Datasets:

---

Fancellu

/

github-issues

like 0

[]

[]

[ "Version 2.3.2 is over one year old, so please use the latest release (2.14.0) to get the expected behavior. Version 2.3.2 does not contain some fixes we made to fix resolving hidden files/directories (starting with a dot)." ]

[]

[]

[ "Hi ! Thanks for reporting, I opened a PR to fix this\r\n\r\nWhat filesystem are you using ?", "Hi @lhoestq ! Thank you so much 🙌 \r\n\r\nIt's a bit of a custom setup, but in practice I am using a [pyarrow.fs.S3FileSystem](https://arrow.apache.org/docs/python/generated/pyarrow.fs.S3FileSystem.html) (wrapped in a `fsspec.implementations.arrow.ArrowFSWrapper` [to make it](https://arrow.apache.org/docs/python/filesystems.html#using-arrow-filesystems-with-fsspec) `fsspec` compatible). I also register it as an entrypoint with `fsspec` so that it's the one that gets automatically resolved when looking for filesystems for the `s3` protocol\r\n\r\nIn my case the `storage_option` that seemed not getting piped through was the filesystem's `endpoint_override` that I use in some tests to point at a mock S3 bucket" ]

[]

[]

[ "You can list the dataset's columns with `ds.column_names` before `.map` to check whether the dataset has an `image` column. If it doesn't, then this is a bug. Otherwise, please paste the line with the `.map` call.\r\n\r\n\r\n", "This is the piece of code I am running:\r\n```\r\ndata_transforms = utils.get_data_augmentation(args)\r\nimage_dataset = utils.load_image_dataset(args.dataset)\r\n\r\ndef resize(examples):\r\n examples[\"pixel_values\"] = [image.convert(\"RGB\").resize((300, 300)) for image in examples[\"image\"]]\r\n return examples\r\n\r\ndef preprocess_train(example_batch):\r\n print(f\"Example batch: \\n{example_batch}\")\r\n example_batch[\"pixel_values\"] = [\r\n data_transforms[\"train\"](image.convert(\"RGB\")) for image in example_batch[\"pixel_values\"]\r\n ]\r\n return example_batch\r\n\r\ndef preprocess_val(example_batch):\r\n example_batch[\"pixel_values\"] = [\r\n data_transforms[\"val\"](image.convert(\"RGB\")) for image in example_batch[\"pixel_values\"]\r\n ]\r\n return example_batch\r\n\r\nimage_dataset = image_dataset.map(resize, remove_columns=[\"image\"], batched=True)\r\n\r\nimage_dataset[\"train\"].set_transform(preprocess_train)\r\nimage_dataset[\"validation\"].set_transform(preprocess_val)\r\n```\r\n\r\nWhen I print ds.column_names I get the following\r\n`{'train': ['image', 'label'], 'validation': ['image', 'label'], 'test': ['image', 'label']}`\r\n\r\nThe `print(f\"Example batch: \\n{example_batch}\")` in the `preprocess_train` function outputs only labels without images:\r\n```\r\nExample batch: \r\n{'label': [0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3]}\r\n```\r\n\r\nThe weird part of it all is that a sample code runs in a jupyter lab notebook without any bugs, but when I run my scripts from the terminal I get the bug. The same code.", "The `remove_columns=[\"image\"]` argument in the `.map` call removes the `image` column from the output, so drop this argument to preserve it.", "The problem is not with the removal of the image key. The bug is why only the labels are sent to be process, instead of all the featues or dictionary keys.\r\n\r\nP.S. I just dropped the removal argument as you've suggested, but that didn't solve the problem, because only the labels are being sent to be processed", "All the `image_dataset.column_names` after the `map` call should also be present in `preprocess_train `/`preprocess_val` unless (input) `columns` in `set_transform` are specified.\r\n\r\nIf that's not the case, we need a full reproducer (not snippets) with the environment info.", "I have resolved the error after including a collate function as indicated in the Quick Start session of the Datasets docs.:\r\n\r\nHere is what I did:\r\n```\r\ndata_transforms = utils.get_data_augmentation(args)\r\nimage_dataset = utils.load_image_dataset(args.dataset)\r\n\r\ndef preprocess_train(example_batch):\r\n example_batch[\"pixel_values\"] = [\r\n data_transforms[\"train\"](image.convert(\"RGB\")) for image in example_batch[\"image\"]\r\n ]\r\n return example_batch\r\n\r\ndef preprocess_val(example_batch):\r\n example_batch[\"pixel_values\"] = [\r\n data_transforms[\"val\"](image.convert(\"RGB\")) for image in example_batch[\"image\"]\r\n ]\r\n return example_batch\r\n\r\ndef collate_fn(examples):\r\n images = []\r\n labels = []\r\n for example in examples:\r\n images.append((example[\"pixel_values\"]))\r\n labels.append(example[\"label\"])\r\n\r\n pixel_values = torch.stack(images)\r\n labels = torch.tensor(labels)\r\n return {\"pixel_values\": pixel_values, \"label\": labels}\r\n\r\ntrain_dataset = image_dataset[\"train\"].with_transform(preprocess_train)\r\nval_dataset = image_dataset[\"validation\"].with_transform(preprocess_val)\r\n\r\nimage_datasets = {\r\n \"train\": train_dataset,\r\n \"val\": val_dataset\r\n}\r\n\r\nsamplers = {\r\n \"train\": data.RandomSampler(train_dataset),\r\n \"val\": data.SequentialSampler(val_dataset),\r\n}\r\n\r\ndataloaders = {\r\n x: data.DataLoader(\r\n image_datasets[x],\r\n collate_fn=collate_fn,\r\n batch_size=batch_size,\r\n sampler=samplers[x],\r\n num_workers=args.num_workers,\r\n worker_init_fn=utils.set_seed_for_worker,\r\n generator=g,\r\n pin_memory=True,\r\n )\r\n for x in [\"train\", \"val\"]\r\n}\r\n\r\ntrain_loader, val_loader = dataloaders[\"train\"], dataloaders[\"val\"]\r\n```\r\nEverything runs fine without any bug now. " ]

[]

[]

[ "Hi ! I opened https://github.com/huggingface/datasets/pull/6067 to add the missing `_lock`\r\n\r\nWe'll do a patch release soon, but feel free to install `datasets` from source in the meantime", "I have tested the latest main, it does not work.\r\n\r\nI add more logs to reproduce this issue, it looks like a multi threading bug:\r\n\r\n```python\r\n@contextmanager\r\ndef ensure_lock(tqdm_class, lock_name=\"\"):\r\n \"\"\"get (create if necessary) and then restore `tqdm_class`'s lock\"\"\"\r\n import os\r\n import threading\r\n print(os.getpid(), threading.get_ident(), \"ensure_lock\", tqdm_class, lock_name)\r\n old_lock = getattr(tqdm_class, '_lock', None) # don't create a new lock\r\n lock = old_lock or tqdm_class.get_lock() # maybe create a new lock\r\n lock = getattr(lock, lock_name, lock) # maybe subtype\r\n tqdm_class.set_lock(lock)\r\n print(os.getpid(), threading.get_ident(), \"set_lock\")\r\n yield lock\r\n if old_lock is None:\r\n print(os.getpid(), threading.get_ident(), \"del tqdm_class\")\r\n del tqdm_class._lock\r\n else:\r\n tqdm_class.set_lock(old_lock)\r\n```\r\noutput\r\n```\r\n64943 8424758784 ensure_lock <datasets.utils.logging._tqdm_cls object at 0x2aa7fb250> \r\n64943 8424758784 set_lock\r\n64943 8424758784 del tqdm_class\r\n64943 8424758784 ensure_lock <datasets.utils.logging._tqdm_cls object at 0x2aa7fb250> \r\n64943 8424758784 set_lock\r\n64943 8424758784 del tqdm_class\r\n64943 11638370304 ensure_lock <datasets.utils.logging._tqdm_cls object at 0x2aa7fb250> \r\n64943 11638370304 set_lock\r\n64943 11568967680 ensure_lock <datasets.utils.logging._tqdm_cls object at 0x2aa7fb250> \r\n64943 11568967680 set_lock\r\n64943 11638370304 del tqdm_class\r\n64943 11638370304 ensure_lock <datasets.utils.logging._tqdm_cls object at 0x2aa7fb250> \r\n64943 11638370304 set_lock\r\n64943 11638370304 del tqdm_class\r\n64943 11568967680 del tqdm_class\r\n```\r\n\r\nThread `11638370304` del the _lock from tqdm_class first, then thread `11568967680` del _lock failed.", "Maybe it is a bug of tqdm? I think simply use `try ... except AttributeError ...` wraps `del tqdm_class._lock` should work.", "Yes it looks like a bug on their end indeed, do you want to open a PR on tqdm ?\r\n\r\nLet me see if I can find a workaround in the meantime", "I opened https://github.com/huggingface/datasets/pull/6068 if you want to try it out", "> I opened #6068 if you want to try it out\r\n\r\nThis fix works! Thanks.", "Awesome ! closing this then :)\r\nWe'll do a patch release today or tomorrow" ]

[]

[]

[ "Sorry for asking a duplicate question about `num_proc`, I searched the forum and find the solution.\r\n\r\nBut I still can't make the trick with `torch.distributed.barrier()` to only map at the main process work. The [post on forum]( https://discuss.huggingface.co/t/slow-processing-with-map-when-using-deepspeed-or-fairscale/7229/7) didn't help.", "If it does the `map` twice then it means the hash of your map function is not some same between your two processes.\r\n\r\nCan you make sure your map functions have the same hash in different processes ?\r\n\r\n```python\r\nfrom datasets.fingerprint import Hasher\r\n\r\nprint(Hasher.hash(lambda x: cut_reorder_keys(x, num_stations_list=args.num_stations_list, is_pad=True, is_train=True)))\r\nprint(Hasher.hash(lambda x: random_shift(x, shift_range=(-160, 0), feature_scale=16)))\r\n```\r\n\r\nYou can also set the fingerprint used to reload the resulting dataset by passing `new_finegrprint=` in `map`, see https://huggingface.co/docs/datasets/v2.13.1/en/about_cache#the-cache. This will force the different processes to use the same fingerprint used to locate the resulting dataset in the cache.", "Thanks for help! I find the fingerprint between processes don't have same hash:\r\n```\r\nRank 0: Gpu 0 cut_reorder_keys fingerprint c7f47f40e9a67657\r\nRank 0: Gpu 0 random_shift fingerprint 240a0ce79831e7d4\r\n\r\nRank 1: Gpu 1 cut_reorder_keys fingerprint 20edd3d9cf284001\r\nRank 1: Gpu 1 random_shift fingerprint 819f7c1c18e7733f\r\n```\r\nBut my functions only process the example one by one and don't need rank or other arguments. After all it can work in the test for dataset and dataloader.\r\nI'll try to set `new_fingerprint` to see if it works and figure out the reason of different hash.", "I finally figure it out. The fingerprint of the function will change if other key-value pairs change in `args` even the `args.num_stations_list` is not changed.\r\n\r\n```python\r\nlambda x: cut_reorder_keys(x, num_stations_list=args.num_stations_list, is_pad=True, is_train=True)\r\n```\r\n\r\nMy `args` contains the key `rank` which refers the rank of its GPU, so the fingerprints change among the GPUs.\r\nI use `partial` in `functools` to generate a partial function that fixs the argument `num_stations_list=args.num_stations_list`, and the fingerprint of this partial function keeps among the GPUs. Finally I can reuse the mapped cache." ]

[ { "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" } ]

[]

[]

[]

[]

[ "This can also mean one of the files was not downloaded correctly.\r\n\r\nWe log an erroneous file's name before raising the reader's error, so this is how you can find the problematic file. Then, you should delete it and call `load_dataset` again.\r\n\r\n(I checked all the uploaded files, and they seem to be valid Parquet files, so I don't think this is a bug on their side)\r\n" ]

[]

[]

[ "Hi!\r\n\r\nIt's hard to explain this behavior without more information. Can you profile the slower version with the following code\r\n```python\r\nimport cProfile, pstats\r\nfrom datasets import load_dataset\r\n\r\nwith cProfile.Profile() as profiler:\r\n ds = load_dataset(...)\r\n\r\nstats = pstats.Stats(profiler).sort_stats(\"cumtime\")\r\nstats.print_stats()\r\n```\r\nand share the output?" ]

[]

[]

[]

[ { "id": 1935892865, "node_id": "MDU6TGFiZWwxOTM1ODkyODY1", "url": "https://api.github.com/repos/huggingface/datasets/labels/duplicate", "name": "duplicate", "color": "cfd3d7", "default": true, "description": "This issue or pull request already exists" } ]

[]

[ "A duplicate of https://github.com/huggingface/datasets/issues/5929" ]

[ { "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" } ]

[]

[ "This would break a lot of existing code, so we can't really do this." ]

[]

[]

[ "Hi! `_select_contiguous` fetches a (zero-copy) slice of the dataset's Arrow table to build a shard, so I don't think this part is the problem. To me, the issue seems to be the step where we embed external image files' bytes (a lot of file reads). You can use `.map` with multiprocessing to perform this step before `push_to_hub` in a faster manner and cache it to disk:\r\n```python\r\nfrom datasets.table import embed_table_storage\r\n# load_dataset(...)\r\nformat = dataset.format\r\ndataset = dataset.with_format(\"arrow\")\r\ndataset = dataset.map(embed_table_storage, batched=True)\r\ndataset = dataset.with_format(**format)\r\n# push_to_hub(...)\r\n```\r\n\r\n(In Datasets 3.0, these external bytes will be written to an Arrow file when generating a dataset to avoid this \"embed\" step)", "Hi, thanks, this solution saves some time.\r\nBut can't we avoid embedding all external image files bytes with each push, skipping the images that have already been pushed into the repo?\r\n\r\nEdit: Ok I missed the part of cache it manually on the disk the first time, this solves the problem. Thank you" ]

[]

[]

[ "The `audiofolder` loader is not available in version `2.3.2`, hence the error. Please run the `pip install -U datasets` command to update the `datasets` installation to make `load_dataset(\"audiofolder\", ...)` work." ]

[ { "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": 3761482852, "node_id": "LA_kwDODunzps7gM6xk", "url": "https://api.github.com/repos/huggingface/datasets/labels/good%20second%20issue", "name": "good second issue", "color": "BDE59C", "default": false, "description": "Issues a bit more difficult than \"Good First\" issues" } ]

[ { "login": "zutarich", "id": 95092167, "node_id": "U_kgDOBar9xw", "avatar_url": "https://avatars.githubusercontent.com/u/95092167?v=4", "gravatar_id": "", "url": "https://api.github.com/users/zutarich", "html_url": "https://github.com/zutarich", "followers_url": "https://api.github.com/users/zutarich/followers", "following_url": "https://api.github.com/users/zutarich/following{/other_user}", "gists_url": "https://api.github.com/users/zutarich/gists{/gist_id}", "starred_url": "https://api.github.com/users/zutarich/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/zutarich/subscriptions", "organizations_url": "https://api.github.com/users/zutarich/orgs", "repos_url": "https://api.github.com/users/zutarich/repos", "events_url": "https://api.github.com/users/zutarich/events{/privacy}", "received_events_url": "https://api.github.com/users/zutarich/received_events", "type": "User", "site_admin": false } ]

[ "hii @lhoestq can you assign this issue to me?\r\n", "You can reply \"#self-assign\" to this issue to automatically get assigned to it :)\r\nLet me know if you have any questions or if I can help", "#2289 ", "Actually i am quite new to figure it out how everything goes and done \r\n\r\n> You can reply \"#self-assign\" to this issue to automatically get assigned to it :)\r\n> Let me know if you have any questions or if I can help\r\n\r\nwhen i wrote #self-assign it automatically got converted to some number is it correct or i have done it some wrong way, I am quite new to open source thus wanna try to learn and explore it", "#2289 #self-assign ", "Ah yea github tries to replace the #self-assign with an issue link. I guess you can try to copy-paste instead to see if it works\r\n\r\nAnyway let me assign you manually", "thanks a lot @lhoestq ! though i have a very lil idea of the issue, i am new. as i said before, but gonna try my best shot to do it.\r\ncan you please suggest some tips or anything from your side, how basically we approach it will be really helpfull.\r\nWill try my best!", "The HfFileSystem from the `huggingface_hub` package can already read the HTTP_PROXY and HTTPS_PROXY environment variables. So the remaining thing missing is the `user_agent` that the user may include in a `DownloadConfig` object. The user agent can be used for regular http calls but also calls to the HfFileSystem.\r\n\r\n- for http, the `user_agent` isn't passed from `DownloadConfig` to `get_datasets_user_agent` in `_prepare_single_hop_path_and_storage_options` in `streaming_download_manager.py` so we need to include it\r\n- for HfFileSystem I think it requires a PR in https://github.com/huggingface/huggingface_hub to include it in the `HfFileSystem.__init__`" ]

[]

[]

[ "Hello @exs-avianello, I have reproduced the bug successfully and have understood the problem. But I am confused regarding this part of the statement, \"`pandas.DataFrame.to_csv` is always called with a buf-like `path_or_buf`\".\r\n\r\nCan you please elaborate on it?\r\n\r\nThanks!", "Hi @aryanxk02 ! Sure, what I actually meant is that when passing a path-like `path_or_buf` here\r\n\r\nhttps://github.com/huggingface/datasets/blob/14f6edd9222e577dccb962ed5338b79b73502fa5/src/datasets/arrow_dataset.py#L4708-L4714 \r\n\r\nit gets converted to a file object behind the scenes here\r\n\r\nhttps://github.com/huggingface/datasets/blob/14f6edd9222e577dccb962ed5338b79b73502fa5/src/datasets/io/csv.py#L92-L94\r\n\r\nand the eventual pandas `.to_csv()` calls that write to it always get `path_or_buf=None`, making pandas ignore the `compression` kwarg in the `to_csv_kwargs`\r\n\r\nhttps://github.com/huggingface/datasets/blob/14f6edd9222e577dccb962ed5338b79b73502fa5/src/datasets/io/csv.py#L107-L109", "@exs-avianello When `path_or_buf` is set to None, the `to_csv()` method will return the CSV data as a string instead of saving it to a file. Hence the compression doesn't take place. I think setting `path_or_buf=self.path_or_buf` should work. What you say?" ]

[]

[]

[]

[]

[]

[ "Instead of writing the loading script, you can use the built-in loader to [load JSON files](https://huggingface.co/docs/datasets/loading#json):\r\n```python\r\nfrom datasets import load_dataset\r\nds = load_dataset(\"json\", data_files={\"train\": os.path.join(data_dir[\"train\"]), \"dev\": os.path.join(data_dir[\"dev\"])})\r\n```" ]

[]

[]

[ "These docs are outdated (version 1.2.1 is over two years old). Please refer to [this](https://huggingface.co/docs/datasets/dataset_script) version instead.\r\n\r\nInitially, we hosted datasets in this repo, but now you can find them [on the HF Hub](https://huggingface.co/datasets) (e.g. the [`ag_news`](https://huggingface.co/datasets/ag_news/blob/main/ag_news.py) script)", "Sorry I thought I'd selected the latest version." ]

[]

[]

[ "Even if a dataset is cached, we still make requests to check whether the cache is up-to-date. [This](https://huggingface.co/docs/datasets/v2.13.1/en/loading#offline) section in the docs explains how to avoid them and directly load the cached version.", "Thanks - that works! However it doesn't resolve the original issue (but I am not sure if it is a WSL problem)", "We use `requests` to make HTTP requests (and `aiohttp` in the streaming mode), so I don't think we can provide much help regarding the socket issue (it probably has something to do with WSL). " ]

[]

[]

[]

[]

[]

[ "`HfApi` comes from the `huggingface_hub` package. You can use [this](https://huggingface.co/docs/huggingface_hub/v0.16.3/en/package_reference/utilities#huggingface_hub.configure_http_backend) utility to change the `huggingface_hub`'s `Session` proxies (see the example).\r\n\r\nWe plan to implement https://github.com/huggingface/datasets/issues/5080 and make this behavior more consistent eventually.", "> this\r\n\r\nThanks. I will try `huggingface_hub.configure_http_backend` to change session's config.", "@mariosasko are you saying if I do the following:\r\n\r\n```\r\ndef backend_factory() -> requests.Session:\r\n session = requests.Session()\r\n session.proxies = {\r\n \"https\": \"127.0.0.1:8887\",\r\n \"http\": \"127.0.0.1:8887\",\r\n }\r\n session.verify = \"/etc/ssl/certs/ca-certificates.crt\"\r\n return session\r\n\r\n# Set it as the default session factory\r\nconfigure_http_backend(backend_factory=backend_factory)\r\n```\r\n\r\nwhich works nicely with transformer library:\r\n\r\n```\r\ndef download_gpt_2_model():\r\n tokenizer = GPT2Tokenizer.from_pretrained(\r\n \"gpt2\", force_download=True, resume_download=False\r\n )\r\n text = \"Replace me by any text you'd like.\"\r\n encoded_input = tokenizer(text, return_tensors=\"pt\")\r\n print(encoded_input)\r\n\r\n model = GPT2Model.from_pretrained(\r\n \"gpt2\", force_download=True, resume_download=False\r\n )\r\n output = model(**encoded_input)\r\n```\r\n\r\nshould work for datasets library as well ?\r\n\r\nIn my case if I just do:\r\n\r\n```\r\ndef download_sts12_sts_dataset():\r\n dataset = load_dataset(\r\n \"mteb/sts12-sts\",\r\n download_mode=\"force_redownload\",\r\n verification_mode=\"basic_checks\",\r\n revision=\"main\",\r\n )\r\n\r\n```\r\nI am getting:\r\n`ConnectionError: Couldn't reach https://huggingface.co/datasets/mteb/sts12-sts/resolve/main/dataset_infos.json (ConnectTimeout(MaxRetryError(\"HTTPSConnectionPool(host='huggingface.co', port=443): Max retries exceeded with url: /datasets/mteb/sts12-sts/resolve/main/dataset_infos.json (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7f429e87a3a0>, 'Connection to huggingface.co timed out. (connect timeout=100)'))\")))`\r\n\r\nwhich is typical when the proxy server is not defined. Looks like what is set in configure_http_backend(backend_factory=backend_factory) is ignore.\r\n\r\nIf I use env variable instead, it is working \r\n```\r\ndef download_sts12_sts_dataset():\r\n\r\n os.environ[\"https_proxy\"] = \"127.0.0.1:8887\"\r\n os.environ[\"http_proxy\"] = \"127.0.0.1:8887\"\r\n os.environ[\"REQUESTS_CA_BUNDLE\"] = \"/etc/ssl/certs/ca-certificates.crt\"\r\n\r\n dataset = load_dataset(\r\n \"mteb/sts12-sts\",\r\n download_mode=\"force_redownload\",\r\n verification_mode=\"basic_checks\",\r\n revision=\"main\",\r\n )\r\n```\r\n\r\nShould I add something ?\r\n\r\nI am using `huggingface_hub 0.15.1`, `datasets 2.13.0`, `transformers 4.30.2`", "`huggingface_hub.configure_http_backend` works for `transformers` because they only use the `huggingface_hub` lib for downloads. Our download logic is a bit more complex (e.g., we also support downloading non-Hub files), so we are not aligned with them yet. In the meantime, it's best to use the env vars.", "@mariosasko I fully understand that the logic for dataset is different. I see 2 issues with the current implementation of the env variables:\r\n\r\n- having the same https_proxy/http_prox/no_proxy env variables for all tools is not good in some case. For example I have 2 differents proxy server. In 2019 we had discussion with the Tensorflow teams and they recommended to do the following: TFDS_HTTP_PROXY, TFDS_HTTPS_PROXY ...\r\n- with recent version of requests, it is not possible to deactivate TLS interception (verify=false) by using env variable. This is useful to debug things and in some case TLS is not working and you need to ignore verifying the SSL certificate (probably not recommended) \r\n\r\nOne of the best way is to able to pass our requests.Session() directly\r\n```\r\nimport openai\r\nsession = requests.Session()\r\nsession.cert = CERT\r\nsession.verify = False\r\nopenai.requestssession = session\r\n```\r\n\r\nMy 2 cents in this discussion" ]

[]

[]

[ "Yes, this is intended." ]

[]

[]

[ "I've opened a PR with a fix. In the meantime, you can avoid the error by deleting `task_templates` with `dataset.info.task_templates = None` before the `interleave_datasets` call.\r\n` " ]

[]

[]

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

[]

[]

[ "This scenario currently requires explicitly passing the target features (to avoid the error): \r\n```python\r\nimport datasets\r\n\r\n...\r\n\r\nfeatures = dataset.features\r\nfeatures[\"output\"] = = [{\"test\": datasets.Value(\"int64\")}]\r\ntest2 = dataset.map(lambda row, idx: test_func(row, idx), with_indices=True, num_proc=32, features=features)\r\n```", "I just encountered the same error in the same situation (multiprocessing with variable length outputs).\r\n\r\nThe funny (or dangerous?) thing is, that this error only showed up when testing with a small test dataset (16 examples, ValueError with `num_proc` >1) but the same code works fine for the full dataset (~70k examples).\r\n\r\n@mariosasko Any idea on how to do that with a nested feature with lists of variable lengths containing dicts?\r\n\r\nEDIT: Was able to narrow it down: >200 Examples: no error, <150 Examples: Error. \r\nNow idea what to make of this but pretty obvious that this is a bug....", "This error also occurs while concatenating the datasets." ]

[ { "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" } ]

[ { "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 } ]

[]

[ { "id": 1935892865, "node_id": "MDU6TGFiZWwxOTM1ODkyODY1", "url": "https://api.github.com/repos/huggingface/datasets/labels/duplicate", "name": "duplicate", "color": "cfd3d7", "default": true, "description": "This issue or pull request already exists" } ]

[]

[ "Hi ! The huggingface/dataset-viewer code was not maintained anymore because we switched to a new dataset viewer that is deployed available for each dataset the Hugging Face website.\r\n\r\nWhat are you using this old repository for ?", "I think these parts are outdated:\r\n\r\n* https://github.com/huggingface/datasets-viewer/blob/8efad8eae313a891f713469983bf4c744786f26e/run.py#L126-L131\r\n* https://github.com/huggingface/datasets-viewer/blob/8efad8eae313a891f713469983bf4c744786f26e/run.py#L145-L150\r\n\r\nTo make the viewer work, the first one should be replaced with the following:\r\n```python\r\ndataset_module = datasets.load.dataset_module_factory(path)\r\nbuilder_cls = datasets.load.import_main_class(dataset_module.module_path)\r\nconfs = builder_cls.BUILDER_CONFIGS\r\n```\r\nAnd the second one:\r\n```python\r\ndataset_module = datasets.load.dataset_module_factory(path)\r\nbuilder_cls = datasets.load.import_main_class(dataset_module.module_path)\r\nif conf:\r\n builder_instance = builder_cls(name=conf, cache_dir=path if path_to_datasets is not None else None)\r\nelse:\r\n builder_instance = builder_cls(cache_dir=path if path_to_datasets is not None else None)\r\n```\r\n\r\nBut as @lhoestq suggested, it's better to use the `datasets-server` API nowadays to [fetch the rows](https://huggingface.co/docs/datasets-server/rows).", "> The dataset viewer on the Hugging Face website is incredibly useful\r\n\r\n@mariosasko i think @lilyorlilypad wants to run the new dataset-viewer, not the old one", "> wants to run the new dataset-viewer, not the old one\r\n\r\nThanks for the clarification for me. I do want to run the new dataset-viewer. ", "It should be possible to run it locally using the HF datasets-server API (docs [here](https://huggingface.co/docs/datasets-server)) but the front end part is not open source (yet ?)\r\n\r\nThe back-end is open source though if you're interested: https://github.com/huggingface/datasets-server\r\nIt automatically converts datasets on HF to Parquet, which is the format we use to power the viewer.", "the new frontend would probably be hard to open source, as is, as it's quite intertwined with the Hub's code.\r\n\r\nHowever, at some point it would be amazing to have a community-driven open source implementation of a frontend to datasets-server! ", "For the frontend viewer, see https://github.com/huggingface/datasets/issues/6139.\r\n\r\nAlso mentioned in https://github.com/huggingface/datasets-server/issues/213 and https://github.com/huggingface/datasets-server/issues/441\r\n\r\nClosing as a duplicate of https://github.com/huggingface/datasets/issues/6139" ]

[ { "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": 3761482852, "node_id": "LA_kwDODunzps7gM6xk", "url": "https://api.github.com/repos/huggingface/datasets/labels/good%20second%20issue", "name": "good second issue", "color": "BDE59C", "default": false, "description": "Issues a bit more difficult than \"Good First\" issues" } ]

[]

[ "You can use the `remove_columns` parameter in `map` to avoid duplicating the columns (and save disk space) and then concatenate the original dataset with the map result:\r\n```python\r\nfrom datasets import concatenate_datasets\r\n# dummy example\r\nds_new = ds.map(lambda x: {\"new_col\": x[\"col\"] + 2}, remove_columns=ds.column_names)\r\nds_combined = concatenate_datasets([ds, ds_new], axis=1)\r\n```\r\n\r\nDoing this automatically is hard to implement efficiently unless we know ahead of time which existing columns will be modified by a `map` transform. We have this info when `input_columns` are specified, so I think this is the only case we can optimize." ]

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

[]

[ "You can use `with_transform` to get a new dataset object.\r\n\r\nSupport for lazy `map` has already been discussed [here](https://github.com/huggingface/datasets/issues/3385) a little bit. Personally, I'm not a fan, as this would make `map` even more complex. ", "> You can use `with_transform` to get a new dataset object.\r\n> \r\n> Support for lazy `map` has already been discussed [here](https://github.com/huggingface/datasets/issues/3385) a little bit. Personally, I'm not a fan, as this would make `map` even more complex.\r\n\r\nI read about IterableDataset, and it seems to have lazy mapping. But I can't figure out how to convert an IterableDataset into a normal one when needed.\r\n\r\n`with_transform` still does not chain AFAIU.", "> I read about IterableDataset, and it seems to have lazy mapping. But I can't figure out how to convert an IterableDataset into a normal one when needed.\r\n\r\nYou must cache an `IterableDataset` to disk to load it as a `Dataset`. One way to do this is with `Dataset.from_generator`:\r\n```python\r\nfrom functools import partial\r\nfrom datasets import Dataset\r\n\r\ndef gen_from_iterable_dataset(iterable_ds)\r\n yield from iterable_ds\r\n\r\nds = Dataset.from_generator(partial(gen_from_iterable_dataset, iterable_ds), features=iterable_ds.features})\r\n```\r\n\r\n> with_transform still does not chain AFAIU.\r\n\r\nYes, not supported yet - the solution is to combine the transforms into a single one.", "I wonder if it would be beneficial to have a dedicated method to do that ? Maybe a `.save_to_disk()` so that the user can reload the resulting dataset later ?", "> ```python\r\n> from functools import partial\r\n> from datasets import Dataset\r\n> \r\n> def gen_from_iterable_dataset(iterable_ds)\r\n> yield from iterable_ds\r\n> \r\n> ds = Dataset.from_generator(partial(gen_from_iterable_dataset, iterable_ds), features=iterable_ds.features})\r\n> ```\r\n\r\n@mariosasko With these complex mapping functions, what hash will be used to cache this dataset?\r\n", "The params passed to `Dataset.from_generator` will be used to compute the hash (`partial` encapsulates the `iterable_ds` value, so changing it will also change the hash)", "Hi, I think this feature would be very useful. I want to concatenate large datasets with heterogeneous columns. I dislike `map` since I don't want multiple copy of that datasets locally. I tried to use \"set_transform\" on each dataset to convert it to a standard features format, but `datasets.concatenate_datasets` ignores the updated format of the datasets.  A work around is to use `torch.utils.data.ConcatDataset`. Is there a neat way to do it using HF datasets?" ]

[]

[]

[ "Hi! You can do `ds.get_index(\"embeddings\").faiss_index.metric_type` to get the metric type and then match the result with the FAISS metric [enum](https://github.com/facebookresearch/faiss/blob/43d86e30736ede853c384b24667fc3ab897d6ba9/faiss/MetricType.h#L22-L36) (should be L2).", "Ah! Thank you for pointing this out. FYI: the enum indicates it's using the inner product. Using `torch.inner` or `torch.dot` still produces a discrepancy compared to the built-in score. I think this is because of the compression/quantization that occurs with the FAISS index." ]

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

[]

[ "I want to take a shot at this if possible ", "Yes, feel free to work on this.\r\n\r\nYou can check the PyArrow Table `__repr__` and Polars DataFrame `__repr__`/`_repr_html_` implementations for some pointers/ideas.", "@mariosasko are there any other similar issues that I could work on? I see this has been already solved. " ]

[]

[]

[ "By default, we write data to disk (so it can be memory-mapped) every 1000 rows/samples. You can control this with the `writer_batch_size` parameter. Also, when working with fixed-size arrays, the `ArrayXD` feature types yield better performance (e.g., in your case, `features=datasets.Features({\"i\": datasets.Array3D(shape=(512,512,3), dtype=\"float32\")})` should be faster).\r\n\r\nOur support for multi-dim arrays could be better, and we plan to improve it as part of https://github.com/huggingface/datasets/issues/5272.", "> By default, we write data to disk (so it can be memory-mapped) every 1000 rows/samples. You can control this with the `writer_batch_size` parameter. Also, when working with fixed-size arrays, the `ArrayXD` feature types yield better performance (e.g., in your case, `features=datasets.Features({\"i\": datasets.Array3D(shape=(512,512,3), dtype=\"float32\")})` should be faster).\r\n> \r\n> Our support for multi-dim arrays could be better, and we plan to improve it as part of #5272.\r\n\r\nThanks for the explanation! The Image array was just for demonstration, I use PIL Images in practice. Does that make a difference? What's the best approach for a dataset with PIL Images as rows?", "It's best to use the `datasets.Image()` feature type for PIL images (to save space) :)" ]

[ { "id": 5705560427, "node_id": "LA_kwDODunzps8AAAABVBPxaw", "url": "https://api.github.com/repos/huggingface/datasets/labels/arrow", "name": "arrow", "color": "c2e0c6", "default": false, "description": "Related to Apache Arrow" } ]

[]

[ "This error means that one of the int32 (`Value(\"int32\")`) columns in the dataset has a value that is out of the valid (int32) range.\r\n\r\nI'll open a PR to print the name of a problematic column to make debugging such errors easier.", "I am afraid int32 is not the reason for this error.\r\n\r\nI have submitted a commit to use int64 for all ints in the dataset:\r\nhttps://huggingface.co/datasets/liwu/MNBVC/commit/857ac00d9eab96a6708ad6a82bd9001686042a9e\r\n\r\nand I have updated my env to the latest datasets release:\r\nCopy-and-paste the text below in your GitHub issue.\r\n\r\n- `datasets` version: 2.13.1\r\n- Platform: macOS-13.2.1-arm64-arm-64bit\r\n- Python version: 3.11.2\r\n- Huggingface_hub version: 0.13.4\r\n- PyArrow version: 11.0.0\r\n- Pandas version: 1.5.3\r\n\r\nBut the error still exist\r\n\r\n```\r\nDownloading and preparing dataset mnbvc/news_peoples_daily to /Users/silver/.cache/huggingface/datasets/liwu___mnbvc/news_peoples_daily/0.0.1/ee380f6309fe9b8b0d1fb14d77118f132444f22c8c4b28bf5c1645312688e051...\r\nDownloading data files: 100%|███████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 12/12 [00:00<00:00, 9070.40it/s]\r\nExtracting data files: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 12/12 [00:00<00:00, 2697.16it/s]\r\n---------------------------------------------------------------------------\r\nOverflowError Traceback (most recent call last)\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:1647, in GeneratorBasedBuilder._prepare_split_single(self, gen_kwargs, fpath, file_format, max_shard_size, split_info, check_duplicate_keys, job_id)\r\n 1646 example = self.info.features.encode_example(record) if self.info.features is not None else record\r\n-> 1647 writer.write(example, key)\r\n 1648 num_examples_progress_update += 1\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:490, in ArrowWriter.write(self, example, key, writer_batch_size)\r\n 488 self.hkey_record = []\r\n--> 490 self.write_examples_on_file()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:448, in ArrowWriter.write_examples_on_file(self)\r\n 444 batch_examples[col] = [\r\n 445 row[0][col].to_pylist()[0] if isinstance(row[0][col], (pa.Array, pa.ChunkedArray)) else row[0][col]\r\n 446 for row in self.current_examples\r\n 447 ]\r\n--> 448 self.write_batch(batch_examples=batch_examples)\r\n 449 self.current_examples = []\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:553, in ArrowWriter.write_batch(self, batch_examples, writer_batch_size)\r\n 552 typed_sequence = OptimizedTypedSequence(col_values, type=col_type, try_type=col_try_type, col=col)\r\n--> 553 arrays.append(pa.array(typed_sequence))\r\n 554 inferred_features[col] = typed_sequence.get_inferred_type()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:236, in pyarrow.lib.array()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:110, in pyarrow.lib._handle_arrow_array_protocol()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:189, in TypedSequence.__arrow_array__(self, type)\r\n 188 trying_cast_to_python_objects = True\r\n--> 189 out = pa.array(cast_to_python_objects(data, only_1d_for_numpy=True))\r\n 190 # use smaller integer precisions if possible\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:320, in pyarrow.lib.array()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:39, in pyarrow.lib._sequence_to_array()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/error.pxi:144, in pyarrow.lib.pyarrow_internal_check_status()\r\n\r\nOverflowError: Python int too large to convert to C long\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nOverflowError Traceback (most recent call last)\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:1656, in GeneratorBasedBuilder._prepare_split_single(self, gen_kwargs, fpath, file_format, max_shard_size, split_info, check_duplicate_keys, job_id)\r\n 1655 num_shards = shard_id + 1\r\n-> 1656 num_examples, num_bytes = writer.finalize()\r\n 1657 writer.close()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:584, in ArrowWriter.finalize(self, close_stream)\r\n 583 self.hkey_record = []\r\n--> 584 self.write_examples_on_file()\r\n 585 # If schema is known, infer features even if no examples were written\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:448, in ArrowWriter.write_examples_on_file(self)\r\n 444 batch_examples[col] = [\r\n 445 row[0][col].to_pylist()[0] if isinstance(row[0][col], (pa.Array, pa.ChunkedArray)) else row[0][col]\r\n 446 for row in self.current_examples\r\n 447 ]\r\n--> 448 self.write_batch(batch_examples=batch_examples)\r\n 449 self.current_examples = []\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:553, in ArrowWriter.write_batch(self, batch_examples, writer_batch_size)\r\n 552 typed_sequence = OptimizedTypedSequence(col_values, type=col_type, try_type=col_try_type, col=col)\r\n--> 553 arrays.append(pa.array(typed_sequence))\r\n 554 inferred_features[col] = typed_sequence.get_inferred_type()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:236, in pyarrow.lib.array()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:110, in pyarrow.lib._handle_arrow_array_protocol()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/arrow_writer.py:189, in TypedSequence.__arrow_array__(self, type)\r\n 188 trying_cast_to_python_objects = True\r\n--> 189 out = pa.array(cast_to_python_objects(data, only_1d_for_numpy=True))\r\n 190 # use smaller integer precisions if possible\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:320, in pyarrow.lib.array()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/array.pxi:39, in pyarrow.lib._sequence_to_array()\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/pyarrow/error.pxi:144, in pyarrow.lib.pyarrow_internal_check_status()\r\n\r\nOverflowError: Python int too large to convert to C long\r\n\r\nThe above exception was the direct cause of the following exception:\r\n\r\nDatasetGenerationError Traceback (most recent call last)\r\nCell In[2], line 1\r\n----> 1 dataset = load_dataset(\"liwu/MNBVC\", 'news_peoples_daily', split='train')\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/load.py:1809, in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, verification_mode, ignore_verifications, keep_in_memory, save_infos, revision, use_auth_token, task, streaming, num_proc, storage_options, **config_kwargs)\r\n 1806 try_from_hf_gcs = path not in _PACKAGED_DATASETS_MODULES\r\n 1808 # Download and prepare data\r\n-> 1809 builder_instance.download_and_prepare(\r\n 1810 download_config=download_config,\r\n 1811 download_mode=download_mode,\r\n 1812 verification_mode=verification_mode,\r\n 1813 try_from_hf_gcs=try_from_hf_gcs,\r\n 1814 num_proc=num_proc,\r\n 1815 storage_options=storage_options,\r\n 1816 )\r\n 1818 # Build dataset for splits\r\n 1819 keep_in_memory = (\r\n 1820 keep_in_memory if keep_in_memory is not None else is_small_dataset(builder_instance.info.dataset_size)\r\n 1821 )\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:909, in DatasetBuilder.download_and_prepare(self, output_dir, download_config, download_mode, verification_mode, ignore_verifications, try_from_hf_gcs, dl_manager, base_path, use_auth_token, file_format, max_shard_size, num_proc, storage_options, **download_and_prepare_kwargs)\r\n 907 if num_proc is not None:\r\n 908 prepare_split_kwargs[\"num_proc\"] = num_proc\r\n--> 909 self._download_and_prepare(\r\n 910 dl_manager=dl_manager,\r\n 911 verification_mode=verification_mode,\r\n 912 **prepare_split_kwargs,\r\n 913 **download_and_prepare_kwargs,\r\n 914 )\r\n 915 # Sync info\r\n 916 self.info.dataset_size = sum(split.num_bytes for split in self.info.splits.values())\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:1670, in GeneratorBasedBuilder._download_and_prepare(self, dl_manager, verification_mode, **prepare_splits_kwargs)\r\n 1669 def _download_and_prepare(self, dl_manager, verification_mode, **prepare_splits_kwargs):\r\n-> 1670 super()._download_and_prepare(\r\n 1671 dl_manager,\r\n 1672 verification_mode,\r\n 1673 check_duplicate_keys=verification_mode == VerificationMode.BASIC_CHECKS\r\n 1674 or verification_mode == VerificationMode.ALL_CHECKS,\r\n 1675 **prepare_splits_kwargs,\r\n 1676 )\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:1004, in DatasetBuilder._download_and_prepare(self, dl_manager, verification_mode, **prepare_split_kwargs)\r\n 1000 split_dict.add(split_generator.split_info)\r\n 1002 try:\r\n 1003 # Prepare split will record examples associated to the split\r\n-> 1004 self._prepare_split(split_generator, **prepare_split_kwargs)\r\n 1005 except OSError as e:\r\n 1006 raise OSError(\r\n 1007 \"Cannot find data file. \"\r\n 1008 + (self.manual_download_instructions or \"\")\r\n 1009 + \"\\nOriginal error:\\n\"\r\n 1010 + str(e)\r\n 1011 ) from None\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:1508, in GeneratorBasedBuilder._prepare_split(self, split_generator, check_duplicate_keys, file_format, num_proc, max_shard_size)\r\n 1506 job_id = 0\r\n 1507 with pbar:\r\n-> 1508 for job_id, done, content in self._prepare_split_single(\r\n 1509 gen_kwargs=gen_kwargs, job_id=job_id, **_prepare_split_args\r\n 1510 ):\r\n 1511 if done:\r\n 1512 result = content\r\n\r\nFile ~/git/venv/lib/python3.11/site-packages/datasets/builder.py:1665, in GeneratorBasedBuilder._prepare_split_single(self, gen_kwargs, fpath, file_format, max_shard_size, split_info, check_duplicate_keys, job_id)\r\n 1663 if isinstance(e, SchemaInferenceError) and e.__context__ is not None:\r\n 1664 e = e.__context__\r\n-> 1665 raise DatasetGenerationError(\"An error occurred while generating the dataset\") from e\r\n 1667 yield job_id, True, (total_num_examples, total_num_bytes, writer._features, num_shards, shard_lengths)\r\n\r\nDatasetGenerationError: An error occurred while generating the dataset\r\n```\r\n\r\nBesides, it works fine when I am using streamed dataset.", "`simhash` is the problematic column - it has values such as `18329103420363166823` that are out of the int64 range. You can fix this by setting the feature type to `Value(\"string\")` (it's advised to use this type for hash values in general)\r\n\r\n> Besides, it works fine when I am using streamed dataset.\r\n\r\nStreaming yields Python dictionaries from the script without converting them to the Arrow representation, as this conversion step is not that cheap performance-wise.", "i am using uint64 for simhash\r\n\r\nuint64 ranges up to about 3.69E19.\r\n\r\n18329103420363166823 is less than this value.\r\n\r\nmoreover, our simhash algorithm use 64 bits. it should fit in uint64.\r\n\r\n\r\n\r\n", "You are right. I overlooked the feature type.\r\n\r\nThis is a reproducer:\r\n```python\r\nimport pyarrow as pa\r\nfrom datasets.arrow_writer import TypedSequence\r\n\r\npa.array(TypedSequence([18329103420363166823], type=Value(\"uint64\")))\r\n```\r\n\r\n`pa.array([18329103420363166823])` also fails with the same error, so it seems PyArrow does not always infer the correct type as NumPy does (`uint64` in this case).\r\n\r\nI'll report this issue in the Arrow repo.\r\n\r\n`pa.array([18329103420363166823], pa.uint64)` works, so maybe we can implement a temporary fix (supporting complex input such as `[{\"image\": pil_image, \"num\": uint64_value}]` would be hard though).\r\n\r\nIn the meantime, you should be able to bypass this error by returning the `simhash` values as NumPy scalars in the script:\r\n```python\r\ndef _generate_examples(self, ...):\r\n ...\r\n yield {..., \"simhash\": np.uint64(simhash), ...}\r\n```", "Thank you for checking this issue in detail.\r\n\r\nHowever, it seems that using `np.uint64(simhash)` does not work. The same issue still exists.\r\n\r\nhttps://huggingface.co/datasets/liwu/MNBVC/commit/1e44f1e400b7e61052647d44c99cdae3bae9c830\r\n\r\nAnyway, we decide to use string type for these simhash values. Hope pyarrow can fix their bug soon.", "Arrow issue: https://github.com/apache/arrow/issues/36520", "May be something read your training data line by line.\r\nThen your training data just only one line. \r\nIt is so large.\r\nI guess.\r\n" ]

[]

[]

[ "issue due to corrupted download files. resolved after cleaning download cache. sorry for any inconvinence." ]

[]

[]

[]

[]

[ { "login": "albertvillanova", "id": 8515462, "node_id": "MDQ6VXNlcjg1MTU0NjI=", "avatar_url": "https://avatars.githubusercontent.com/u/8515462?v=4", "gravatar_id": "", "url": "https://api.github.com/users/albertvillanova", "html_url": "https://github.com/albertvillanova", "followers_url": "https://api.github.com/users/albertvillanova/followers", "following_url": "https://api.github.com/users/albertvillanova/following{/other_user}", "gists_url": "https://api.github.com/users/albertvillanova/gists{/gist_id}", "starred_url": "https://api.github.com/users/albertvillanova/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/albertvillanova/subscriptions", "organizations_url": "https://api.github.com/users/albertvillanova/orgs", "repos_url": "https://api.github.com/users/albertvillanova/repos", "events_url": "https://api.github.com/users/albertvillanova/events{/privacy}", "received_events_url": "https://api.github.com/users/albertvillanova/received_events", "type": "User", "site_admin": false } ]

[ "Thanks for reporting, @Praful932.\r\n\r\nLet's continue the conversation on the Hub: https://huggingface.co/datasets/xglue/discussions/5" ]

[]

[]

[ "Thanks for reporting, @wangyuxinwhy. " ]

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

[]

[ "I just noticed the [docs](https://github.com/huggingface/datasets/blob/main/src/datasets/arrow_dataset.py#L2881C11-L2881C200) say:\r\n\r\n>If batched is `True` and `batch_size` is `n > 1`, then the function takes a batch of `n` examples as input and can return a batch with `n` examples, or with an arbitrary number of examples.\r\n\r\nso maybe this is a bug then.", "All the values in a batch must be of the same length. So one solution is dropping all the input columns:\r\n```python\r\ndata = data.map(lambda samples: tokenizer(samples[\"text\"], max_length=tokenizer.model_max_length, truncation=True, stride=4, return_overflowing_tokens=True), batched=True, remove_columns=data.column_names)\r\n```\r\n\r\nAnother is padding/transforming the input columns to the tokenizer output's length (447). " ]

[]

[ { "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 } ]

[ "We'll do a new release of `datasets` soon to make the fix available :)\r\n\r\nIn the meantime you can use `datasets` from source (main)", "Thank you very much @lhoestq ! 🚀 " ]

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

[]

[]

[]

[]

[ "in this case we need to decide what to do with the existing datasets with white space characters (there shouldn't be a lot of them I think)", "I imagine that we should stop supporting them, and help the user fix them?", "See a report where the datasets server fails: https://huggingface.co/datasets/poloclub/diffusiondb/discussions/2#6374ff55b93cbdf65675f564\r\n\r\nThe config name is `random_10k [2m]`!" ]

[]

[]

[ "Unfortunately, I can't reproduce the error. What does the following code return for you?\r\n```python\r\nimport requests\r\nfrom huggingface_hub import hf_hub_url\r\nr = requests.get(hf_hub_url(\"codeparrot/codeparrot-clean-train\", \"dataset_infos.json\", repo_type=\"dataset\"))\r\n```\r\n\r\nAlso, can you provide more info about your network (region, proxies, etc.)?" ]

[]

[]

[ "Can you explain your use case for `max_shard_size`? \r\n\r\nOn some systems, there is a limit to the size of a memory-mapped file, so we could consider exposing this parameter in `load_dataset`.", "In my use case, users may choose a proper size to balance the cost and benefit of using large shard size. (On azure blob or hdfs which may automatically download the shard from background)", "But `load_dataset` doesn't support caching (and reading) Arrow datasets from remote storage. \r\n\r\n`load_datset_builder` + `download_and_prepare` is not equal to `load_dataset`. The latter has one more step, `builder.as_dataset`, that memory-maps Arrow files, which only works for local files.", "Thanks. So if I want to use `IterableDataset` and control the size of single arrow file, how should I organize the data loader? Maybe `load_dataset_build` + `download_and_prepare` + `builder.as_dataset` + `dataset.to_iterable_dataset`?", "Yes, this should work.\r\n\r\nI think we can expose `max_shard_size` in `load_dataset`, so feel free to open a PR." ]

[ { "id": 1935892865, "node_id": "MDU6TGFiZWwxOTM1ODkyODY1", "url": "https://api.github.com/repos/huggingface/datasets/labels/duplicate", "name": "duplicate", "color": "cfd3d7", "default": true, "description": "This issue or pull request already exists" } ]

[]

[ "This is a known issue: https://github.com/huggingface/datasets/issues/3847.\r\n\r\nFixing this requires significant work - rewriting the `tokenizers` lib to make them immutable.\r\n\r\nThe current solution is to pass `cache_file_name` to `map` to use that file for caching or calling a tokenizer before `map` (with the same set of parameters as the ones in the map transform)", "Closing since this is a duplicate" ]

[ { "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" } ]

[]

[ "For this to be possible, we would have to switch from the \"Streaming\" Arrow format to the \"Random Access\" (IPC/Feather) format, which allows reading arbitrary record batches (explained [here](https://arrow.apache.org/docs/python/ipc.html)). We could then use these batches to construct shards.\r\n\r\n@lhoestq @albertvillanova Do you think this use case is worth the switch? Also, we currently shard files, not inner row groups/chunks. Should we also support sharding row groups (e.g. if the number of input files is 1)?\r\n\r\nPS: I don't expect significant speed-up for local, uncompressed Arrow files.", "Alternatively we could support multiprocessing map for iterable datasets and let the user do the CPU intensive task there ?\r\n\r\nThis way it would work on arrow data but also on any iterable dataset", "> For this to be possible, we would have to switch from the \"Streaming\" Arrow format to the \"Random Access\" (IPC/Feather) format, which allows reading arbitrary record batches (explained [here](https://arrow.apache.org/docs/python/ipc.html)). We could then use these batches to construct shards.\r\n> \r\n> @lhoestq @albertvillanova Do you think this use case is worth the switch? Also, we currently shard files, not inner row groups/chunks. Should we also support sharding row groups (e.g. if the number of input files is 1)?\r\n> \r\n> PS: I don't expect significant speed-up for local, uncompressed Arrow files.\r\n\r\nCould you explain why you'd need to change the arrow format?\r\n\r\nWhen we use streaming datasets we simply determine the number of worker shards and then add some modulo logic at the appropriate place. Worst case scenario, you'd skip streaming entries according to the number of shards.\r\n\r\nFor PyTorch, I'd be happy to provide an implementation or a sketch thereof, if you point me toward what the testing requirements would be for such a PR.", "> Could you explain why you'd need to change the arrow format?\r\n\r\nThis way workers have random access to the location of the file where its dataset subset starts. Currently we're using the Arrow streaming format which doesn't include the metadata of the record batches offsets. This is needed here to efficiently split a dataset made of one single file.", "> > Could you explain why you'd need to change the arrow format?\r\n> \r\n> This way workers have random access to the location of the file where its dataset subset starts. Currently we're using the Arrow streaming format which doesn't include the metadata of the record batches offsets. This is needed here to efficiently split a dataset made of one single file.\r\n\r\nI guess I don't understand why you'd need to subset the dataset in the first place. \r\nIt seems sufficient to figure out how to offset or skip rows.\r\n\r\nFor instance, using pyArrow, you could use RecordBatchStreamReader to zero-copy iterate over records with read_next_batch and then only initiate the next step for records modulo worker shard.\r\nThat's one way to do it, where of course you'd need to account for gpu sharding as well.\r\n\r\n\r\nOtherwise, how did you implement worker/node/GPU sharding for iterable/streaming data where you do not have index information or prior splits (e.g. files)?", "> For instance, using pyArrow, you could use RecordBatchStreamReader to zero-copy iterate over records with read_next_batch and then only initiate the next step for records modulo worker shard.\r\n\r\nThat works indeed ! And what we meant is that you can make it even faster to instantiate. Indeed using RecordBatchStreamReader you need to get the list of all the record batches in each worker, whereas you could just get the list of record batches per worker if you use the record batches locations in the Arrow IPC file footer. This would be especially appreciated to have a fast instantiation in case you have tens of thousands of Arrow files for example.", "Any recent updates on this ? " ]

[]

[]

[ "This wasn’t working for me a bit earlier, but it looks to be back up now", "We had a minor issue updating the docs after the latest release. It should work now :)." ]

[]

[]

[ "I think it's more likely that this issue is related to PyTorch than Datasets, as PyTorch (on import) registers functions to execute when forking a process. Maybe this is the culprit: https://github.com/pytorch/pytorch/issues/99625", "From reading that ticket, it may be down in mkl? Is it worth hotfixing in the meantime, with the express intention of turning it off? I know that's a horribly crufty solution, but it's also deeply frustrating to be limited to 2 cores for operations as simple as filtration.", "This is too specific and unrelated to `datasets`, so this shouldn't be fixed here.", "@mariosasko @mmr-crexi I had the exact same problem on my kubernetes cluster. the datasets subprocess only user 1 and 17 core" ]

[]

[]

[ "Can you try again? Maybe there was a minor outage.", "Yes, it seems to be working now. In case it's helpful, the outage lasted several days. It was failing as late as yesterday morning. ", "we fixed something on the server side, glad it's fixed now" ]

[]

[]

[ "Duplicate of #", "Hi ! can you try to set the upper case environment variables `HTTP_PROXY` and `HTTPS_PROXY` ?\r\n\r\nWe use `aiohttp` for streaming and it uses case sensitive environment variables", "Hi, thanks for the quick reply.\r\n\r\nI set the uppercase env variables with\r\n\r\n`\r\nos.environ['HTTP_PROXY'] = \"http://example.com:xxxx\" \r\nos.environ['HTTPS_PROXY'] = \"http://example.com:xxxx\" \r\n`\r\n\r\nHowever, I still get the same error.\r\n\r\nOne thing that could be helpfull: When downloading a dataset without streaming i get the following message:\r\n_HF google storage unreachable. Downloading and preparing it from source_.\r\nThe download does however work as expected.\r\n", "Are you able to use `aiohttp` to get the file at `https://huggingface.co/datasets/facebook/voxpopuli/resolve/main/data/n_files.json` using your proxy ?", "It only works when passing trust_env=True when creating the ClientSession, as well as setting ssl=False.\r\n\r\nWorking Example:\r\n\r\n```\r\nimport os\r\n\r\nos.environ['HTTP_PROXY'] = \"xyz\"\r\nos.environ['HTTPS_PROXY'] = \"xyz\"\r\n\r\nimport asyncio\r\nimport aiohttp\r\n\r\nasync def download_pep(url):\r\n async with aiohttp.ClientSession(trust_env=True) as session:\r\n print(\"1\")\r\n async with session.get(url, ssl=False) as resp:\r\n print(\"2\")\r\n content = await resp.text()\r\n print(content)\r\n return content\r\n\r\nasyncio.run(download_pep(\"https://huggingface.co/datasets/facebook/voxpopuli/resolve/main/data/n_files.json\"))\r\n```\r\n\r\n\r\n\r\nSSL Verification has been a problem with other packages as well. Usually I circumvent the problem by setting\r\n```\r\nimport ssl\r\nssl._create_default_https_context = ssl._create_unverified_context\r\n```\r\n(probably not the best idea for security), although here aiohttp does not seem to use this default context.", "We do pass `trust_env` as well. Could you share the full stack trace you get when streaming using `datasets` ? That could help locate where we might have forgotten to pass `trust_env`", "Is there a way to disable ssl verification when streaming a dataset. I suspect this might be the isssue with my proxy.\r\n\r\n\r\nHere you go:\r\n\r\n```\r\nFileNotFoundError Traceback (most recent call last)\r\nCell In[8], line 3\r\n 1 from datasets import load_dataset\r\n----> 3 ds = load_dataset(\"facebook/voxpopuli\", name=\"de\", streaming=True)\r\n 5 sample = next(iter(ds))\r\n\r\nFile [~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/load.py:1790](https://vscode-remote+ssh-002dremote-002bml-002er-002dsoftware-002eat.vscode-resource.vscode-cdn.net/home/wrsbri/projects/audio_course/~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/load.py:1790), in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, verification_mode, ignore_verifications, keep_in_memory, save_infos, revision, use_auth_token, task, streaming, num_proc, storage_options, **config_kwargs)\r\n 1788 # Return iterable dataset in case of streaming\r\n 1789 if streaming:\r\n-> 1790 return builder_instance.as_streaming_dataset(split=split)\r\n 1792 # Some datasets are already processed on the HF google storage\r\n 1793 # Don't try downloading from Google storage for the packaged datasets as text, json, csv or pandas\r\n 1794 try_from_hf_gcs = path not in _PACKAGED_DATASETS_MODULES\r\n\r\nFile [~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/builder.py:1281](https://vscode-remote+ssh-002dremote-002bml-002er-002dsoftware-002eat.vscode-resource.vscode-cdn.net/home/wrsbri/projects/audio_course/~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/builder.py:1281), in DatasetBuilder.as_streaming_dataset(self, split, base_path)\r\n 1274 dl_manager = StreamingDownloadManager(\r\n 1275 base_path=base_path or self.base_path,\r\n 1276 download_config=DownloadConfig(use_auth_token=self.use_auth_token, storage_options=self.storage_options),\r\n 1277 dataset_name=self.name,\r\n 1278 data_dir=self.config.data_dir,\r\n 1279 )\r\n 1280 self._check_manual_download(dl_manager)\r\n-> 1281 splits_generators = {sg.name: sg for sg in self._split_generators(dl_manager)}\r\n 1282 # By default, return all splits\r\n 1283 if split is None:\r\n\r\nFile [~/.cache/huggingface/modules/datasets_modules/datasets/facebook--voxpopuli/b5ff837284f0778eefe0f642734e142d8c3f574eba8c9c8a4b13602297f73604/voxpopuli.py:120](https://vscode-remote+ssh-002dremote-002bml-002er-002dsoftware-002eat.vscode-resource.vscode-cdn.net/home/wrsbri/projects/audio_course/~/.cache/huggingface/modules/datasets_modules/datasets/facebook--voxpopuli/b5ff837284f0778eefe0f642734e142d8c3f574eba8c9c8a4b13602297f73604/voxpopuli.py:120), in Voxpopuli._split_generators(self, dl_manager)\r\n 118 def _split_generators(self, dl_manager):\r\n 119 n_shards_path = dl_manager.download_and_extract(_N_SHARDS_FILE)\r\n--> 120 with open(n_shards_path) as f:\r\n 121 n_shards = json.load(f)\r\n 123 if self.config.name == \"en_accented\":\r\n\r\nFile [~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/streaming.py:71](https://vscode-remote+ssh-002dremote-002bml-002er-002dsoftware-002eat.vscode-resource.vscode-cdn.net/home/wrsbri/projects/audio_course/~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/streaming.py:71), in extend_module_for_streaming..wrap_auth..wrapper(*args, **kwargs)\r\n 69 @wraps(function)\r\n 70 def wrapper(*args, **kwargs):\r\n---> 71 return function(*args, use_auth_token=use_auth_token, **kwargs)\r\n\r\nFile [~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/download/streaming_download_manager.py:517](https://vscode-remote+ssh-002dremote-002bml-002er-002dsoftware-002eat.vscode-resource.vscode-cdn.net/home/wrsbri/projects/audio_course/~/.conda/envs/audio_hf/lib/python3.10/site-packages/datasets/download/streaming_download_manager.py:517), in xopen(file, mode, use_auth_token, *args, **kwargs)\r\n 515 except FileNotFoundError:\r\n 516 if file.startswith(config.HF_ENDPOINT):\r\n--> 517 raise FileNotFoundError(\r\n 518 file + \"\\nIf the repo is private or gated, make sure to log in with `huggingface-cli login`.\"\r\n 519 ) from None\r\n 520 else:\r\n 521 raise\r\n\r\nFileNotFoundError: https://huggingface.co/datasets/facebook/voxpopuli/resolve/main/data/n_files.json\r\nIf the repo is private or gated, make sure to log in with `huggingface-cli login`.\r\n```", "> Is there a way to disable ssl verification when streaming a dataset.\r\n\r\nI don't think so.\r\n\r\nWe use `fsspec` HTTPFileSystem implementation that is based on `aiohttp`. If you register a subclass of HTTPFileSystem that has SSL disabled by default it could work, but I wouldn't recommended it because it can raise security issues.", "Okay thanks for your help! I guess I have to figure out how to improve the proxy environment / see if I can make it work with ssl connections." ]

[ { "id": 1935892861, "node_id": "MDU6TGFiZWwxOTM1ODkyODYx", "url": "https://api.github.com/repos/huggingface/datasets/labels/documentation", "name": "documentation", "color": "0075ca", "default": true, "description": "Improvements or additions to documentation" } ]

[ { "login": "benjaminbrown038", "id": 35114142, "node_id": "MDQ6VXNlcjM1MTE0MTQy", "avatar_url": "https://avatars.githubusercontent.com/u/35114142?v=4", "gravatar_id": "", "url": "https://api.github.com/users/benjaminbrown038", "html_url": "https://github.com/benjaminbrown038", "followers_url": "https://api.github.com/users/benjaminbrown038/followers", "following_url": "https://api.github.com/users/benjaminbrown038/following{/other_user}", "gists_url": "https://api.github.com/users/benjaminbrown038/gists{/gist_id}", "starred_url": "https://api.github.com/users/benjaminbrown038/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/benjaminbrown038/subscriptions", "organizations_url": "https://api.github.com/users/benjaminbrown038/orgs", "repos_url": "https://api.github.com/users/benjaminbrown038/repos", "events_url": "https://api.github.com/users/benjaminbrown038/events{/privacy}", "received_events_url": "https://api.github.com/users/benjaminbrown038/received_events", "type": "User", "site_admin": false } ]

[ "Loading a local dataset also works the same way when `data_files` are not specified, so I agree we should make this info easier to discover \r\n\r\ncc @stevhliu ", "Is this issue open? If so, I will self assign. ", "@benjaminbrown038 Yes, it is. Maybe @stevhliu can give some pointers on improving this doc page's discoverability.", "I think we can add a version of the [Main use-case](https://huggingface.co/docs/datasets/repository_structure#main-usecase) section to the [Share a dataset to the Hub](https://huggingface.co/docs/datasets/upload_dataset) tutorial. \r\n\r\nCurrently, it doesn't tell you *how* to structure the repository; it only tells you how to create it. So adding the \"main use-case\" will help bridge the gap and make it easier to find. We should also add a link to the [Structure your repository](https://huggingface.co/docs/datasets/repository_structure) guide for users who want to learn about the other options.", "#self-assign" ]

[]

[]

[ "Here's a minimal way to reproduce the bug, for the sake of convenience.\r\n````\r\nfrom datasets import Dataset, DatasetInfo, load_dataset\r\n\r\n\r\nepisodes_dict = {\"test\":[1,2,3],\"test2\": [1,2,4]}\r\n\r\nhugging_face_dataset = Dataset.from_dict(\r\n episodes_dict, info=DatasetInfo(description=\"test_str\")\r\n)\r\nprint(hugging_face_dataset.info)\r\n\r\nhugging_face_dataset.push_to_hub(\"balisujohn/minari_test\", private=True)\r\n\r\nredownloaded_dataset= load_dataset(\"balisujohn/minari_test\")[\"train\"]\r\n\r\n\r\nprint(redownloaded_dataset.info)\r\n````\r\n", "Thanks for reporting !\r\n\r\nFor now I would recommend uploading a separate JSON file for your metadata.\r\n\r\nAlternatively you can upload a second configuration of the dataset containing your metadata but this feature is not released yet (though you can already use it from [here](https://github.com/huggingface/datasets/pull/5331), it will be released soon)" ]

[]

[]

[ "cc @pcuenca as well. \r\n\r\nNot super urgent btw", "The issue commes from the dataset itself and is not related to the `datasets` lib\r\n\r\nsee https://huggingface.co/datasets/mozilla-foundation/common_voice_6_1/blob/2c475b3b88e0f2e5828f830a4b91618a25ff20b7/common_voice_6_1.py#L148-L152", "Let's remove these lines in the dataset no? cc @anton-l @Vaibhavs10 ", "Addressed in:\r\n\r\n* `mozilla-foundation/common_voice_1_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_1_0/discussions/4)\r\n* `mozilla-foundation/common_voice_2_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_2_0/discussions/3)\r\n* `mozilla-foundation/common_voice_3_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_3_0/discussions/3)\r\n* `mozilla-foundation/common_voice_4_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_4_0/discussions/3)\r\n* `mozilla-foundation/common_voice_5_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_5_0/discussions/3)\r\n* `mozilla-foundation/common_voice_5_1` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_5_1/discussions/3)\r\n* `mozilla-foundation/common_voice_6_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_6_0/discussions/3)\r\n* `mozilla-foundation/common_voice_6_1` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_6_1/discussions/3)\r\n* `mozilla-foundation/common_voice_7_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_7_0/discussions/3)\r\n* `mozilla-foundation/common_voice_8_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_8_0/discussions/7)\r\n* `mozilla-foundation/common_voice_9_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_9_0/discussions/8)\r\n* `mozilla-foundation/common_voice_10_0` [PR](https://huggingface.co/datasets/mozilla-foundation/common_voice_10_0/discussions/7)" ]

[]

[]

[ "This must be due to DatasetInfo.from_merge which drops them and is used in `concatenate_datasets`.\r\n\r\nAnd you're experiencing this issue because multiprocessing does concatenate the resulting datasets from each process.\r\n\r\nMaybe they should be kept if all the subdatasets share the same values for config_name and split", "That sounds like a clean workaround!" ]

[]

[ { "login": "mariosasko", "id": 47462742, "node_id": "MDQ6VXNlcjQ3NDYyNzQy", "avatar_url": "https://avatars.githubusercontent.com/u/47462742?v=4", "gravatar_id": "", "url": "https://api.github.com/users/mariosasko", "html_url": "https://github.com/mariosasko", "followers_url": "https://api.github.com/users/mariosasko/followers", "following_url": "https://api.github.com/users/mariosasko/following{/other_user}", "gists_url": "https://api.github.com/users/mariosasko/gists{/gist_id}", "starred_url": "https://api.github.com/users/mariosasko/starred{/owner}{/repo}", "subscriptions_url": "https://api.github.com/users/mariosasko/subscriptions", "organizations_url": "https://api.github.com/users/mariosasko/orgs", "repos_url": "https://api.github.com/users/mariosasko/repos", "events_url": "https://api.github.com/users/mariosasko/events{/privacy}", "received_events_url": "https://api.github.com/users/mariosasko/received_events", "type": "User", "site_admin": false } ]

[ "Thanks for reporting! \r\n\r\nSpecifying the target features explicitly should avoid this error:\r\n```python\r\ndataset = dataset.map(\r\n batch_process,\r\n batched=True,\r\n batch_size=1,\r\n num_proc=1,\r\n remove_columns=dataset.column_names,\r\n features=datasets.Features({\"texts\": datasets.Sequence(datasets.Value(\"string\"))})\r\n)\r\n```\r\n\r\nThis error stems from our type promotion not handling the nested case. But this promotion/casting allocates memory in most scenarios, which can be problematic for large datasets, so explicitly passing the features is the optimal solution.", "Hi @mariosasko thanks for the context, this is helpful to know. Would it be worth having some logic to generate this explicit feature specification automatically if a type annotation for a .map returns a dataclass that can be inferred?\r\n\r\nFeels like something that would be easy to implement and could save memory / deal with this case in a standardized way.", "> . Would it be worth having some logic to generate this explicit feature specification automatically if a type annotation for a .map returns a dataclass that can be inferred?\r\n\r\nInteresting proposal! Yes, we could consider doing this if the (return) type hint is `TypedDict`, and raise an error that type hints are incorrect if the cast using the inferred types fails.", "@mariosasko Put up an initial PR to implement this proposal. Let me know your thoughts on direction and what else should be in-scope here." ]

[]

[]

[ "i got error using method from_spark when using multi-node Spark cluster. seems could only use \"from_spark\" in local?", "@lhoestq ", "cc @maddiedawson it looks like there an issue with `_validate_cache_dir` ?\r\n\r\nIt looks like the function passed to mapPartitions has a reference to the Spark dataset builder, and therefore contains the SparkContext itself.\r\n\r\nI think it can be fixed by defining `create_cache_and_write_probe` outside the Spark dataset builder, and pass a `partial(create_cache_and_write_probe, cache_dir=self._cache_dir)` to `mapPartitions`", "Just saw this; thanks for flagging! Your proposed solution sounds good. I can prepare a PR", "@maddiedawson can you show me the demo ,so i can test in local .before your PR" ]

[]

[]

[]

[]

[]

[ "Does \"number of shards\" refer to the total number of data?\r\n\r\nmy config:\r\nnproc_per_node=2\r\nds=ds['train'] = load_dataset(streaming=True).take(50000)\r\n\r\nI'm test again: in prepare_data(), data have the same for each GPU\r\n", "The number of shards is `ds.n_shards`. It corresponds generally to the number of files the dataset is made of, to be able to distribute to several nodes.\r\n\r\n**You don't end up with the same data per GPU**. But all the samples are going through your preprocessing function you pass to map. They are just skipped afterwards to only keep 1 sample out of n(GPUs)", "For each GPU, although see the same data in prepare_data(), the actual training data will not be the same in the end. \r\nIs my understanding correct?\r\n\r\nWhere can I print the actual training data for each GPU?", "> For each GPU, although see the same data in prepare_data(), the actual training data will not be the same in the end.\r\nIs my understanding correct?\r\n\r\nYes exactly :)\r\n\r\n> Where can I print the actual training data for each GPU?\r\n\r\nYou should call print in the data_collator", "I print out n_shards, and under multiple GPUs, this value is always 1.\r\nIs this value correct?", "Yes it's correct, and it explains why you always have the same data passed to your map function (the data can't be split).\r\n\r\nBut after being passed to `map`, each GPU keeps one example out of n(GPUs) so that you don't end up with duplicate data across GPUs", "> > For each GPU, although see the same data in prepare_data(), the actual training data will not be the same in the end.\r\n> > Is my understanding correct?\r\n> \r\n> Yes exactly :)\r\n> \r\n> > Where can I print the actual training data for each GPU?\r\n> \r\n> You should call print in the data_collator\r\n\r\nOK, when printing the train data in the data collator, each GPU sees different data.\r\n\r\nThanks for your reply", "Do we have a solution for this one? Or it's required to get \"number of shards is a factor of number of GPUs: in that case the shards are evenly distributed per GPU\"", "For now it's required to have a number of shards that is a factor of the number of GPUs to not have all the workers process the same data (and then skip the right ones to not end up training on duplicate data).\r\n\r\nIt would be quite complex to implement a strategy that would utilize all the GPUs with an arbitrary number of shards even at the end of training" ]

[]

[]

[ "Sorry, I solve this by call `evaluate.load('glue_metric.py','sst-2')`\r\n" ]

[]

[]

[ "This is the actual error:\r\n```\r\nFailed to read file '/home/lakala/hjc/code/pycode/glm/temp.json' with error <class 'pyarrow.lib.ArrowInvalid'>: cannot mix list and non-list, non-null values\r\n```\r\nWhich means some samples are incorrectly formatted.\r\n\r\nPyArrow, a storage backend that we use under the hood, requires that all the list elements have the same level of nesting (same number of dimensions) or are `None`.\r\n```python\r\nimport pyarrow as pa\r\npa.array([[1, 2, 3], 2]) # ArrowInvalid: cannot mix list and non-list, non-null values\r\npa.array([[1, 2, 3], [2]]) # works\r\n``` ", "@mariosasko \r\nI used the same operation to check the original data before and after slicing.\r\nThis is reflected in my code.\r\n160000 is not a specific number.\r\nI can also get output using 150000.\r\nThis doesn't seem to align very well with what you said.\r\nBecause if only some sample formats are incorrect.\r\nSo there should be an error in one of the front and back slices.\r\nthank you for your reply.", "Our JSON loader does the following in your case:\r\n\r\n```python\r\nimport json\r\nimport pyarrow as pa\r\n\r\nwith open(file, encoding=\"utf-8\") as f:\r\n dataset = json.load(f)\r\nkeys = set().union(*[row.keys() for row in dataset])\r\nmapping = {col: [row.get(col) for row in dataset] for col in keys}\r\npa_table = pa.Table.from_pydict(mapping) # the ArrowInvalid error comes from here\r\n```\r\n\r\nSo if this code throws an error with correctly-formatted JSON, then this is an Arrow bug and should be reported in their repo.\r\n\r\n> I used the same operation to check the original data before and after slicing.\r\nThis is reflected in my code.\r\n160000 is not a specific number.\r\nI can also get output using 150000.\r\nThis doesn't seem to align very well with what you said.\r\nBecause if only some sample formats are incorrect.\r\nSo there should be an error in one of the front and back slices.\r\n\r\nYou should shuffle the data to make sure that's not the case", "@mariosasko \r\nThank you.\r\nI will try again." ]

[]

[]

[ "cc @sanchit-gandhi @Vaibhavs10 @lhoestq - this is mainly for demos that use Common Voice datasets as done here: https://github.com/facebookresearch/fairseq/tree/main/examples/mms#-transformers\r\n", "Hi ! the error for me is\r\n\r\n```\r\nFileNotFoundError: Couldn't find a dataset script at /content/mozilla-foundation/common_voice_13_0/common_voice_13_0.py or any data file in the same directory. Couldn't find 'mozilla-foundation/common_voice_13_0' on the Hugging Face Hub either: FileNotFoundError: Dataset 'mozilla-foundation/common_voice_13_0' doesn't exist on the Hub. If the repo is private or gated, make sure to log in with `huggingface-cli login`.\r\n```\r\n\r\nAnd tbh idk how you managed to get your error. \"n_shards.json\" is not even a thing in `datasets`", "Okay, I am able to reproduce @patrickvonplaten's original error: https://github.com/Vaibhavs10/scratchpad/blob/main/cv13_datasets_test.ipynb\r\n\r\nAlso not sure why it looks for `n_shards.json`", "Ok I see, this file is downloaded from the CV dataset script - let me investigate", "Ok I see: when you log out you no longer have access to the repository.\r\n\r\nTherefore the dataset script is loaded from cache:\r\n```\r\nWARNING:datasets.load:Using the latest cached version of the module from /root/.cache/huggingface/modules/datasets_modules/datasets/mozilla-foundation--common_voice_13_0/22809012aac1fc9803eaffc44122e4149043748e93933935d5ea19898587e4d7 (last modified on Wed Jun 14 10:13:17 2023) since it couldn't be found locally at mozilla-foundation/common_voice_13_0., or remotely on the Hugging Face Hub.\r\n```\r\n\r\nand the script tries to download the n_shards.json but fails", "Is this ok for you https://github.com/huggingface/datasets/pull/5954 ?\r\n\r\nI'll do a release this afternoon", "Cool! ", "this is included in the new release 2.13.0" ]