Datasets:

---

miesnerjacob

/

github-issues

like 0

[]

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

[ "Thanks for reporting and for all the details and suggestions.\r\n\r\nI'm totally in favor of having a HF_DATASETS_OFFLINE env variable to disable manually all the connection checks, remove retries etc.\r\n\r\nMoreover you may know that the use case that you are mentioning is already supported from `datasets` 1.3.0, i.e. you already can:\r\n- first load datasets and metrics from an instance with internet connection\r\n- then be able to reload datasets and metrics from another instance without connection (as long as the filesystem is shared)\r\n\r\nThis is already implemented, but currently it only works if the requests return a `ConnectionError` (or any error actually). Not sure why it would hang instead of returning an error.\r\n\r\nMaybe this is just a issue with the timeout value being not set or too high ?\r\nIs there a way I can have access to one of the instances on which there's this issue (we can discuss this offline) ?\r\n", "I'm on master, so using all the available bells and whistles already.\r\n\r\nIf you look at the common issues - it for example tries to look up files if they appear in `_PACKAGED_DATASETS_MODULES` which it shouldn't do.\r\n\r\n--------------\r\n\r\nYes, there is a nuance to it. As I mentioned it's firewalled - that is it has a network but making any calls outside - it just hangs in:\r\n\r\n```\r\nsin_addr=inet_addr(\"xx.xx.xx.xx\")}, [28->16]) = 0\r\nclose(5) = 0\r\nsocket(AF_INET, SOCK_STREAM|SOCK_CLOEXEC, IPPROTO_TCP) = 5\r\nconnect(5, {sa_family=AF_INET, sin_port=htons(3128), sin_addr=inet_addr(\"yy.yy.yy.yy\")}, 16^C) = ? ERESTARTSYS (To be restarted if SA_RESTART is set)\r\n```\r\nuntil it times out.\r\n\r\nThat's why we need to be able to tell the software that there is no network to rely on even if there is one (good for testing too).\r\n\r\nSo what I'm thinking is that this is a simple matter of pre-ambling any network call wrappers with:\r\n\r\n```\r\nif HF_DATASETS_OFFLINE:\r\n assert \"Attempting to make a network call under Offline mode\"\r\n```\r\n\r\nand then fixing up if there is anything else to fix to make it work.\r\n\r\n--------------\r\n\r\nOtherwise I think the only other problem I encountered is that we need to find a way to pre-cache metrics, for some reason it's not caching it and wanting to fetch it from online.\r\n\r\nWhich is extra strange since it already has those files in the `datasets` repo itself that is on the filesystem.\r\n\r\nThe workaround I had to do is to copy `rouge/rouge.py` (with the parent folder) from the datasets repo to the current dir - and then it proceeded.", "Ok understand better the hanging issue.\r\nI guess catching connection errors is not enough, we should also avoid all the hangings.\r\nCurrently the offline mode tests are only done by simulating an instant connection fail that returns an error, let's have another connection mock that hangs instead.\r\n\r\nI'll also take a look at why you had to do this for `rouge`.\r\n", "FWIW, I think instant failure on the behalf of a network call is the simplest solution to correctly represent the environment and having the caller to sort it out is the next thing to do, since here it is the case of having no functional network, it's just that the software doesn't know this is the case, because there is some network. So we just need to help it to bail out instantly rather than hang waiting for it to time out. And afterwards everything else you said.", "Update on this: \r\n\r\nI managed to create a mock environment for tests that makes the connections hang until timeout.\r\nI managed to reproduce the issue you're having in this environment.\r\n\r\nI'll update the offline test cases to also test the robustness to connection hangings, and make sure we set proper timeouts where it's needed in the code. This should cover the _automatic_ section you mentioned.", "Fabulous! I'm glad you were able to reproduce the issues, @lhoestq!", "I lost access to the firewalled setup, but I emulated it with:\r\n\r\n```\r\nsudo ufw enable\r\nsudo ufw default deny outgoing\r\n```\r\n(thanks @mfuntowicz)\r\n\r\nI was able to test `HF_DATASETS_OFFLINE=1` and it worked great - i.e. didn't try to reach out with it and used the cached files instead.\r\n\r\nThank you!" ]

[]

[]

[]

[ { "id": 2107841032, "node_id": "MDU6TGFiZWwyMTA3ODQxMDMy", "url": "https://api.github.com/repos/huggingface/datasets/labels/nlp-viewer", "name": "nlp-viewer", "color": "94203D", "default": false, "description": "" } ]

[]

[ "Facing the same issue for [Squad](https://huggingface.co/datasets/viewer/?dataset=squad) and [TriviaQA](https://huggingface.co/datasets/viewer/?dataset=trivia_qa) datasets as well.", "We just fixed the issue, thanks for reporting !" ]

[]

[]

[ "Thanks ! could you maybe add a few tests in test_arrow_dataset.py to make sure from_pandas works as expected with categorical types ?\r\n\r\nIn particular I'm pretty sure that if you now try to `cast` the dataset to the same features at its current features, it will break instead of just being a no-op.\r\nThis is because `features.type` returns an arrow int64 type for the classlabel column instead of the arrow dictionary type that you have in the arrow table. There are two issues in this case:\r\n- it will try to replace the arrow type from dictionary to int64 instead of being a no-op\r\n- it will crash because pyarrow is not able to cast a dictionary to int64 (even if it's actually possible do cast the column by hand by accessing the sub-array of the dictionary array containing the indices/integers)\r\n\r\nIt would be awesome to fix this case ! Ideally the arrow `pa_type` of classlabel ([here](https://github.com/huggingface/datasets/blob/7072e1becd69d421d863374b825e3da4c6551798/src/datasets/features.py#L558)) should be an arrow dictionary type. This should fix the issue. Then we can start working on backward compatibility.\r\n\r\nLet me know if you have questions or if I can help.\r\nIn particular if there is some glue-ing to do I can take care of that if you want ;)\r\n\r\n--------------\r\n\r\nAlso just a few information regarding the functions you mentioned\r\n\r\n`int2str` and `str2int` are used by users to transforms the labels if they want to. Here sine ClassLabel is instantiated without the class names, they would crash. I was about to make a PR to disallow the creation of an empty ClassLabel feature type.\r\nTherefore can you provide class_names= when creating the ClassLabel ?\r\n\r\n`encode_example` is mostly used with a dataset builder (e.g. squad.py) so it's not used when using .from_pandas.\r\n\r\n\r\n", "Got it - that's super helpful, I was trying to figure out what would break!\r\n\r\nI think there are two issues we're discussing here:\r\n\r\n1. modifying the pa_type of ClassLabel: totally agree with you on that one if that's OK from a back-compat perspective. (i.e. are users of `datasets` not supposed to access or use the .pa_type attribute of ClassLabel?)\r\n2. creating a ClassLabel requires information that's not present on the pa.DictionaryType object: I think the crux of the problem is that at this line (https://github.com/huggingface/datasets/pull/1936/files#diff-54081ede051fd0a7ef65748c481cc06f90209f01bb89968747089d13a2ca052bR933) - you only have access to the `pa_type`, which is `DictionaryType[int8, string]`. I've unpacked it and looked at all of the available methods, and I don't believe that any of the actual values (\"names\") are present - those are stored on the `pyarrow.DictArray.dictionary` attribute (i.e. as data, not on the pyarrow.DataType) - so in order to actually be able to instantiate the ClassLabel with the names= parameter, we need to pass in more information to this method.\r\n\r\nWe *could* mostly accomplish this by modifying https://github.com/huggingface/datasets/pull/1936/files#diff-54081ede051fd0a7ef65748c481cc06f90209f01bb89968747089d13a2ca052bR909 to accept a pyarrow Table in addition to the type, and it's not too difficult to do, but it feels a little bit off to me:\r\n\r\n- It feels a bit off that a \"schema\" definition will change depending on what data gets added to the dataset. In particular, if someone adds rows or concatenates two datasets, the ClassLabel \"names\" will also need to change, right? I think maybe we're getting around this because a Dataset is immutable (I think?) and so any new dataset is freshly constructed, but for example - I think this check wouldn't work for `ClassLabel`s if we were to compare the `Dataset.features` instead of the underlying pyarrow type https://github.com/huggingface/datasets/blob/master/src/datasets/arrow_dataset.py#L2664\r\n- To that end I wonder if ClassLabel should actually just be the \"type\" akin to Category, and the \"names\" should be considered \"data\" and not part of the \"type\"? Similar to how pyarrow maintains two data objects - the array of indices and the array of string values.\r\n\r\nWith that in mind, I'm wondering if you *should* allow an empty ClassLabel (and`int2str`, etc. can be updated to have more descriptive error messages if labels aren't provided or inferred), and if the underlying data is a pa.DictionaryType, then the names can be inferred and applied at these points in the code:\r\n- https://github.com/huggingface/datasets/blob/96578adface7e4bc1f3e8bafbac920d72ca1ca60/src/datasets/arrow_dataset.py#L274\r\n- https://github.com/huggingface/datasets/blob/96578adface7e4bc1f3e8bafbac920d72ca1ca60/src/datasets/arrow_dataset.py#L686\r\n- https://github.com/huggingface/datasets/blob/96578adface7e4bc1f3e8bafbac920d72ca1ca60/src/datasets/arrow_dataset.py#L673\r\n\r\nI think perhaps the mismatch here is when the data is stored on disk as an int there should be a convenient way of saying \"this is a dictionary and here are some explicitly provided labels\", whereas when it's stored as a string, we'd ideally like to say \"this is a Category and please condense the representation and automatically infer the labels\".\r\n\r\nSorry for the long comment! Hopefully my thoughts make sense - thanks for taking the time to discuss!", "Yes that makes sense. I completely forgot that the label names of an arrow Dictionary type were not stored in the type but in the DictionaryArray.\r\n\r\nThis is made me realize that it's actually pretty unpractical and I feel that handling this can add unnecessary complexity in the handling of dtypes.\r\nMore specifically:\r\n- it's not possible to create a DictionaryArray from a call to pyarrow.array with python objects, which is the function we use to convert python objects to pyarrow objects (or we would need to convert the python objects to pandas categorical series beforehand but it doesn't work for nested types)\r\n- casting nested types containing Dictionary types would require a lot of array manipulations since it's not compatible with pyarrow.array.cast\r\n\r\nI feel like the original feature request (support of pandas Categorical) should be addressable without adding so much complexity to the library.\r\n\r\nIf we admit that we don't want to deal with arrow Dictionary type, maybe we can simply convert the pandas categorical series to an int64 series and set the feature type to the right ClassLabel in `from_pandas`. We can have the reverse operation in `to_pandas`. This way we don't need to support the arrow DictionaryType and so we can keep simple/accessible code for conversion from python to arrow and also for type casting. Let me know what you think.\r\n\r\nIn the future depending on the usage of the ClassLabel types with pandas/pyarrow we might reconsider this but for now I believe this simple solution is enough.", "I like that idea! Let me try working up a PR for this", "OK! I just whipped up the `from_pandas()` portion of this PR, and it works, though I'm not *super* familiar with the available APIs so I'm not sure if there's a more \"vectorized\" way of doing all of these updates - so happy to get some feedback and iterate!\r\n\r\nApologies for multiple commits - I realized how to solve a few different problems right after I gave up and pushed with the intent to ask for help :-)\r\n\r\nI wanted to get some guidance on how to handle the reverse direction: I think there are two main areas to look at, `.to_pandas()` and also `.set_format('pandas')` and then pulling out a dataframe like so: `dataset[:]`. Is there a single place where I can handle both of these cases at once or do these need to be handled independently?", "Thanks ! This is awesome :) \r\nCould you also add a test ? There is already `test_to_pandas` in test_arrow_dataset.py\r\nFeel free to complete this test to make sure it works for Categorical :)\r\n\r\nTo make it work with the \"pandas\" formating (when you do `set_format(\"pandas\")` and then query `dataset[0]`, `dataset[:]`, etc.), you can take a look and the `PandasFormatter` in formatting.py\r\nIt takes a pyarrow table as input of its formatting methods (one method for rows, one for columns and one for batches) and returns a pandas DataFrame (or a Series for the method for formatting a column). You can cast to Categorical in each one of the formatter methods and it should work directly when you use a pandas-formatted dataset.\r\n\r\nThis formatter can then also be used in `to_pandas` (currently it does `pa_table.to_pandas()` but `PandasFormatter().format_batch(pa_table)` can be used instead)." ]

[]

[]

[ "@patrickvonplaten \r\nI removed the mp3 files, dummy_data is much smaller now!" ]

[ { "id": 2067376369, "node_id": "MDU6TGFiZWwyMDY3Mzc2MzY5", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20request", "name": "dataset request", "color": "e99695", "default": false, "description": "Requesting to add a new dataset" } ]

[]

[ "Dataset added in release [1.5.0](https://github.com/huggingface/datasets/releases/tag/1.5.0), I think I can close this." ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Hi, it seems I am somewhat stuck here. The failed test `ci/circleci: run_dataset_script_tests_pyarrow_1_WIN` seems to be caused by some broken connection (`ConnectionResetError: [WinError 10054] An existing connection was forcibly closed by the remote host`). Any help on this is appreciated. \r\n\r\nEdit: Seems to be resolved now.", "I guess the `dummy_data.zip` is too large. I can reduce the languages that are contained there, but when testing it, it obviously throws an error, as not all files can be found. I guess I can either i) change the default value regarding which languages are loaded or ii) let the `_generate_examples` silently skip any language for which it cannot find files. Both solutions are not really pretty - is there another way around this?", "Thanks for the review and the constructive comments :) ! I tried to address them, and reduced the number of lines in the dummy data to 1 to reduce its size. " ]

[]

[]

[ "Hi @JieyuZhao ! Have you had a chance to add the different configurations ?\r\nThanks again for your help on this !", "> Hi @JieyuZhao ! Have you had a chance to add the different configurations ?\r\n> Thanks again for your help on this !\r\n\r\nHi @lhoestq Yes, I've updated the code. Now the configuration will have dev/test splits.", "> Cool thanks !\r\n> This looks perfect this way.\r\n> \r\n> Now we just need to update the dataset_infos.json (it contains the metadata of the dataset) and add dummy data to be able to test this script automatically.\r\n> \r\n> To update the dataset_infos.json you just need delete the current one at `./datasets/wino_biais/dataset_infos.json`, and then run this command:\r\n> \r\n> ```\r\n> datasets-cli test ./datasets/wino_biais --save_infos --all_configs --ignore_verifications\r\n> ```\r\n> \r\n> To add the dummy data there's also a tool to add them automatically.\r\n> First delete the folder at `./datasets/wino_biais/dummy` and then run\r\n> \r\n> ```\r\n> datasets-cli dummy_data ./datasets/wino_biais --auto_generate --match_text_files \"*conll\" --n_lines 15\r\n> ```\r\n> \r\n> Let me know if you have questions :)\r\n> Also don't forget to run `make style` to format the code properly.\r\n\r\nThanks for the instruction! I've updated the metadata and the dummy data and also do the formatting. Please let me know if more is needed. :)" ]

[]

[]

[ "@lhoestq Thanks for the quick review.", "I merged master to this branch to re-run the CI before merging :)" ]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Hi @lhoestq ,\r\n\r\nI am running into an issue now when trying to run RAG. Running exactly as described [here](https://huggingface.co/facebook/rag-token-nq?fbclid=IwAR3bTfhls5U_t9DqsX2Vzb7NhtRHxJxfQ-uwFT7VuCPMZUM2AdAlKF_qkI8#usage) I get the error below. Wondering if it's related to this.\r\n\r\nRunning Transformers 4.3.2 with datasets installed from source from `master` branch.\r\n\r\n```bash\r\n(venv) sergey_mkrtchyan datasets (master) $ python\r\nPython 3.8.6 (v3.8.6:db455296be, Sep 23 2020, 13:31:39)\r\n[Clang 6.0 (clang-600.0.57)] on darwin\r\nType \"help\", \"copyright\", \"credits\" or \"license\" for more information.\r\n>>> from transformers import RagTokenizer, RagRetriever, RagTokenForGeneration\r\n>>> tokenizer = RagTokenizer.from_pretrained(\"facebook/rag-token-nq\")\r\n>>> retriever = RagRetriever.from_pretrained(\"facebook/rag-token-nq\", index_name=\"exact\", use_dummy_dataset=True)\r\nUsing custom data configuration dummy.psgs_w100.nq.no_index-dummy=True,with_index=False\r\nReusing dataset wiki_dpr (/Users/sergey_mkrtchyan/.cache/huggingface/datasets/wiki_dpr/dummy.psgs_w100.nq.no_index-dummy=True,with_index=False/0.0.0/8a97e0f4fa5bc46e179474db6a61b09d5d2419d2911835bd3f91d110c936d8bb)\r\nUsing custom data configuration dummy.psgs_w100.nq.exact-50b6cda57ff32ab4\r\nReusing dataset wiki_dpr (/Users/sergey_mkrtchyan/.cache/huggingface/datasets/wiki_dpr/dummy.psgs_w100.nq.exact-50b6cda57ff32ab4/0.0.0/8a97e0f4fa5bc46e179474db6a61b09d5d2419d2911835bd3f91d110c936d8bb)\r\n 0%| | 0/10 [00:00<?, ?it/s]\r\nTraceback (most recent call last):\r\n File \"<stdin>\", line 1, in <module>\r\n File \"/Users/sergey_mkrtchyan/workspace/cformers/venv/lib/python3.8/site-packages/transformers/models/rag/retrieval_rag.py\", line 425, in from_pretrained\r\n return cls(\r\n File \"/Users/sergey_mkrtchyan/workspace/cformers/venv/lib/python3.8/site-packages/transformers/models/rag/retrieval_rag.py\", line 387, in __init__\r\n self.init_retrieval()\r\n File \"/Users/sergey_mkrtchyan/workspace/cformers/venv/lib/python3.8/site-packages/transformers/models/rag/retrieval_rag.py\", line 458, in init_retrieval\r\n self.index.init_index()\r\n File \"/Users/sergey_mkrtchyan/workspace/cformers/venv/lib/python3.8/site-packages/transformers/models/rag/retrieval_rag.py\", line 284, in init_index\r\n self.dataset = load_dataset(\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/load.py\", line 750, in load_dataset\r\n ds = builder_instance.as_dataset(split=split, ignore_verifications=ignore_verifications, in_memory=keep_in_memory)\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/builder.py\", line 734, in as_dataset\r\n datasets = utils.map_nested(\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/utils/py_utils.py\", line 195, in map_nested\r\n return function(data_struct)\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/builder.py\", line 769, in _build_single_dataset\r\n post_processed = self._post_process(ds, resources_paths)\r\n File \"/Users/sergey_mkrtchyan/.cache/huggingface/modules/datasets_modules/datasets/wiki_dpr/8a97e0f4fa5bc46e179474db6a61b09d5d2419d2911835bd3f91d110c936d8bb/wiki_dpr.py\", line 205, in _post_process\r\n dataset.add_faiss_index(\"embeddings\", custom_index=index)\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/arrow_dataset.py\", line 2516, in add_faiss_index\r\n super().add_faiss_index(\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/search.py\", line 416, in add_faiss_index\r\n faiss_index.add_vectors(self, column=column, train_size=train_size, faiss_verbose=faiss_verbose)\r\n File \"/Users/sergey_mkrtchyan/workspace/huggingface/datasets/src/datasets/search.py\", line 281, in add_vectors\r\n self.faiss_index.add(vecs)\r\n File \"/Users/sergey_mkrtchyan/workspace/cformers/venv/lib/python3.8/site-packages/faiss/__init__.py\", line 104, in replacement_add\r\n self.add_c(n, swig_ptr(x))\r\n File \"/Users/sergey_mkrtchyan/workspace/cformers/venv/lib/python3.8/site-packages/faiss/swigfaiss.py\", line 3263, in add\r\n return _swigfaiss.IndexHNSW_add(self, n, x)\r\nRuntimeError: Error in virtual void faiss::IndexHNSW::add(faiss::Index::idx_t, const float *) at /Users/runner/work/faiss-wheels/faiss-wheels/faiss/faiss/IndexHNSW.cpp:356: Error: 'is_trained' failed\r\n>>>\r\n```\r\n\r\nThe error message is hinting that it could be related to this, but I might be wrong. Any ideas?\r\n\r\n\r\nEdit: Can confirm it's working fine with datasets==1.2.0\r\n\r\nDouble Edit: Did some further digging. The issue is related to this commit: 8c5220307c33f00e01c3bf7b8. I opened a separate issue #1941 for proper tracking." ]

[]

[]

[ "Hi !\r\nI guess you can add a dataset without the fields that must be kept anonymous, and then update those when the anonymity period is over.\r\nYou can also make the PR from an anonymous org.\r\nPinging @yjernite just to make sure it's ok", "Hello,\r\nI would prefer to do the reverse: adding a link to an anonymous paper without the people names/institution in the PR. Would it be conceivable ?\r\nCheers\r\n", "Sure, I think it's ok on our side", "Yup, sounds good!" ]

[]

[]

[]

[]

[]

[ "Hi @JieyuZhao !\r\n\r\nYou can edit the dataset card of wino_bias to update the URL via a Pull Request. This would be really appreciated :)\r\n\r\nThe dataset card is the README.md file you can find at https://github.com/huggingface/datasets/tree/master/datasets/wino_bias\r\nAlso the homepage url is also mentioned in the wino_bias.py so feel free to update it there as well.\r\n\r\nYou can create a Pull Request directly from the github interface by editing the files you want and submit a PR, or from a local clone of the repository.\r\n\r\nThanks for noticing !" ]

[]

[]

[ "@lhoestq - apologies for the multiple PRs, my previous one (#1905) got mangled due to some merge conflicts that I had trouble resolving so I just cherry-picked my changes onto a fresh branch here." ]

[]

[]

[ "So I was curious why the issue reported at #1919 wasn't caught in [this test](https://github.com/huggingface/datasets/blob/248104c4bdb2e01c036b7578867199191fbff181/tests/test_arrow_dataset.py#L209), so I did some digging.\r\nI tried to save to a temporary directory (just like in the test), like this:\r\n```python\r\nwith tempfile.TemporaryDirectory() as requested_tempdir:\r\n squad.save_to_disk(requested_tempdir) # no error\r\n```\r\nand it executes succesfuly without problems.\r\nSo why does it work, but this doesn't?\r\n```python\r\nsquad.save_to_disk(\"./squad\") # error\r\n```\r\nIt's because `save_to_disk` also creates a temporary directory (let's call it `tempdir`), and since `tempdir` and `requested_tempdir` share the same parents, the `Path.joinpath` method [(here)](https://github.com/huggingface/datasets/blob/248104c4bdb2e01c036b7578867199191fbff181/src/datasets/arrow_dataset.py#L469) will keep `requested_tempdir` as it is and the *train* directory will be created under `requested_tempdir` and hence no errors will arise.\r\n\r\nBut in the second case (where we are saving to a local dir), the *train* directory is created under *squad* which in turn is created under `tempdir`, not under `.` (current dir).\r\n\r\nSo, all of this probably doesn't help solving the issue but it might help creating a better test, and it also makes me wonder why are we saving to a temporary dir in `save_to_disk` anyway? I mean, won't it be removed with all its contents upon execution completion? what's the point then? ", "CLosing in favor of #1923" ]

[]

[]

[ "Hi thanks for reporting and for proposing a fix :)\r\n\r\nI just merged a fix, feel free to try it from the master branch !", "Closing since this has been fixed by #1923" ]

[]

[]

[]

[]

[]

[ "upgraded to php 3.9.2 and it works!" ]

[]

[]

[ "Hmmm this one broke master. I'm fixing it.\r\n\r\nMaybe because your branch was outdated ?", "Sorry @lhoestq, I forgot to update the imports... :/", "It's fine, the CI should have caught this tbh. Not sure why it did't fail" ]

[]

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

[ "Thanks for reporting ! This is a bug. For now feel free to set `ignore_verifications=False` in `load_dataset`.\r\nI'm working on a fix", "I just merged a fix :)\r\n\r\nWe'll do a patch release soon. In the meantime feel free to try it from the master branch\r\nThanks again for reporting !", "Closing since this has been fixed by #1925" ]

[]

[]

[]

[]

[]

[ "Just so I understand how it can be used in practice, do you have an example showing how to load a text dataset with this option?", "Sure ! Here is an example:\r\n```python\r\nfrom datasets import load_dataset\r\n\r\nload_dataset(\"text\", keep_linebreaks=True, data_files=...)\r\n```\r\n\r\nI'll update the documentation to explain this", "Perfect!" ]

[]

[]

[ "So much better - thank you for doing that, @lhoestq!", "Also fixed the `uncorpus` urls for wmt19 ru-en and zh-en for https://github.com/huggingface/datasets/issues/1893", "Thanks!\r\nCan this be merged sooner? \r\nI manually update it and it works well." ]

[]

[]

[ "@thomwolf @lhoestq can you guys please take a look and recommend some solution.", "am suspicious of this thing? what's the purpose of this? pickling and unplickling\r\n`self = pickle.loads(pickle.dumps(self))`\r\n\r\n```\r\n def save_to_disk(self, dataset_path: str, fs=None):\r\n \"\"\"\r\n Saves a dataset to a dataset directory, or in a filesystem using either :class:`datasets.filesystem.S3FileSystem` or any implementation of ``fsspec.spec.AbstractFileSystem``.\r\n\r\n Args:\r\n dataset_path (``str``): path (e.g. ``dataset/train``) or remote uri (e.g. ``s3://my-bucket/dataset/train``) of the dataset directory where the dataset will be saved to\r\n fs (Optional[:class:`datasets.filesystem.S3FileSystem`,``fsspec.spec.AbstractFileSystem``], `optional`, defaults ``None``): instance of :class:`datasets.filesystem.S3FileSystem` or ``fsspec.spec.AbstractFileSystem`` used to download the files from remote filesystem.\r\n \"\"\"\r\n assert (\r\n not self.list_indexes()\r\n ), \"please remove all the indexes using `dataset.drop_index` before saving a dataset\"\r\n self = pickle.loads(pickle.dumps(self))\r\n ```", "It's been 24 hours and sadly it's still running. With not a single byte written", "Tried finding the root cause but was unsuccessful.\r\nI am using lazy tokenization with `dataset.set_transform()`, it works like a charm with almost same performance as pre-compute.", "Hi ! This very probably comes from the hack you used.\r\n\r\nThe pickling line was added an a sanity check because save_to_disk uses the same assumptions as pickling for a dataset object. The main assumption is that memory mapped pyarrow tables must be reloadable from the disk. In your case it's not possible since you altered the pyarrow table.\r\nI would suggest you to rebuild a valid Dataset object from your new pyarrow table. To do so you must first save your new table to a file, and then make a new Dataset object from that arrow file.\r\n\r\nYou can save the raw arrow table (without all the `datasets.Datasets` metadata) by calling `map` with `cache_file_name=\"path/to/outut.arrow\"` and `function=None`. Having `function=None` makes the `map` write your dataset on disk with no data transformation.\r\n\r\nOnce you have your new arrow file, load it with `datasets.Dataset.from_file` to have a brand new Dataset object :)\r\n\r\nIn the future we'll have a better support for the fast filtering method from pyarrow so you don't have to do this very unpractical workaround. Since it breaks somes assumptions regarding the core behavior of Dataset objects, this is very discouraged.", "Thanks, @lhoestq for your response. Will try your solution and let you know." ]

[]

[]

[ "It looks like this PR now includes changes to many other files than the ones for CoNLLpp.\r\n\r\nTo fix that feel free to create another branch and another PR.\r\n\r\nThis was probably caused by a git rebase. You can avoid this issue by using git merge if you've already pushed your branch." ]

[]

[]

[ "Hi ! :)\r\n\r\nThis looks like the same issue as https://github.com/huggingface/datasets/issues/1856 \r\nBasically google drive has quota issues that makes it inconvenient for downloading files.\r\n\r\nIf the quota of a file is exceeded, you have to wait 24h for the quota to reset (which is painful).\r\n\r\nThe error says that the checksum of the downloaded file doesn't match because google drive returns a text file with the \"Quota Exceeded\" error instead of the actual data file.", "Thanks @lhoestq! Yes, it seems back to normal after a couple of days." ]

[ { "id": 1935892871, "node_id": "MDU6TGFiZWwxOTM1ODkyODcx", "url": "https://api.github.com/repos/huggingface/datasets/labels/enhancement", "name": "enhancement", "color": "a2eeef", "default": true, "description": "New feature or request" }, { "id": 2067400324, "node_id": "MDU6TGFiZWwyMDY3NDAwMzI0", "url": "https://api.github.com/repos/huggingface/datasets/labels/generic%20discussion", "name": "generic discussion", "color": "c5def5", "default": false, "description": "Generic discussion on the library" } ]

[]

[ "We already have a ClassLabel type that does this kind of mapping between the label ids (integers) and actual label values (strings).\r\n\r\nI wonder if actually we should use the DictionaryType from Arrow and the Categorical type from pandas for the `datasets` ClassLabel feature type.\r\nCurrently ClassLabel corresponds to `pa.int64()` in pyarrow and `dtype('int64')` in pandas (so the label names are lost during conversions).\r\n\r\nWhat do you think ?", "Now that I've heard you explain ClassLabel, that makes a lot of sense! While DictionaryType for Arrow (I think) can have arbitrarily typed keys, so it won't cover all potential cases, pandas' Category is *probably* the most common use for that pyarrow type, and ClassLabel should match that perfectly?\r\n\r\nOther thoughts:\r\n\r\n- changing the resulting patype on ClassLabel might be backward-incompatible? I'm not totally sure if users of the `datasets` library tend to directly access the `patype` attribute (I don't think we really do, but we haven't been using it for very long yet).\r\n- would ClassLabel's dtype change to `dict[int64, string]`? It seems like in practice a ClassLabel (when not explicitly specified) would be constructed from the DictionaryType branch of `generate_from_arrow_type`, so it's not totally clear to me that anyone ever actually accesses/uses that dtype?\r\n- I don't quite know how `.int2str` and `.str2int` are used in practice - would those be kept? Perhaps the implementation might actually be substantially smaller if we can just delegate to pyarrow's dict methods?\r\n\r\nAnother idea that just occurred to me: add a branch in here to generate a ClassLabel if the dict key is int64 and the values are string: https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L932 , and then don't touch anything else.\r\n\r\nIn practice, I don't think this would be backward-incompatible in a way anyone would care about since the current behavior just throws an exception, and this way, we could support *reading* a pandas Categorical into a `Dataset` as a ClassLabel. I *think* from there, while it would require some custom glue it wouldn't be too hard to convert the ClassLabel into a pandas Category if we want to go back - I think this would improve on the current behavior without risking changing the behavior of ClassLabel in a backward-incompat way.\r\n\r\nThoughts? I'm not sure if this is overly cautious. Whichever approach you think is better, I'd be happy to take it on!\r\n", "I think we can first keep the int64 precision but with an arrow Dictionary for ClassLabel, and focus on the connection with arrow and pandas.\r\n\r\nIn this scope, I really like the idea of checking for the dictionary type:\r\n\r\n> Another idea that just occurred to me: add a branch in here to generate a ClassLabel if the dict key is int64 and the values are string: https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L932 , and then don't touch anything else.\r\n\r\nThis looks like a great start.\r\n\r\nThen as you said we'd have to add the conversion from classlabel to the correct arrow dictionary type. Arrow is already able to convert from arrow Dictionary to pandas Categorical so it should be enough.\r\n\r\nI can see two things that we must take case of to make this change backward compatible:\r\n- first we must still be able to load an arrow file with arrow int64 dtype and `datasets` ClassLabel type without crashing. This can be fixed by casting the arrow int64 array to an arrow Dictionary array on-the-fly when loading the table in the ArrowReader.\r\n- then we still have to return integers when accessing examples from a ClassLabel column. Currently it would return the strings values since it's based on the pandas behavior for converting from pandas to python/numpy. To do so we just have to adapt the python/numpy extractors in formatting.py (it takes care of converting an arrow table to a dictionary of python objects by doing arrow table -> pandas dataframe -> python dictionary)\r\n\r\nAny help on this matter is very much welcome :)" ]

[]

[]

[ "Also - I took a stab at updating the docs, but I'm not sure how to actually check the outputs to see if it's formatted properly." ]

[]

[]

[ "Thanks!" ]

[]

[]

[ "@patrickvonplaten could you please review and help me close this PR?", "@lhoestq Thank you so much for your comments and for patiently reviewing the code. Have _hopefully_ included all the suggested changes. Let me know if any more changes are required.\r\n\r\nSorry the code had lots of silly errors from my side!:' Will be more careful from next time! :)\r\n\r\n\r\n" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "OK! Thank you for the review - I will follow up with a separate PR for the comments here (https://github.com/huggingface/datasets/pull/1900#discussion_r578319725)!" ]

[]

[]

[]

[ { "id": 2067388877, "node_id": "MDU6TGFiZWwyMDY3Mzg4ODc3", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20bug", "name": "dataset bug", "color": "2edb81", "default": false, "description": "A bug in a dataset script provided in the library" } ]

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

[ "Thanks for reporting. This looks like a very bad issue. I'm looking into it", "I just merged a fix, we'll do a patch release soon. Thanks again for reporting, and sorry for the inconvenience.\r\nIn the meantime you can load `ALT` using `datasets` from the master branch", "Thanks!!! works perfectly in the bleading edge master version", "Closed by #1899" ]

[]

[]

[]

[]

[]

[ "Thanks for reporting !\r\n\r\nYou're right, `string_to_arrow` should be able to take `\"timestamp[ns]\"` as input and return the right pyarrow timestamp type.\r\nFeel free to suggest a fix for `string_to_arrow` and open a PR if you want to contribute ! This would be very appreciated :)\r\n\r\nTo give you more context:\r\n\r\nAs you may know we define the features types of a dataset using the `Features` object in combination with feature types like `Value`. For example\r\n```python\r\nfeatures = Features({\r\n \"age\": Value(\"int32\")\r\n})\r\n```\r\nHowever under the hood we are actually using pyarrow to store the data, and so we have a mapping between the feature types of `datasets` and the types of pyarrow.\r\n\r\nFor example, the `Value` feature types are created from a pyarrow type with `Value(str(pa_type))`.\r\nHowever it looks like the conversion back to a pyarrow type doesn't work with `\"timestamp[ns]\"`.\r\nThis is the `string_to_arrow` function you highlighted that does this conversion, so we should fix that.\r\n\r\n", "Thanks for the clarification @lhoestq !\r\n\r\nThis may be a little bit of a stupid question, but I wanted to clarify one more thing before I took a stab at this:\r\n\r\nWhen the features get inferred, I believe they already have a pyarrow schema (https://github.com/huggingface/datasets/blob/master/src/datasets/arrow_dataset.py#L234).\r\n\r\nWe then convert it to a string (https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L778) only to convert it back into the arrow type (https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L143, and https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L35). Is there a reason for this round-trip?\r\n\r\nI'll open a PR later to add `timestamp` support to `string_to_arrow`, but I'd be curious to understand since it feels like there may be some opportunities to simplify!", "The objective in terms of design is to make it easy to create Features in a pythonic way. So for example we use a string to define a Value type.\r\nThat's why when inferring the Features from an arrow schema we have to find the right string definitions for Value types. I guess we could also have a constructor `Value.from_arrow_type` to avoid recreating the arrow type, but this could create silent errors if the pyarrow type doesn't have a valid mapping with the string definition. The \"round-trip\" is used to enforce that the ground truth is the string definition, not the pyarrow type, and also as a sanity check.\r\n\r\nLet me know if that makes sense ", "OK I think I understand now:\r\n\r\nFeatures are datasets' internal representation of a schema type, distinct from pyarrow's schema.\r\nValue() corresponds to pyarrow's \"primitive\" types (e.g. `int` or `string`, but not things like `list` or `dict`).\r\n`get_nested_type()` (https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L698) and `generate_from_arrow_type()` (https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L778) *should* be inverses of each other, and similarly, for the primitive values, `string_to_arrow()` and `Value.__call__` (https://github.com/huggingface/datasets/blob/master/src/datasets/features.py#L146) should be inverses of each other?\r\n\r\nThanks for taking the time to answer - I just wanted to make sure I understood before opening a PR so I'm not disrupting anything about how the codebase is expected to work!", "Yes you're totally right :)" ]

[]

[]

[ "Hi sam !\r\nIndeed we can expect the performances to be very close since both MMapIndexedDataset and the `datasets` implem use memory mapping. With memory mapping what determines the I/O performance is the speed of your hard drive/SSD.\r\n\r\nIn terms of performance we're pretty close to the optimal speed for reading text, even though I found recently that we could still slightly improve speed for big datasets (see [here](https://github.com/huggingface/datasets/issues/1803)).\r\n\r\nIn terms of number of examples and example sizes, the only limit is the available disk space you have.\r\n\r\nI haven't used `psrecord` yet but it seems to be a very interesting tool for benchmarking. Currently for benchmarks we only have github actions to avoid regressions in terms of speed. But it would be cool to have benchmarks with comparisons with other dataset tools ! This would be useful to many people", "Also I would be interested to know what data types `MMapIndexedDataset` supports. Is there some documentation somewhere ?", "no docs haha, it's written to support integer numpy arrays.\r\n\r\nYou can build one in fairseq with, roughly:\r\n```bash\r\n\r\nwget https://s3.amazonaws.com/research.metamind.io/wikitext/wikitext-103-raw-v1.zip\r\nunzip wikitext-103-raw-v1.zip\r\nexport dd=$HOME/fairseq-py/wikitext-103-raw\r\n\r\nexport mm_dir=$HOME/mmap_wikitext2\r\nmkdir -p gpt2_bpe\r\nwget -O gpt2_bpe/encoder.json https://dl.fbaipublicfiles.com/fairseq/gpt2_bpe/encoder.json\r\nwget -O gpt2_bpe/vocab.bpe https://dl.fbaipublicfiles.com/fairseq/gpt2_bpe/vocab.bpe\r\nwget -O gpt2_bpe/dict.txt https://dl.fbaipublicfiles.com/fairseq/gpt2_bpe/dict.txt\r\nfor SPLIT in train valid; do \\\r\n python -m examples.roberta.multiprocessing_bpe_encoder \\\r\n --encoder-json gpt2_bpe/encoder.json \\\r\n --vocab-bpe gpt2_bpe/vocab.bpe \\\r\n --inputs /scratch/stories_small/${SPLIT}.txt \\\r\n --outputs /scratch/stories_small/${SPLIT}.bpe \\\r\n --keep-empty \\\r\n --workers 60; \\\r\ndone\r\n\r\nmkdir -p $mm_dir\r\nfairseq-preprocess \\\r\n --only-source \\\r\n --srcdict gpt2_bpe/dict.txt \\\r\n --trainpref $dd/wiki.train.bpe \\\r\n --validpref $dd/wiki.valid.bpe \\\r\n --destdir $mm_dir \\\r\n --workers 60 \\\r\n --dataset-impl mmap\r\n```\r\n\r\nI'm noticing in my benchmarking that it's much smaller on disk than arrow (200mb vs 900mb), and that both incur significant cost by increasing the number of data loader workers. \r\nThis somewhat old [post](https://ray-project.github.io/2017/10/15/fast-python-serialization-with-ray-and-arrow.html) suggests there are some gains to be had from using `pyarrow.serialize(array).tobuffer()`. I haven't yet figured out how much of this stuff `pa.Table` does under the hood.\r\n\r\nThe `MMapIndexedDataset` bottlenecks we are working on improving (by using arrow) are:\r\n1) `MMapIndexedDataset`'s index, which stores offsets, basically gets read in its entirety by each dataloading process.\r\n2) we have separate, identical, `MMapIndexedDatasets` on each dataloading worker, so there's redundancy there; we wonder if there is a way that arrow can somehow dedupe these in shared memory.\r\n\r\nIt will take me a few hours to get `MMapIndexedDataset` benchmarks out of `fairseq`/onto a branch in this repo, but I'm happy to invest the time if you're interested in collaborating on some performance hacking." ]

[ { "id": 2067388877, "node_id": "MDU6TGFiZWwyMDY3Mzg4ODc3", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20bug", "name": "dataset bug", "color": "2edb81", "default": false, "description": "A bug in a dataset script provided in the library" } ]

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

[ "This was also mentioned in https://github.com/huggingface/datasets/issues/488 \r\n\r\nThe bucket where is data was stored seems to be unavailable now. Maybe we can change the URL to the ones in https://conferences.unite.un.org/uncorpus/en/downloadoverview ?", "Closing since this has been fixed by #1912" ]

[]

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

[ "Yes that would be awesome. Not only the download speeds are awful, but also some files are missing.\r\nWe list all the URLs in the datasets/wmt19/wmt_utils.py so we can make a script to download them all and host on S3.\r\nAlso I think most of the materials are under the CC BY-NC-SA 3.0 license (must double check) so it should be possible to redistribute the data with no issues.\r\n\r\ncc @patrickvonplaten who knows more about the wmt scripts", "Yeah, the scripts are pretty ugly! A big refactor would make sense here...and I also remember that the datasets were veeery slow to download", "I'm downloading them.\r\nI'm starting with the ones hosted on http://data.statmt.org which are the slowest ones", "@lhoestq better to use our new git-based system than just raw S3, no? (that way we have built-in CDN etc.)", "Closing since the urls were changed to mirror urls in #1912 ", "Hi there! What about mirroring other datasets like [CCAligned](http://www.statmt.org/cc-aligned/) as well? All of them are really slow to download..." ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Next step is going to add these two in the documentation ^^" ]

[]

[]

[ "Close #1872" ]

[]

[]

[ "@lhoestq I stumbled upon an interesting failure when adding tests for CSV serialization of `ArrayXD` features (see the failing unit tests in the CI)\r\n\r\nIt's due to the fact that booleans cannot be converted from arrow format to numpy without copy: https://arrow.apache.org/docs/python/generated/pyarrow.Array.html#pyarrow.Array.to_numpy", "Good catch ! I must be able to fix that one by allowing copies for this kind of arrays.\r\nThis is the kind of surprise you get sometimes when playing with arrow x)", "Raising this error for booleans was introduced in https://issues.apache.org/jira/browse/ARROW-2871?jql=text%20~%20%22boolean%20to_numpy%22 without much explanations unfortunately.\r\nSo \"no copy\" only works for primitive types - except booleans.\r\nThis is confirmed in the source code at https://github.com/wesm/arrow/blob/c07b9b48cf3e0bbbab493992a492ae47e5b04cad/python/pyarrow/array.pxi#L621\r\n\r\nI'm opening a PR to allow copies for booleans...", "I just merged the fix for boolean ArrayXD, feel free to merge from master to see if it fixes the ci :)", "@lhoestq unfirtunately, arrays of strings (or any other non-primitive type) require a copy too\r\n\r\nA list of primitive types can be found here: https://github.com/wesm/arrow/blob/c07b9b48cf3e0bbbab493992a492ae47e5b04cad/python/pyarrow/types.pxi#L821\r\n\r\npyarrow provides a `is_primitive` function to check whether a type is primitive , I used it to set `zero_copy_only`\r\n\r\nAlso, `PandasArrayExtensionArray.isna` was using `numpy.isnan` which fails for arrays of strings. I replaced it with `pandas.isna`. Let me know what you think! :) " ]

[]

[]

[ "Does it make sense to make the domains as the different languages?\r\nA problem is that you need to download the datasets from the browser.\r\nOne idea would be to either contact Mozilla regarding API access to the dataset or make use of a headless browser for downloading the datasets (might be hard since we have to figure out how to host them). An even more creative idea would be to host the dataset inside a torrent and figure out a way to download specific datasets from within that torrent.\r\n\r\nHere is some information about the download authorization. They are hosting the data on S3.\r\n\r\nhttps://docs.aws.amazon.com/AmazonS3/latest/API/sigv4-auth-using-authorization-header.html\r\n\r\nHere is an example of how a download link looks.\r\n\r\nhttps://mozilla-common-voice-datasets.s3.dualstack.us-west-2.amazonaws.com/cv-corpus-6.1-2020-12-11/nl.tar.gz?X-Amz-Algorithm=AWS4-HMAC-SHA256&X-Amz-Credential=ASIAQ3GQRTO3ND4UAQXB%2F20210217%2Fus-west-2%2Fs3%2Faws4_request&X-Amz-Date=20210217T080740Z&X-Amz-Expires=43200&X-Amz-Security-Token=FwoGZXIvYXdzEGIaDCC6ALh%2FwIK9ovvRdCKSBCs5WaSJNsZ2h0SnhpnWFv4yiAJHJTe%2BY6pBcCqadRMs0RABHeQ2n1QDACJ5V9WOqIHfMfT0AI%2Bfe6iFkTGLgRrJOMYpgV%2FmIBcXCjeb72r4ZvudMA8tprkSxZsEh53bJkIDQx1tXqfpz0yoefM0geD3461suEGhHnLIyiwffrUpRg%2BkNZN9%2FLZZXpF5F2pogieKKV533Jetkd1xlWOR%2Bem9R2bENu2RV563XX3JvbWxSYN9IHkVT1xwd4ZiOpUtX7%2F2RoluJUKw%2BUPpyml3J%2FOPPGdr7CyPLjqNxdq9ceRi8lRybty64XvNYZGt45VNTQ3pkTTz4VpUCJAGkgxq95Ve%2BOwW%2Fsc8JtblTFKrH11vej62NB7C0n7JPPS4SLKXHKW%2B7ZbybcNf3BnsAVouPdsGTMslcgkD81b9trnjyXJdOZkzdHUf2KcWVXVceEsZnMhcCZQ1cJpI7qXPEk8QrKCQcNByPLHmPIEdHpj9IrIBKDkl2qO7VX7CCB65WDt2eZRltOcNHXWVFXFktMdQOQztI1j0XSZz2iOX4jPKKaqz193VEytlAqmehNi8pePOnxkP9Z1SP7d3I6rayuBF3phmpHxw499tY3ECYYgoCnJ6QSFa3KxMjFmEpQlmjxuwEMHd4CDL2FJYGcCiIxbCcL1r8ZE3%2BbGdcu7PRsVCHX3Huh%2FqGIaF4h40FgteN6teyKCHKOebs4EGMipb9xmEMZ9ZbVopz4bkhLdMTrjKon9w624Xem0MTPqN7XY%2BB6lRgrW8rd4%3D&X-Amz-Signature=28eabdfce72a472a70b0f9e1e2c37fe1471b5ec8ed60614fbe900bfa97ae1ac8&X-Amz-SignedHeaders=host\r\n\r\nIt could be that we simply need to make a http-request with the right parameters and we can download the datasets.", "> Wow, this looks great already! It's really a difficult dataset so thanks a lot for opening a PR.\r\n> I think the tagging tool is not too important for now and we can take a look at that later!\r\n> \r\n> At the moment, it would be very good to correctly generate some dummy data for all the possible languages. I think the structure of the `.tsv` file as you've noted in the PR is the one we want to use as the structure for `features = datasets.Features(`\r\n> \r\n> The splits `'Train\"`, `\"Test\"`, `\"Validation\"` look great to me! Because this is a special dataset that also has files called `\"Invalidated\"` I think the best option is to also add those as splits, _i.e._ `\"other\"`, `\"invalidated\"`, `\"reported\"`, `\"validated\"` . Those split names can be gives as shown here for example:\r\n> \r\n> https://github.com/huggingface/datasets/blob/28be129db862ec89a87ac9349c64df6b6118aff4/datasets/librispeech_asr/librispeech_asr.py#L124\r\n> \r\n> Also putting @lhoestq in cc here to hear his opinion on the different splits. @lhoestq Common Voicie is a crowd collected dataset where if a collected data sample did not receive enough \"up_votes\" from the community -> then it is (If I understood it correctly) marked as invalid -> hence the file `\"invalidated.tsv\"`. I think this is still useful data, so I would include it what do you think?\r\n> \r\n> @BirgerMoell let me know if you have any more questions :-)\r\n\r\nI think reporting is a separate feature. People can help annotate the data and then they can report things while annotating.\r\nhttps://commonvoice.mozilla.org/sv-SE/listen\r\n\r\nHere is the interface that shows reporting and the thumbs up and down which gives upvotes and downvotes.\r\n<img src=\"https://i.imgur.com/utWjszt.png\" height=\"800px\">\r\n", "I added splits and features. I'm not sure how you want me to generate dummy data for all the languages?", "Hey @BirgerMoell,\r\n\r\nI tweaked your dataset file a bit to have a first working version. To test this dataset downloading script, you can do the following:\r\n\r\n- 1) Download the Common Voice Georgian dataset from https://commonvoice.mozilla.org/en/datasets (It's pretty small which is why I chose it)\r\n- 2) Run the following command using this branch: \r\n\r\n```python\r\nfrom datasets import load_dataset\r\n\r\nds = load_dataset(\"./../datasets/datasets/common_voice\", \"Georgian\", data_dir=\"./cv-corpus-6.1-2020-12-11/ka/\", split=\"train\")\r\n```\r\n\r\nNote that I'm loading a local version of the dataset script (`\"./../datasets/datasets/common_voice/\"` points to the folder in your branch) and that I also insert the downloaded data with the `data_dir` arg.\r\n\r\n-> You'll see that the data is correctly loaded and that `ds` contains all the information we need.\r\n\r\nNow there are a lot of different datasets on Common Voice, so it probably takes too much time to test all of those, but maybe you can test whether the current script works as well *e.g.* for Swedish, 3,4 other languages.\r\n\r\nIt would be very nice if we can use the exact same structure for all languages, meaning that we don't have to change the `datasets.Features(...)` structure depending on the language, but can use the exact same one for every language.\r\n\r\nIf everything works as expected we can then go over to cleaning the script and seeing how to add dummy data tests for it." ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "@lhoestq I am not sure how to test `dictionary_encode_column` (in-place version was not tested before)", "I can take a look at dictionary_encode_column tomorrow.\r\nAlthough it's likely that it doesn't work then. It was added at the beginning of the lib and never tested nor used afaik.", "Now let's update the documentation to use the new methods x)" ]

[]

[]

[ "@lhoestq I have refactorized the logic. Instead of the previous hierarchy call (local temp file opening -> remote call -> use again temp local file logic but from within the remote caller scope), now it is flattened. Schematically:\r\n```python\r\nwith src.open() as src_file, dst.open() as dst_file:\r\n src_file.fetch(dst_file)\r\n```\r\n\r\nI have created `RemotePath` (analogue to Path) with method `.open()` that returns `FtpFile`/`HttpFile` (analogue to file-like).\r\n\r\nNow I am going to implement `RemotePath.exists()` method (analogue to the Path's method) to check if remote resource is accessible, using `Ftp/Http.head()`.", "Quick update on this one:\r\nwe discussed offline with @albertvillanova on this PR and I think using `fsspec` can help a lot, since it already implements many parts of the abstraction we need to have nice download tools for both http and ftp (and others !)" ]

[]

[]

[]

[]

[]

[]

[]

[]

[ "Hi @lhoestq. If you agree to merge this, I will start separating the logic for NestedDataStructure.map ;)" ]

[]

[]

[ "Hey @anton-l,\r\n\r\nThanks a lot for the very clean integration!\r\n\r\n1) I think we should now start having \"automatic-speech-recognition\" as a label in the dataset tagger (@yjernite is it easy to add?). But we can surely add this dataset with the tag you've added and then later change the label to `asr` \r\n\r\n2) That's perfect! Yeah good question - we're currently thinking about a better design with @lhoestq \r\n\r\n3) Again tagging @yjernite & @lhoestq here - guess we should add this license though!", "Thanks @anton-l for adding this one :)\r\nAbout the points you mentioned:\r\n1. Sure as soon as we've updated the tag sets in https://github.com/huggingface/datasets-tagging/blob/main/task_set.json, we can update the tags in this dataset card and also in the other audio dataset card.\r\n2. For now we just try to have them as small as possible but we may switch to S3/LFS at one point indeed\r\n3. If it's not part of the license set at https://github.com/huggingface/datasets-tagging/blob/main/license_set.json we can add it to this license set\r\n\r\nFor now it's ok to have the other-* tags but we'll update them very soon", "Let's merge this one and then we'll update the tags for the audio datasets. We'll probably also add something like this:\r\n```\r\ntype:\r\n- text\r\n- audio\r\n```\r\n\r\nThank you so much for adding this one, good job !" ]

[]

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

[ "I started working on this. My idea is to first add the pyarrow Table wrappers InMemoryTable and MemoryMappedTable that both implement what's necessary regarding copy/pickle. Then have another wrapper that takes the concatenation of InMemoryTable/MemoryMappedTable objects.\r\n\r\nWhat's important here is that concatenating two tables into one doesn't double the memory used (`total_allocated_bytes()` stays the same).", "Hi @lhoestq @albertvillanova,\r\n\r\nI checked the linked issues and PR, this seems like a great idea. Would you mind elaborating on the in-memory and memory-mapped datasets? \r\nBased on my understanding, it is something like this, please correct me if I am wrong:\r\n1. For in-memory datasets, we don't have any dataset files so the entire dataset is pickled to the cache during loading, and then whenever required it is unpickled .\r\n2. For on-disk/memory-mapped datasets, we have the data files provided, so they can be re-loaded from the paths, and only the file-paths are stored while pickling.\r\n\r\nIf this is correct, will the feature also handle pickling/unpickling of a concatenated dataset? Will this be cached?\r\n\r\nThis also leads me to ask whether datasets are chunked during pickling? \r\n\r\nThanks,\r\nGunjan", "Hi ! Yes you're totally right about your two points :)\r\n\r\nAnd in the case of a concatenated dataset, then we should reload each sub-table depending on whether it's in-memory or memory mapped. That means the dataset will be made of several blocks in order to keep track of what's from memory and what's memory mapped. This allows to pickle/unpickle concatenated datasets", "Hi @lhoestq\r\n\r\nThanks, that sounds nice. Can you explain where the issue of the double memory may arise? Also, why is the existing `concatenate_datasets` not sufficient for this purpose?", "Hi @lhoestq,\r\n\r\nWill the `add_item` feature also help with lazy writing (or no caching) during `map`/`filter`?", "> Can you explain where the issue of the double memory may arise?\r\n\r\nWe have to keep each block (in-memory vs memory mapped) separated in order to be able to reload them with pickle.\r\nOn the other hand we also need to have the full table from mixed in-memory and memory mapped data in order to iterate or extract data conveniently. That means that each block is accessible twice: once in the full table, and once in the separated blocks. But since pyarrow tables concatenation doesn't double the memory, then building the full table doesn't cost memory which is what we want :)\r\n\r\n> Also, why is the existing concatenate_datasets not sufficient for this purpose?\r\n\r\nThe existing `concatenate_datasets` doesn't support having both in-memory and memory mapped data together (there's no fancy block separation logic). It works for datasets fully in-memory or fully memory mapped but not a mix of the two.\r\n\r\n> Will the add_item feature also help with lazy writing (or no caching) during map/filter?\r\n\r\nIt will enable the implementation of the fast, masked filter from this discussion: https://github.com/huggingface/datasets/issues/1949\r\nHowever I don't think this will affect map." ]

[]

[]

[ "Thanks for reporting !\r\nThis is due to the changes made in the data files in the multiwoz repo: https://github.com/budzianowski/multiwoz/pull/59\r\nI'm opening a PR to update the checksums of the data files.", "I just merged the fix. It will be available in the new release of `datasets` later today.\r\nYou'll be able to get the new version with\r\n```\r\npip install --upgrade datasets\r\n```", "Hi, I still meet the error when loading the datasets after upgradeing datasets.\r\n\r\nraise NonMatchingChecksumError(error_msg + str(bad_urls))\r\ndatasets.utils.info_utils.NonMatchingChecksumError: Checksums didn't match for dataset source files:\r\n['https://github.com/budzianowski/multiwoz/raw/master/data/MultiWOZ_2.2/dialog_acts.json', 'https://github.com/budzianowski/multiwoz/raw/master/data/MultiWOZ_2.2/test/dialogues_001.json']", "This must be related to https://github.com/budzianowski/multiwoz/pull/72\r\nThose files have changed, let me update the checksums for this dataset.\r\n\r\nFor now you can use `ignore_verifications=True` in `load_dataset` to skip the checksum verification." ]