasets.load_dataset(\"mnist\")\r\ndataset.set_format(\"jax\")\r\nX_train = dataset[\"train\"][\"image\"]\r\n```", "Hi! We've recently introduced a new Image feature that yields PIL Images (and caches transforms on them) instead of arrays.\r\n\r\nHowever, this feature requires a custom transform to yield np arrays directly:\r\n```python\r\nddict = datasets.load_dataset(\"mnist\")\r\n\r\ndef pil_image_to_array(batch):\r\n return {\"image\": [np.array(img) for img in batch[\"image\"]]} # or jnp.array(img) for Jax\r\n\r\nddict.set_transform(pil_image_to_array, columns=\"image\", output_all_columns=True)\r\n```\r\n\r\n[Docs](https://huggingface.co/docs/datasets/master/process.html#format-transform) on `set_transform`.\r\n\r\nAlso, the approach proposed by @cgarciae is not the best because it loads the entire column in memory.\r\n\r\n@albertvillanova @lhoestq WDYT? The Audio and the Image feature currently don't support the TF/Jax/PT Formatters, but for the Numpy Formatter maybe it makes more sense to return np arrays (and not a dict in the case of the Audio feature or a PIL Image object in the case of the Image feature).", "Yes I agree it should return arrays and not a PIL image (and possible an array instead of a dict for audio data).\r\nI'm currently finishing some code refactoring of the image and audio and opening a PR today. Maybe we can look into that after the refactoring", "This has been fixed in https://github.com/huggingface/datasets/pull/5072, which is included in the latest release of `datasets`." ]

[ "Closed by:\r\n- #4259" ]