Datasets:

---

Francesco-A

/

github-issues_huggingface-datasets

like 0

[ { "color": "0e8a16", "default": false, "description": "Contribution to a dataset script", "id": 4564477500, "name": "dataset contribution", "node_id": "LA_kwDODunzps8AAAABEBBmPA", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset%20contribution" } ]

[ "Hi ! Thanks for adding this dataset :)\r\n\r\nRegarding your questions:\r\n1. We can have them as different configuration of the `rico` dataset\r\n2. Yes please use the path to the image and not open the image directly, so that we can let users open the image one at at time during training if they want to for example. In the future we'll have an Image feature type that will decode the encoded image data on the fly when accessing the examples.\r\n3. Feel free to keep the hierarchies as strings if they don't follow a fixed format\r\n4. You can just return the path\r\n\r\n", "Thanks for your contribution, @ncoop57. Are you still interested in adding this dataset?\r\n\r\nWe are removing the dataset scripts from this GitHub repo and moving them to the Hugging Face Hub: https://huggingface.co/datasets\r\n\r\nWe would suggest you create this dataset there. Please, feel free to tell us if you need some help." ]

[]

[]

[]

[]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "Cool! Can you please remove `Fix #3541` from the description as this PR doesn't add support for streaming/`iter_archive`, so it only partially addresses the issue?\r\n\r\nSide note:\r\nI think we can use `libarchive` (`libarchive-c` is a Python package with the bindings) for streaming 7z archives. The only issue with this lib is that it's tricky to install on Windows/Mac." ]

[ { "color": "7057ff", "default": true, "description": "Good for newcomers", "id": 1935892877, "name": "good first issue", "node_id": "MDU6TGFiZWwxOTM1ODkyODc3", "url": "https://api.github.com/repos/huggingface/datasets/labels/good%20first%20issue" } ]

[ "Hi, I would be glad to help with this. Is there anyone else working on it?", "Hi, I would be glad to work on this too.", "#self-assign", "Hi @Carlosbogo,\r\n\r\nwould you be interested in replacing the `.format` and `%` syntax with f-strings in the modules in the `datasets` directory since @Mehdi2402 has opened a PR that does that for all the other directories?", "Oh I see. I will be glad to help with the `datasets` directory then." ]

[]

[ "Hi ! Did you try to set the output format to pytorch ? (or tensorflow if you're using tensorflow)\r\nIt can be done with `dataset.set_format(\"torch\", columns=columns)` (or \"tensorflow\").\r\n\r\nNote that for pytorch, string columns can't be converted to `torch.Tensor`, so you have to specify in `columns=` the list of columns you want to keep (`input_ids` for example)", "Hello . Yes, I did set the output format as below for the two columns \r\n\r\n `train_dataset.set_format('torch',columns=['Text','Label'])`\r\n ", "I think you're having this issue because you try to format strings as pytorch tensors, which is not possible.\r\nIndeed by having \"Text\" in `columns=['Text','Label']`, you try to convert the text values to pytorch tensors.\r\n\r\nInstead I recommend you to first tokenize your dataset using a tokenizer from transformers. For example\r\n\r\n```python\r\nfrom transformers import BertTokenizer\r\ntokenizer = BertTokenizer.from_pretrained(\"bert-base-uncased\")\r\n\r\ntrain_dataset.map(lambda x: tokenizer(x[\"Text\"]), batched=True)\r\ntrain_dataset.set_format(\"torch\", column=[\"input_ids\"])\r\n```\r\n\r\nAnother way to fix your issue would be to not set the format to pytorch, and leave the dataset as it is by default. In that case, the strings are returned normally when you get examples from your dataloader. It means that you would have to tokenize the examples in the training loop (or using a data collator) though.\r\n\r\nLet me know if you have other questions", "Hi, actually the thing is I am getting the same error and even after tokenizing them I am passing them through batch_encode_plus.\r\nI dont know what seems to be the problem is. I even converted it into 'pt' while passing them through batch_encode_plus but when I am evaluating my model , i am getting this error\r\n\r\n\r\n---------------------------------------------------------------------------\r\nTypeError Traceback (most recent call last)\r\n<ipython-input-145-ca218223c9fc> in <module>()\r\n----> 1 val_loss, predictions, true_val = evaluate(dataloader_validation)\r\n 2 val_f1 = f1_score_func(predictions, true_val)\r\n 3 tqdm.write(f'Validation loss: {val_loss}')\r\n 4 tqdm.write(f'F1 Score (Weighted): {val_f1}')\r\n\r\n6 frames\r\n/usr/local/lib/python3.6/dist-packages/torch/utils/data/dataset.py in <genexpr>(.0)\r\n 160 \r\n 161 def __getitem__(self, index):\r\n--> 162 return tuple(tensor[index] for tensor in self.tensors)\r\n 163 \r\n 164 def __len__(self):\r\n\r\nTypeError: new(): invalid data type 'str' ", "> Hi, actually the thing is I am getting the same error and even after tokenizing them I am passing them through batch_encode_plus.\r\n> I dont know what seems to be the problem is. I even converted it into 'pt' while passing them through batch_encode_plus but when I am evaluating my model , i am getting this error\r\n> \r\n> TypeError Traceback (most recent call last)\r\n> in ()\r\n> ----> 1 val_loss, predictions, true_val = evaluate(dataloader_validation)\r\n> 2 val_f1 = f1_score_func(predictions, true_val)\r\n> 3 tqdm.write(f'Validation loss: {val_loss}')\r\n> 4 tqdm.write(f'F1 Score (Weighted): {val_f1}')\r\n> \r\n> 6 frames\r\n> /usr/local/lib/python3.6/dist-packages/torch/utils/data/dataset.py in (.0)\r\n> 160\r\n> 161 def **getitem**(self, index):\r\n> --> 162 return tuple(tensor[index] for tensor in self.tensors)\r\n> 163\r\n> 164 def **len**(self):\r\n> \r\n> TypeError: new(): invalid data type 'str'\r\n\r\nI got the same error and fix it .\r\nyou can check your input where there may be string contained.\r\nsuch as\r\n```\r\na = [1,2,3,4,'<unk>']\r\ntorch.tensor(a)\r\n```", "I didn't know tokenizers could return strings in the token ids. Which tokenizer are you using to get this @Doragd ?", "> I didn't know tokenizers could return strings in the token ids. Which tokenizer are you using to get this @Doragd ?\r\n\r\ni'm sorry that i met this issue in another place (not in huggingface repo). ", "@akhilkapil do you have strings in your dataset ? When you set the dataset format to \"pytorch\" you should exclude columns with strings as pytorch can't make tensors out of strings", "Closing due to inactivity." ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

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

[]

[]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "I have received a reply from Zenodo support:\r\n> We are currently investigating and fixing this issue related to GitHub releases. As soon as we have solved it we will reach back to you.", "Other repo maintainers had the same problem with Zenodo. \r\n\r\nThere is an open issue on their GitHub repo: zenodo/zenodo#2181", "I have received the following request from Zenodo support:\r\n> Could you send us the link to the repository as well as the release tag?\r\n\r\nMy reply:\r\n> Sure, here it is:\r\n> - Link to the repository: https://github.com/huggingface/datasets\r\n> - Link to the repository at the release tag: https://github.com/huggingface/datasets/releases/tag/1.8.0\r\n> - Release tag: 1.8.0", "Zenodo issue has been fixed. The 1.8.0 release DOI can be found here: https://zenodo.org/record/4946100#.YMd6vKj7RPY" ]

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

[ "Hi @mariosasko unless someone's already working on it, I guess I can tackle it!", "Hi! Sure, feel free to tackle this.", "#self-assign", "I'm assuming this should just cover `to_csv`, `to_parquet`, and `to_json`, right? As `to_list` and `to_dict` just return Python objects, `to_pandas` returns a `pandas.DataFrame` and `to_sql` just inserts into a SQL DB, is that right?" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008167 / 0.011353 (-0.003186) | 0.004404 / 0.011008 (-0.006605) | 0.100462 / 0.038508 (0.061954) | 0.028835 / 0.023109 (0.005726) | 0.326759 / 0.275898 (0.050861) | 0.355150 / 0.323480 (0.031670) | 0.007200 / 0.007986 (-0.000786) | 0.003293 / 0.004328 (-0.001035) | 0.078006 / 0.004250 (0.073756) | 0.033298 / 0.037052 (-0.003754) | 0.307119 / 0.258489 (0.048630) | 0.337689 / 0.293841 (0.043848) | 0.033016 / 0.128546 (-0.095530) | 0.011383 / 0.075646 (-0.064263) | 0.321989 / 0.419271 (-0.097283) | 0.039793 / 0.043533 (-0.003740) | 0.295388 / 0.255139 (0.040249) | 0.322694 / 0.283200 (0.039494) | 0.082989 / 0.141683 (-0.058694) | 1.496701 / 1.452155 (0.044546) | 1.548861 / 1.492716 (0.056145) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.176587 / 0.018006 (0.158580) | 0.397660 / 0.000490 (0.397170) | 0.001063 / 0.000200 (0.000863) | 0.000070 / 0.000054 (0.000015) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.022386 / 0.037411 (-0.015025) | 0.096380 / 0.014526 (0.081854) | 0.103032 / 0.176557 (-0.073525) | 0.135050 / 0.737135 (-0.602086) | 0.105941 / 0.296338 (-0.190397) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.430989 / 0.215209 (0.215780) | 4.310309 / 2.077655 (2.232654) | 2.142596 / 1.504120 (0.638477) | 1.952043 / 1.541195 (0.410848) | 1.817803 / 1.468490 (0.349312) | 0.690026 / 4.584777 (-3.894751) | 3.315413 / 3.745712 (-0.430299) | 3.370336 / 5.269862 (-1.899525) | 1.668707 / 4.565676 (-2.896970) | 0.081860 / 0.424275 (-0.342415) | 0.012493 / 0.007607 (0.004886) | 0.527779 / 0.226044 (0.301735) | 5.318732 / 2.268929 (3.049804) | 2.467029 / 55.444624 (-52.977596) | 2.247171 / 6.876477 (-4.629306) | 2.270825 / 2.142072 (0.128752) | 0.802288 / 4.805227 (-4.002939) | 0.148895 / 6.500664 (-6.351770) | 0.064967 / 0.075469 (-0.010503) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.259304 / 1.841788 (-0.582484) | 13.662441 / 8.074308 (5.588133) | 14.074662 / 10.191392 (3.883270) | 0.152907 / 0.680424 (-0.527516) | 0.028340 / 0.534201 (-0.505861) | 0.397356 / 0.579283 (-0.181927) | 0.392600 / 0.434364 (-0.041764) | 0.467935 / 0.540337 (-0.072402) | 0.539890 / 1.386936 (-0.847046) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006156 / 0.011353 (-0.005197) | 0.004371 / 0.011008 (-0.006637) | 0.076391 / 0.038508 (0.037883) | 0.026455 / 0.023109 (0.003346) | 0.339816 / 0.275898 (0.063917) | 0.370032 / 0.323480 (0.046552) | 0.004614 / 0.007986 (-0.003372) | 0.003200 / 0.004328 (-0.001129) | 0.075408 / 0.004250 (0.071157) | 0.034100 / 0.037052 (-0.002953) | 0.341232 / 0.258489 (0.082743) | 0.380290 / 0.293841 (0.086449) | 0.031021 / 0.128546 (-0.097525) | 0.011562 / 0.075646 (-0.064084) | 0.085564 / 0.419271 (-0.333708) | 0.041431 / 0.043533 (-0.002102) | 0.359570 / 0.255139 (0.104431) | 0.366919 / 0.283200 (0.083719) | 0.088242 / 0.141683 (-0.053441) | 1.460703 / 1.452155 (0.008548) | 1.534351 / 1.492716 (0.041635) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.225703 / 0.018006 (0.207697) | 0.395014 / 0.000490 (0.394524) | 0.000385 / 0.000200 (0.000185) | 0.000060 / 0.000054 (0.000006) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.023975 / 0.037411 (-0.013436) | 0.098658 / 0.014526 (0.084132) | 0.105043 / 0.176557 (-0.071513) | 0.139988 / 0.737135 (-0.597148) | 0.106854 / 0.296338 (-0.189484) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.442454 / 0.215209 (0.227245) | 4.430860 / 2.077655 (2.353205) | 2.084823 / 1.504120 (0.580704) | 1.870421 / 1.541195 (0.329226) | 1.901618 / 1.468490 (0.433128) | 0.699214 / 4.584777 (-3.885563) | 3.336911 / 3.745712 (-0.408801) | 1.856479 / 5.269862 (-3.413383) | 1.166496 / 4.565676 (-3.399180) | 0.083189 / 0.424275 (-0.341086) | 0.012293 / 0.007607 (0.004686) | 0.543147 / 0.226044 (0.317102) | 5.452030 / 2.268929 (3.183101) | 2.506689 / 55.444624 (-52.937936) | 2.168186 / 6.876477 (-4.708291) | 2.172277 / 2.142072 (0.030205) | 0.813554 / 4.805227 (-3.991673) | 0.152074 / 6.500664 (-6.348590) | 0.066891 / 0.075469 (-0.008579) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.278635 / 1.841788 (-0.563153) | 13.690232 / 8.074308 (5.615924) | 13.403201 / 10.191392 (3.211809) | 0.128171 / 0.680424 (-0.552253) | 0.016687 / 0.534201 (-0.517514) | 0.378645 / 0.579283 (-0.200638) | 0.382922 / 0.434364 (-0.051442) | 0.467483 / 0.540337 (-0.072854) | 0.559026 / 1.386936 (-0.827910) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#b262d411ec0e252615a140c4e3e60e7dbd38eef1 \"CML watermark\")\n" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "Thanks !\r\nI'm in favor of this change, even though it's a breaking change:\r\n\r\nif you had a dataset\r\n```\r\ndata/\r\n train.csv\r\n test.csv\r\n```\r\n\r\nthen running this code would now return both train and test splits:\r\n```python\r\nload_dataset(\"csv\", data_dir=\"data/\")\r\n```\r\nwhereas right now it returns only a train split with the data from both CSV files.\r\n\r\nIn my opinion it's ok do do this breaking change because:\r\n- it makes this behavior consistent with `load_dataset(\"path/to/data\")` that also returns both splits: data_files resolution must be the same\r\n- I don't expect too many affected users (unless people really wanted to group train and test images in the train split on purpose ?) compared to the many new users to come (especially with #4069 )\r\n- this usage will become more and more common as we add packaged builder and imagefolder/audiofolder usage grows, so it may be better to do this change early\r\n\r\nLet me know if you think this is acceptable @mariosasko @albertvillanova or not, and if you think we need to first have a warning for some time before switching to this new behavior", "Also, if people really want to put train and test, say, images in a single train split they could do \r\n`load_dataset(\"imagefolder\", data_files={\"train\": \"/path/to/data/**})`. Probably (arguably :)), if this is a more counterintuitive case, then it should require manual files specification, not a default one (in which we expect that users do want to infer splits from filenames / dir structure but currently they have to pass smth like `{\"train\": \"/path/to/data/train*\", \"test\": \"/path/to/data/test*\"}` explicitly as `data_files`) ", "I also like this change, and I don't think we even need a warning during the transition period, considering I've been asked several times since the release of `imagefolder` why splits are not correctly inferred if the directory structure is as follows:\r\n```\r\ndata_dir\r\n train\r\n label_a\r\n 0.jpg\r\n ...\r\n label_b \r\n 0.jpg\r\n ...\r\n test\r\n label_a\r\n 0.jpg\r\n ...\r\n label_b \r\n 0.jpg\r\n ...\r\n```", "Cool ! Feel free to add a test (maybe something similar to `test_PackagedDatasetModuleFactory_with_data_dir` but with a data_dir that contains several splits) and mark this PR as ready for review then @polinaeterna :)", "@lhoestq @mariosasko do you think it's a good idea to do the same with `HubDatasetModuleFactoryWithoutScript` and `LocalDatasetModuleFactoryWithoutScript` (see the latest change). If we agree on the current change, doing \r\n```python\r\nds = load_dataset(\"polinaeterna/jsonl_test\", data_dir=\"data/\")\r\n```\r\non dataset with the following structure:\r\n```\r\ntrain.jsonl\r\ntest.jsonl\r\ndata/\r\n train.jsonl\r\n test.jsonl\r\n```\r\nwill result in having two splits from files under `data/` dir in specified repo, while master version returns a single train split. \r\nThe same would be for local dataset without script if doing smth like:\r\n```python\r\nds = load_dataset(\"/home/polina/workspace/repos/jsonl_test\", data_dir=\"/home/polina/workspace/repos/jsonl_test/data\")\r\n```\r\n(though I'm not sure I understand this use case :D)\r\nLet me know if you think we should preserve the same logic for all factories or if I should roll back this change.", "@lhoestq to test passing subdirectory (`base_path`) to data_files functions and methods, I extended the temporary test directory with data so that it contains subdirectory. Because of that the number of files in this directory increased, so I had to change some numbers and patterns to account for this change - [907ddf0](https://github.com/huggingface/datasets/pull/4144/commits/907ddf09d3afece5afbae18675c859d6e453f2bf)\r\n\r\nDo you think it's ok? Another option is to create another tmp dir and do all the checks inside it. " ]

[]

[ "Closed via #1252 " ]

[]

[]

[]

[]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "Added tests for xisfile, xgetsize, xlistdir and xglob for private repos, and also tests for xwalk that was untested" ]

[]

[ "And I don't see any speed up when increasing the number of GPUs while calling `get_nearest_examples_batch`.", "Hi ! Yes search_batch uses FAISS search which happens in parallel across the GPUs\r\n\r\n> And I don't see any speed up when increasing the number of GPUs while calling get_nearest_examples_batch.\r\n\r\nThat's unexpected, can you share the code you're running ?", "here is the code snippet\r\n\r\n```python\r\n\r\n# add faiss index\r\nsource_dataset = load_dataset(source_path)\r\nqueries = load_dataset(query_path)\r\ngpu = [0,1,2,3]\r\nsource_dataset.add_faiss_index(\r\n \"embedding\",\r\n device=gpu,\r\n )\r\n\r\n\r\n# batch query\r\nbatch_size = 32\r\nfor i in tqdm(range(0, len(queries), batch_size)):\r\n if i + batch_size >= len(queries):\r\n batched_queries = queries[i:]\r\n else:\r\n batched_queries = queries[i:i+batch_size]\r\n\r\n batched_query_embeddings = np.stack([i for i in batched_queries['embedding']], axis=0)\r\n scores, candidates = source_dataset.get_nearest_examples_batch(\r\n \"embedding\",\r\n batched_query_embeddings,\r\n k=5\r\n )\r\n```", "My version of datasets is `2.4.1.dev0`.", "The code looks all good to me, do you see all the GPUs being utilized ? What version of faiss are you using ?", "I can see the memory usage of all the GPUs.\r\nMy version of `faiss-gpu` is `1.7.2`", "It looks all good to me then ^^ though you said you didn't experienced speed improvements by adding more GPUs ? What size is your source dataset and what time differences did you experience ?", "query set: 1e6\r\nsource dataset: 1e6\r\nembedding size: 768\r\nindex: Flat\r\ntopk: 20\r\nGPU: V100\r\n\r\nThe time taken to traverse the query set once is about 1.5h, which is almost not influenced by the value of query batch size or the number of GPUs according to my experiments.", "Hmmm the number of GPUs should divide the time, something is going wrong. Can you check that adding more GPU does divide the memory used per GPU ? Maybe it can be worth looking at similar issues in the FAISS repository or create a noew issue over there to understand what's going on", "> Can you check that adding more GPU does divide the memory used per GPU \r\n\r\nThe memory used per GPU is unchanged while adding more GPU. Is this unexpected?\r\n\r\nI used to think that every GPU loads all the source vectors and the data parallelism is at the query level. 😆 ", "> I used to think that every GPU loads all the source vectors and the data parallelism is at the query level. 😆\r\n\r\nOh indeed that's possible, I wasn't sure. Anyway you can check that calling get_nearest_examples_batch simply calls search under the hood: \r\n\r\nhttps://github.com/huggingface/datasets/blob/f90f71fbbb33889fe75a3ffc101cdf16a88a3453/src/datasets/search.py#L375", "Here is a runnable script. \r\nMulti-GPU searching still does not work in my experiments.\r\n\r\n\r\n```python\r\nimport os\r\nfrom tqdm import tqdm\r\nimport numpy as np\r\nimport datasets\r\nfrom datasets import Dataset\r\n\r\nclass DPRSelector:\r\n\r\n def __init__(self, source, target, index_name, gpu=None):\r\n self.source = source\r\n self.target = target\r\n self.index_name = index_name\r\n\r\n cache_path = 'embedding.faiss'\r\n\r\n if not os.path.exists(cache_path):\r\n self.source.add_faiss_index(\r\n column=\"embedding\",\r\n index_name=index_name,\r\n device=gpu,\r\n )\r\n self.source.save_faiss_index(index_name, cache_path)\r\n else:\r\n self.source.load_faiss_index(\r\n index_name,\r\n cache_path,\r\n device=gpu\r\n )\r\n print('index builded!')\r\n\r\n def build_dataset(self, top_k, batch_size):\r\n print('start search')\r\n\r\n for i in tqdm(range(0, len(self.target), batch_size)):\r\n if i + batch_size >= len(self.target):\r\n batched_queries = self.target[i:]\r\n else:\r\n batched_queries = self.target[i:i+batch_size]\r\n\r\n\r\n batched_query_embeddings = np.stack([i for i in batched_queries['embedding']], axis=0)\r\n search_res = self.source.get_nearest_examples_batch(\r\n self.index_name,\r\n batched_query_embeddings,\r\n k=top_k\r\n )\r\n \r\n print('finish search')\r\n\r\n\r\ndef get_pseudo_dataset():\r\n pseudo_dict = {\"embedding\": np.zeros((1000000, 768), dtype=np.float32)}\r\n print('generate pseudo data')\r\n\r\n dataset = Dataset.from_dict(pseudo_dict)\r\n def list_to_array(data):\r\n return {\"embedding\": [np.array(vector, dtype=np.float32) for vector in data[\"embedding\"]]} \r\n dataset.set_transform(list_to_array, columns='embedding', output_all_columns=True)\r\n\r\n print('build dataset')\r\n return dataset\r\n\r\n\r\n\r\nif __name__==\"__main__\":\r\n\r\n np.random.seed(42)\r\n\r\n\r\n source_dataset = get_pseudo_dataset()\r\n target_dataset = get_pseudo_dataset()\r\n\r\n gpu = [0,1,2,3,4,5,6,7]\r\n selector = DPRSelector(source_dataset, target_dataset, \"embedding\", gpu=gpu)\r\n\r\n selector.build_dataset(top_k=20, batch_size=32)\r\n```", "@lhoestq Hi, could you please test the code above if you have time? 😄 ", "Maybe @albertvillanova you can take a look ? I won't be available in the following days", "@albertvillanova Hi, can you help with this issue?", "Hi @xwwwwww I'm investigating it, but I'm not an expert in Faiss. In principle, it is weird that your code does not work properly because it seems right...", "Have you tried passing `gpu=-1` and check if there is a speedup?", "> Have you tried passing `gpu=-1` and check if there is a speedup?\r\n\r\nyes, there is a speed up using GPU compared with CPU. ", "When passing `device=-1`, ALL existing GPUs are used (multi GPU): this is the maximum speedup you can get. To know the number of total GPUs:\r\n```\r\nimport faiss\r\n\r\nngpus = faiss.get_num_gpus()\r\nprint(ngpus)\r\n```\r\n\r\nWhen passing a list of integers to `device`, then only that number of GPUs are used (multi GPU as well)\r\n- the speedup should be proportional (more or less) to the ratio of the number of elements passed to `device` over `ngpus`\r\n- if this is not the case, then there is an issue in the implementation of this use case (however, I have reviewed the code and in principle I can't find any evident bug)\r\n\r\nWhen passing a positive integer to `device`, then only a single GPU is used.\r\n- this time should be more or less proportional to the time when passing `device=-1` over `ngpus`", "Thanks for your help!\r\nHave you run the code and replicated the same experimental results (i.e., no speedup while increasing the number of GPUs)?", "@albertvillanova @lhoestq Sorry for the bother, is there any progress on this issue? 😃 ", "I can confirm `add_faiss_index` calls `index = faiss.index_cpu_to_gpus_list(index, gpus=list(device))`.\r\n\r\nCould this be an issue with your environment ? Could you try running with 1 and 8 GPUs with a code similar to[ this one from the FAISS examples](https://github.com/facebookresearch/faiss/blob/main/tutorial/python/5-Multiple-GPUs.py) but using `gpu_index = faiss.index_cpu_to_gpus_list(cpu_index, gpus=list(device))`, and see if the speed changes ?", "Hi, I test the FAISS example and the speed indeed changes. I set `nb=1000000`, `nq=1000000` and `d=64`\r\n\r\n| num GPUS | time cost |\r\n| -------- | --------- |\r\n| 1 | 28.53 |\r\n| 5 | 7.16 |\r\n\r\n\r\n\r\n", "Ok the benchmark is great, not sure why it doesn't speed up the index in your case though. You can try running the benchmark with the same settings as your actual dataset\r\n```\r\nquery set: 1e6\r\nsource dataset: 1e6\r\nembedding size: 768\r\nindex: Flat\r\ntopk: 20\r\nGPU: V100\r\n```\r\n\r\nNote that you can still pass a FAISS index you built yourself to a dataset using https://huggingface.co/docs/datasets/v2.4.0/en/package_reference/main_classes#datasets.Dataset.add_faiss_index_from_external_arrays", "> Here is a runnable script. Multi-GPU searching still does not work in my experiments.\r\n> \r\n> ```python\r\n> import os\r\n> from tqdm import tqdm\r\n> import numpy as np\r\n> import datasets\r\n> from datasets import Dataset\r\n> \r\n> class DPRSelector:\r\n> \r\n> def __init__(self, source, target, index_name, gpu=None):\r\n> self.source = source\r\n> self.target = target\r\n> self.index_name = index_name\r\n> \r\n> cache_path = 'embedding.faiss'\r\n> \r\n> if not os.path.exists(cache_path):\r\n> self.source.add_faiss_index(\r\n> column=\"embedding\",\r\n> index_name=index_name,\r\n> device=gpu,\r\n> )\r\n> self.source.save_faiss_index(index_name, cache_path)\r\n> else:\r\n> self.source.load_faiss_index(\r\n> index_name,\r\n> cache_path,\r\n> device=gpu\r\n> )\r\n> print('index builded!')\r\n> \r\n> def build_dataset(self, top_k, batch_size):\r\n> print('start search')\r\n> \r\n> for i in tqdm(range(0, len(self.target), batch_size)):\r\n> if i + batch_size >= len(self.target):\r\n> batched_queries = self.target[i:]\r\n> else:\r\n> batched_queries = self.target[i:i+batch_size]\r\n> \r\n> \r\n> batched_query_embeddings = np.stack([i for i in batched_queries['embedding']], axis=0)\r\n> search_res = self.source.get_nearest_examples_batch(\r\n> self.index_name,\r\n> batched_query_embeddings,\r\n> k=top_k\r\n> )\r\n> \r\n> print('finish search')\r\n> \r\n> \r\n> def get_pseudo_dataset():\r\n> pseudo_dict = {\"embedding\": np.zeros((1000000, 768), dtype=np.float32)}\r\n> print('generate pseudo data')\r\n> \r\n> dataset = Dataset.from_dict(pseudo_dict)\r\n> def list_to_array(data):\r\n> return {\"embedding\": [np.array(vector, dtype=np.float32) for vector in data[\"embedding\"]]} \r\n> dataset.set_transform(list_to_array, columns='embedding', output_all_columns=True)\r\n> \r\n> print('build dataset')\r\n> return dataset\r\n> \r\n> \r\n> \r\n> if __name__==\"__main__\":\r\n> \r\n> np.random.seed(42)\r\n> \r\n> \r\n> source_dataset = get_pseudo_dataset()\r\n> target_dataset = get_pseudo_dataset()\r\n> \r\n> gpu = [0,1,2,3,4,5,6,7]\r\n> selector = DPRSelector(source_dataset, target_dataset, \"embedding\", gpu=gpu)\r\n> \r\n> selector.build_dataset(top_k=20, batch_size=32)\r\n> ```\r\n\r\nBy the way, have you run this toy example and replicated my experiment results? I think it is a more direct way to figure this out :)", "Hi,\r\n\r\nI have a similar question and would like to know if there's any progress in this issue. \r\n\r\n`dataset.add_faiss_index(column=\"embedding\")`, this takes around 5minutes to add the index.\r\n\r\n`dataset.add_faiss_index(column=\"embedding\", device=-1)`, this ran for more than 10minutes and still didn't complete execution. \r\n\r\nNow, I don't understand why that's the case as I expected for GPU the indexing should be faster" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "@mishig25 note that it's not exactly the same behavior as in nginx as here it interacts a bit with the `version` and the `language`\r\n\r\nShould be close enough, though.", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007212 / 0.011353 (-0.004141) | 0.005125 / 0.011008 (-0.005883) | 0.098460 / 0.038508 (0.059952) | 0.034040 / 0.023109 (0.010931) | 0.320203 / 0.275898 (0.044305) | 0.357787 / 0.323480 (0.034307) | 0.006000 / 0.007986 (-0.001986) | 0.005644 / 0.004328 (0.001316) | 0.072654 / 0.004250 (0.068403) | 0.049393 / 0.037052 (0.012341) | 0.345686 / 0.258489 (0.087196) | 0.362345 / 0.293841 (0.068504) | 0.036597 / 0.128546 (-0.091949) | 0.012303 / 0.075646 (-0.063343) | 0.334374 / 0.419271 (-0.084897) | 0.062010 / 0.043533 (0.018477) | 0.312547 / 0.255139 (0.057408) | 0.336021 / 0.283200 (0.052821) | 0.112304 / 0.141683 (-0.029378) | 1.446706 / 1.452155 (-0.005449) | 1.523256 / 1.492716 (0.030540) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.217658 / 0.018006 (0.199652) | 0.449208 / 0.000490 (0.448718) | 0.002878 / 0.000200 (0.002679) | 0.000091 / 0.000054 (0.000036) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025735 / 0.037411 (-0.011676) | 0.105876 / 0.014526 (0.091350) | 0.114887 / 0.176557 (-0.061669) | 0.170984 / 0.737135 (-0.566152) | 0.121420 / 0.296338 (-0.174918) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.419670 / 0.215209 (0.204461) | 4.189453 / 2.077655 (2.111798) | 1.938236 / 1.504120 (0.434116) | 1.769747 / 1.541195 (0.228553) | 1.910919 / 1.468490 (0.442429) | 0.705046 / 4.584777 (-3.879730) | 3.783774 / 3.745712 (0.038062) | 2.096504 / 5.269862 (-3.173358) | 1.339265 / 4.565676 (-3.226412) | 0.086670 / 0.424275 (-0.337605) | 0.012243 / 0.007607 (0.004636) | 0.524701 / 0.226044 (0.298657) | 5.240689 / 2.268929 (2.971760) | 2.473622 / 55.444624 (-52.971003) | 2.170568 / 6.876477 (-4.705909) | 2.289653 / 2.142072 (0.147581) | 0.848913 / 4.805227 (-3.956314) | 0.168332 / 6.500664 (-6.332332) | 0.064926 / 0.075469 (-0.010543) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.193614 / 1.841788 (-0.648173) | 14.920403 / 8.074308 (6.846095) | 14.475059 / 10.191392 (4.283667) | 0.164458 / 0.680424 (-0.515966) | 0.017613 / 0.534201 (-0.516588) | 0.426311 / 0.579283 (-0.152972) | 0.431478 / 0.434364 (-0.002886) | 0.520280 / 0.540337 (-0.020057) | 0.627738 / 1.386936 (-0.759198) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007458 / 0.011353 (-0.003895) | 0.005363 / 0.011008 (-0.005645) | 0.076713 / 0.038508 (0.038205) | 0.034189 / 0.023109 (0.011079) | 0.359938 / 0.275898 (0.084040) | 0.395532 / 0.323480 (0.072052) | 0.005977 / 0.007986 (-0.002008) | 0.004263 / 0.004328 (-0.000065) | 0.075971 / 0.004250 (0.071721) | 0.051924 / 0.037052 (0.014871) | 0.362818 / 0.258489 (0.104329) | 0.409897 / 0.293841 (0.116056) | 0.035494 / 0.128546 (-0.093053) | 0.012399 / 0.075646 (-0.063247) | 0.088335 / 0.419271 (-0.330937) | 0.047968 / 0.043533 (0.004435) | 0.355744 / 0.255139 (0.100606) | 0.376339 / 0.283200 (0.093139) | 0.104542 / 0.141683 (-0.037141) | 1.464826 / 1.452155 (0.012672) | 1.600665 / 1.492716 (0.107948) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.220841 / 0.018006 (0.202834) | 0.446444 / 0.000490 (0.445954) | 0.000392 / 0.000200 (0.000192) | 0.000057 / 0.000054 (0.000003) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029402 / 0.037411 (-0.008009) | 0.116511 / 0.014526 (0.101986) | 0.122959 / 0.176557 (-0.053598) | 0.171674 / 0.737135 (-0.565462) | 0.129871 / 0.296338 (-0.166468) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.450411 / 0.215209 (0.235202) | 4.471859 / 2.077655 (2.394205) | 2.229439 / 1.504120 (0.725319) | 2.053308 / 1.541195 (0.512114) | 2.142476 / 1.468490 (0.673986) | 0.708299 / 4.584777 (-3.876478) | 3.797830 / 3.745712 (0.052118) | 2.142509 / 5.269862 (-3.127352) | 1.333357 / 4.565676 (-3.232320) | 0.086837 / 0.424275 (-0.337439) | 0.012102 / 0.007607 (0.004495) | 0.548428 / 0.226044 (0.322384) | 5.490611 / 2.268929 (3.221682) | 2.713882 / 55.444624 (-52.730742) | 2.399638 / 6.876477 (-4.476839) | 2.481549 / 2.142072 (0.339477) | 0.839812 / 4.805227 (-3.965415) | 0.168890 / 6.500664 (-6.331774) | 0.065564 / 0.075469 (-0.009906) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.275507 / 1.841788 (-0.566281) | 14.896343 / 8.074308 (6.822035) | 13.159701 / 10.191392 (2.968309) | 0.172065 / 0.680424 (-0.508359) | 0.017507 / 0.534201 (-0.516694) | 0.420031 / 0.579283 (-0.159252) | 0.438835 / 0.434364 (0.004471) | 0.490597 / 0.540337 (-0.049741) | 0.583952 / 1.386936 (-0.802984) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#48c9755d0ae9abe4c4d6cd8c1ce76eff849f0e5c \"CML watermark\")\n" ]

[]

[ "I notice a comment in the code says:\r\n`Having lists of different sizes makes sharding ambigious, raise an error in this case until we decide how to define sharding without ambiguity for users` \r\n \r\n ... which suggests this update was pushed knowing that it might break some things. But, it didn't seem to have a useful error message of an argument that could be passed to avoid the error.", "Sorry for the inconvenience, I opened a PR in your repo to fix this: https://huggingface.co/datasets/sil-ai/bloom-speech/discussions/2\r\n\r\nBasically we've always considered lists in `gen_kwargs` to be a shard list that we can split and pass into different workers to generate the dataset (e.g. if you pass `num_proc=` in `load_dataset()` to generate the dataset in parallel), but it was documented only recently", "@lhoestq Thanks for the help. It looks like that took care of it." ]

[]

[ "I wrongly commited data from another branch in this PR, I'll close this a reopen another PR with the fixed branch" ]

[]

[ "I tried lower batch sizes and it didn't accelerate filter (quite the opposite actually).\r\nThe slow-down also appears for pyarrow 0.17.1 for some reason, not sure it comes from these changes", "I just checked the benchmarks of other PRs and some of them had 300s (!!) for filter. This needs some investigation..", "Merging this one since it's not the cause of the the slow down" ]

[]

[ "Hi !\r\nWe're not using `TextDatasetForNextSentencePrediction` in `datasets`.\r\nAlthough you can probably use the `TextDatasetForNextSentencePrediction.create_examples_from_document` on a dataset to prepare it for next sentence prediction.", "Thanks.\r\n\r\nDo you mean that `TextDatasetForNextSentencePrediction.create_exapmles_from_document` can be applied to dataset object other than `TextDatasetForNextSentencePrediction` e.g. a `Dataset` object which is loaded by `datasets.load_dataset`?", "It would probably require a bit of tweaking, but you can apply it to a dataset, yes.\r\nThis should give you a new dataset with sentence pairs you can train a model on.\r\n\r\nYou can find the documentation about dataset processing here:\r\nhttps://huggingface.co/docs/datasets/processing.html#processing-data-with-map", "Thank you for detail information.\r\n\r\nI'll try to apply `create_examples_from_document` to `Dataset` object.\r\n" ]

[]

[ "If the windows tests in the CI pass, today will be a happy day" ]

[]

[ "Added the dataset card.\r\nRequesting another review." ]

[]

[ "@sebastianGehrmann " ]

[]

[]

[]

[]

[]

[]

[]

[ "is this an issue ?", "Yes @lhoestq, look at the docs site. For example, in https://huggingface.co/docs/datasets/add_dataset.html, look at the indentation in the code block under the sentence:\r\n> Here are the features of the SQuAD dataset for instance, which is taken from the squad dataset loading script:" ]

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

[ "Hi ! For now this is probably the best option.\r\nWe might add a feature like this in the feature as well.\r\n\r\nDo you know any deduplication method that works on arbitrary big datasets without filling up RAM ?\r\nOtherwise we can have do the deduplication in memory like pandas but I feel like this is going to be limiting for some cases", "Yes, I'd like to work on this feature once I'm done with #2500, but first I have to do some research, and see if the implementation wouldn't be too complex.\r\n\r\nIn the meantime, maybe [this lib](https://github.com/TomScheffers/pyarrow_ops) can help. However, note that this lib operates directly on pyarrow tables and relies only on `hash` to find duplicates (e.g. `-1` and `-2` have the same hash in Python 3, so this lib will treat them as duplicates), which doesn't make much sense.", "> Hi ! For now this is probably the best option.\r\n> We might add a feature like this in the feature as well.\r\n> \r\n> Do you know any deduplication method that works on arbitrary big datasets without filling up RAM ?\r\n> Otherwise we can have do the deduplication in memory like pandas but I feel like this is going to be limiting for some cases\r\n\r\nGreat if this is can be done. Thanks!!\r\n\r\nNot sure if you are asking me. In any case I don't know of any unfortunately :( in practice if data is really large we normally do it with spark (only for info. I understand this is not useful in developing this library..)", "Hello,\r\n\r\nI'm also interested in this feature.\r\nHas there been progress on this issue?\r\n\r\nCould we use a similar trick as above, but with a better hashing algorithm like SHA?\r\n\r\nWe could also use a [bloom filter](https://en.wikipedia.org/wiki/Bloom_filter), should we care a lot about collision in this case?", "For reference, we can get a solution fairly easily if we assume that we can hold in memory all unique values. \r\n\r\n```python\r\nfrom datasets import Dataset\r\nfrom itertools import cycle\r\nfrom functools import partial\r\n\r\nmemory = set()\r\ndef is_unique(elem:Any , column: str, memory: set) -> bool:\r\n if elem[column] in memory:\r\n return False\r\n else:\r\n memory.add(elem[column])\r\n return True\r\n\r\n# Example dataset\r\nds = Dataset.from_dict({\"col1\" : [sent for i, sent in zip(range(10), cycle([\"apple\", \"orange\", \"pear\"]))],\r\n \"col2\": [i % 5 for i in range(10)]})\r\n\r\n# Drop duplicates in `ds` on \"col1\"\r\nds2 = ds.filter(partial(is_unique, column=\"col1\", memory=memory))\r\n```\r\n\r\nOf course, we can improve the API so that we can introduce `Dataset.drop_duplicates`.\r\nFor the parallel version, we can use a shared memory set.", "An approach that works assuming you can hold the all the unique document hashes in memory:\r\n```python\r\nfrom datasets import load_dataset\r\n\r\ndef get_hash(example):\r\n \"\"\"Get hash of content field.\"\"\"\r\n return {\"hash\": hash(example[\"content\"])} # can use any hashing function here\r\n \r\ndef check_uniques(example, uniques):\r\n \"\"\"Check if current hash is still in set of unique hashes and remove if true.\"\"\"\r\n if example[\"hash\"] in uniques:\r\n uniques.remove(example[\"hash\"])\r\n return True\r\n else:\r\n return False\r\n\r\nds = load_dataset(\"some_dataset\")\r\nds = ds.map(get_hash)\r\nuniques = set(ds.unique(\"hash\"))\r\nds_filter = ds.filter(check_uniques, fn_kwargs={\"uniques\": uniques})\r\n```\r\nIf the `uniques` could be stored in arrow then no additional memory would used at all but I don't know if this is possible.\r\n", "@lvwerra hey, could you tell me how reliable is this deduplication method. i am currently using the same deduplication strategy to deduplicate a large text corpus to pretrain LLMs ~ 11B to 20B. just needed to ensure if this strategy would be fine on large datasets for LLMs pretraining. " ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "I faced the similar problem. Downgrading datasets to 1.5.0 fixed it.", "Thanks for reporting, I'm looking into it", "I just opened #2438 to fix this :)", "Hi ! This has been fixed in the 1.8.0 release of `datasets`" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "> I think we should also modify one of the metadata files in the `folder_based_builder` tests to make sure \"./\" is ignored now in the `file_name`\r\n\r\n@mariosasko what do you mean here? I'm not sure which metadata file I should modify here", "You can modify this line for instance: https://github.com/huggingface/datasets/blob/2699593b33ee63d17aad2a2bfddedd38a8df57b8/tests/packaged_modules/test_folder_based_builder.py#L135" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007782 / 0.011353 (-0.003571) | 0.004451 / 0.011008 (-0.006557) | 0.099928 / 0.038508 (0.061420) | 0.081534 / 0.023109 (0.058425) | 0.379382 / 0.275898 (0.103484) | 0.410652 / 0.323480 (0.087172) | 0.005967 / 0.007986 (-0.002019) | 0.003702 / 0.004328 (-0.000627) | 0.076359 / 0.004250 (0.072109) | 0.066721 / 0.037052 (0.029669) | 0.383595 / 0.258489 (0.125106) | 0.423854 / 0.293841 (0.130013) | 0.032796 / 0.128546 (-0.095750) | 0.009728 / 0.075646 (-0.065918) | 0.344347 / 0.419271 (-0.074925) | 0.056320 / 0.043533 (0.012788) | 0.379974 / 0.255139 (0.124835) | 0.401294 / 0.283200 (0.118094) | 0.024110 / 0.141683 (-0.117572) | 1.804194 / 1.452155 (0.352039) | 1.860240 / 1.492716 (0.367523) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.233803 / 0.018006 (0.215797) | 0.506893 / 0.000490 (0.506404) | 0.003894 / 0.000200 (0.003694) | 0.000090 / 0.000054 (0.000035) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.033328 / 0.037411 (-0.004083) | 0.098661 / 0.014526 (0.084136) | 0.114971 / 0.176557 (-0.061586) | 0.186815 / 0.737135 (-0.550321) | 0.115490 / 0.296338 (-0.180848) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.422590 / 0.215209 (0.207381) | 4.277189 / 2.077655 (2.199535) | 2.095565 / 1.504120 (0.591445) | 2.040825 / 1.541195 (0.499630) | 2.162562 / 1.468490 (0.694072) | 0.578602 / 4.584777 (-4.006175) | 4.203474 / 3.745712 (0.457762) | 6.674595 / 5.269862 (1.404734) | 3.913251 / 4.565676 (-0.652426) | 0.067777 / 0.424275 (-0.356498) | 0.008716 / 0.007607 (0.001109) | 0.548704 / 0.226044 (0.322660) | 5.162120 / 2.268929 (2.893192) | 2.600250 / 55.444624 (-52.844374) | 2.232730 / 6.876477 (-4.643747) | 2.485617 / 2.142072 (0.343544) | 0.650872 / 4.805227 (-4.154355) | 0.148022 / 6.500664 (-6.352642) | 0.064795 / 0.075469 (-0.010674) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.399439 / 1.841788 (-0.442349) | 22.438959 / 8.074308 (14.364651) | 16.447831 / 10.191392 (6.256439) | 0.202003 / 0.680424 (-0.478421) | 0.026200 / 0.534201 (-0.508001) | 0.472966 / 0.579283 (-0.106317) | 0.491621 / 0.434364 (0.057257) | 0.551580 / 0.540337 (0.011242) | 0.751420 / 1.386936 (-0.635516) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007241 / 0.011353 (-0.004112) | 0.004434 / 0.011008 (-0.006574) | 0.075872 / 0.038508 (0.037364) | 0.080094 / 0.023109 (0.056985) | 0.459244 / 0.275898 (0.183346) | 0.492482 / 0.323480 (0.169002) | 0.005791 / 0.007986 (-0.002194) | 0.003657 / 0.004328 (-0.000671) | 0.075214 / 0.004250 (0.070964) | 0.064208 / 0.037052 (0.027156) | 0.464195 / 0.258489 (0.205706) | 0.497809 / 0.293841 (0.203968) | 0.036301 / 0.128546 (-0.092245) | 0.009855 / 0.075646 (-0.065791) | 0.080826 / 0.419271 (-0.338445) | 0.056700 / 0.043533 (0.013167) | 0.452850 / 0.255139 (0.197711) | 0.490738 / 0.283200 (0.207538) | 0.024145 / 0.141683 (-0.117538) | 1.689911 / 1.452155 (0.237757) | 1.789803 / 1.492716 (0.297087) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.247741 / 0.018006 (0.229735) | 0.486769 / 0.000490 (0.486279) | 0.000418 / 0.000200 (0.000218) | 0.000060 / 0.000054 (0.000005) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.036317 / 0.037411 (-0.001094) | 0.104943 / 0.014526 (0.090417) | 0.120972 / 0.176557 (-0.055585) | 0.188461 / 0.737135 (-0.548674) | 0.120926 / 0.296338 (-0.175412) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.465788 / 0.215209 (0.250579) | 4.662369 / 2.077655 (2.584714) | 2.442241 / 1.504120 (0.938121) | 2.266328 / 1.541195 (0.725133) | 2.438998 / 1.468490 (0.970508) | 0.531384 / 4.584777 (-4.053393) | 4.125286 / 3.745712 (0.379574) | 3.920912 / 5.269862 (-1.348950) | 2.292149 / 4.565676 (-2.273528) | 0.070146 / 0.424275 (-0.354129) | 0.008887 / 0.007607 (0.001280) | 0.598181 / 0.226044 (0.372137) | 5.726454 / 2.268929 (3.457526) | 3.081836 / 55.444624 (-52.362788) | 2.683508 / 6.876477 (-4.192969) | 2.587350 / 2.142072 (0.445278) | 0.604736 / 4.805227 (-4.200491) | 0.141303 / 6.500664 (-6.359362) | 0.065020 / 0.075469 (-0.010449) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.481850 / 1.841788 (-0.359938) | 22.259592 / 8.074308 (14.185284) | 16.304290 / 10.191392 (6.112898) | 0.173514 / 0.680424 (-0.506909) | 0.021590 / 0.534201 (-0.512611) | 0.471753 / 0.579283 (-0.107531) | 0.472132 / 0.434364 (0.037768) | 0.563344 / 0.540337 (0.023007) | 0.738509 / 1.386936 (-0.648427) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#1cb7ae56dbd814945a4982c63bf0e50859a7b93a \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.005910 / 0.011353 (-0.005443) | 0.004372 / 0.011008 (-0.006636) | 0.081583 / 0.038508 (0.043075) | 0.069598 / 0.023109 (0.046488) | 0.346360 / 0.275898 (0.070462) | 0.360733 / 0.323480 (0.037254) | 0.004725 / 0.007986 (-0.003261) | 0.003106 / 0.004328 (-0.001222) | 0.059916 / 0.004250 (0.055666) | 0.053242 / 0.037052 (0.016189) | 0.353551 / 0.258489 (0.095062) | 0.373052 / 0.293841 (0.079211) | 0.029036 / 0.128546 (-0.099510) | 0.007894 / 0.075646 (-0.067753) | 0.284131 / 0.419271 (-0.135140) | 0.049348 / 0.043533 (0.005815) | 0.347409 / 0.255139 (0.092270) | 0.355029 / 0.283200 (0.071830) | 0.022511 / 0.141683 (-0.119171) | 1.454495 / 1.452155 (0.002340) | 1.439551 / 1.492716 (-0.053166) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.218889 / 0.018006 (0.200883) | 0.478734 / 0.000490 (0.478244) | 0.003758 / 0.000200 (0.003558) | 0.000083 / 0.000054 (0.000029) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025759 / 0.037411 (-0.011653) | 0.082511 / 0.014526 (0.067985) | 0.087578 / 0.176557 (-0.088979) | 0.137760 / 0.737135 (-0.599375) | 0.093312 / 0.296338 (-0.203027) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.378963 / 0.215209 (0.163754) | 3.645846 / 2.077655 (1.568191) | 1.741135 / 1.504120 (0.237015) | 1.599166 / 1.541195 (0.057972) | 1.610817 / 1.468490 (0.142327) | 0.459209 / 4.584777 (-4.125568) | 3.484857 / 3.745712 (-0.260855) | 3.928109 / 5.269862 (-1.341752) | 2.419784 / 4.565676 (-2.145892) | 0.051987 / 0.424275 (-0.372288) | 0.006495 / 0.007607 (-0.001112) | 0.427311 / 0.226044 (0.201267) | 4.226378 / 2.268929 (1.957450) | 2.212331 / 55.444624 (-53.232293) | 1.916213 / 6.876477 (-4.960264) | 1.978809 / 2.142072 (-0.163263) | 0.547351 / 4.805227 (-4.257876) | 0.121110 / 6.500664 (-6.379554) | 0.054163 / 0.075469 (-0.021306) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.228594 / 1.841788 (-0.613193) | 19.410901 / 8.074308 (11.336593) | 13.014722 / 10.191392 (2.823330) | 0.156449 / 0.680424 (-0.523975) | 0.021032 / 0.534201 (-0.513169) | 0.403976 / 0.579283 (-0.175307) | 0.413885 / 0.434364 (-0.020479) | 0.470465 / 0.540337 (-0.069873) | 0.641322 / 1.386936 (-0.745614) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007210 / 0.011353 (-0.004143) | 0.003824 / 0.011008 (-0.007185) | 0.058227 / 0.038508 (0.019719) | 0.076211 / 0.023109 (0.053102) | 0.336626 / 0.275898 (0.060728) | 0.420542 / 0.323480 (0.097062) | 0.006178 / 0.007986 (-0.001808) | 0.003332 / 0.004328 (-0.000997) | 0.058073 / 0.004250 (0.053823) | 0.062485 / 0.037052 (0.025432) | 0.386175 / 0.258489 (0.127686) | 0.415659 / 0.293841 (0.121818) | 0.031264 / 0.128546 (-0.097282) | 0.007502 / 0.075646 (-0.068144) | 0.072079 / 0.419271 (-0.347192) | 0.055860 / 0.043533 (0.012327) | 0.343508 / 0.255139 (0.088369) | 0.437844 / 0.283200 (0.154645) | 0.032852 / 0.141683 (-0.108831) | 1.409241 / 1.452155 (-0.042913) | 1.623949 / 1.492716 (0.131233) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.207511 / 0.018006 (0.189504) | 0.464149 / 0.000490 (0.463660) | 0.003248 / 0.000200 (0.003048) | 0.000226 / 0.000054 (0.000172) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030767 / 0.037411 (-0.006645) | 0.079169 / 0.014526 (0.064643) | 0.093111 / 0.176557 (-0.083445) | 0.153369 / 0.737135 (-0.583767) | 0.092939 / 0.296338 (-0.203400) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.375602 / 0.215209 (0.160392) | 3.968612 / 2.077655 (1.890957) | 2.081749 / 1.504120 (0.577629) | 1.899772 / 1.541195 (0.358577) | 1.847923 / 1.468490 (0.379433) | 0.442867 / 4.584777 (-4.141910) | 3.646664 / 3.745712 (-0.099048) | 5.870600 / 5.269862 (0.600739) | 3.356698 / 4.565676 (-1.208979) | 0.051422 / 0.424275 (-0.372853) | 0.006006 / 0.007607 (-0.001601) | 0.442439 / 0.226044 (0.216395) | 4.466256 / 2.268929 (2.197328) | 2.483832 / 55.444624 (-52.960792) | 2.105612 / 6.876477 (-4.770865) | 2.060650 / 2.142072 (-0.081422) | 0.531119 / 4.805227 (-4.274108) | 0.123436 / 6.500664 (-6.377228) | 0.059838 / 0.075469 (-0.015632) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.283042 / 1.841788 (-0.558746) | 19.688251 / 8.074308 (11.613943) | 13.346386 / 10.191392 (3.154994) | 0.197463 / 0.680424 (-0.482961) | 0.018484 / 0.534201 (-0.515717) | 0.391727 / 0.579283 (-0.187556) | 0.425061 / 0.434364 (-0.009303) | 0.448177 / 0.540337 (-0.092160) | 0.653694 / 1.386936 (-0.733242) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#01604752fe89d290479fa406b1a24ac1f346826e \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008966 / 0.011353 (-0.002387) | 0.005195 / 0.011008 (-0.005813) | 0.102879 / 0.038508 (0.064371) | 0.090902 / 0.023109 (0.067792) | 0.434397 / 0.275898 (0.158498) | 0.454013 / 0.323480 (0.130534) | 0.008507 / 0.007986 (0.000521) | 0.005000 / 0.004328 (0.000671) | 0.075789 / 0.004250 (0.071538) | 0.067608 / 0.037052 (0.030555) | 0.435091 / 0.258489 (0.176602) | 0.469411 / 0.293841 (0.175570) | 0.050859 / 0.128546 (-0.077687) | 0.013560 / 0.075646 (-0.062086) | 0.345473 / 0.419271 (-0.073799) | 0.094974 / 0.043533 (0.051441) | 0.429626 / 0.255139 (0.174487) | 0.434290 / 0.283200 (0.151090) | 0.052269 / 0.141683 (-0.089413) | 1.700549 / 1.452155 (0.248395) | 1.890693 / 1.492716 (0.397976) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.296618 / 0.018006 (0.278612) | 0.613908 / 0.000490 (0.613419) | 0.000484 / 0.000200 (0.000284) | 0.000086 / 0.000054 (0.000032) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.034346 / 0.037411 (-0.003065) | 0.096836 / 0.014526 (0.082310) | 0.113332 / 0.176557 (-0.063224) | 0.194464 / 0.737135 (-0.542671) | 0.111732 / 0.296338 (-0.184606) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.624954 / 0.215209 (0.409745) | 6.442193 / 2.077655 (4.364538) | 2.818331 / 1.504120 (1.314211) | 2.529607 / 1.541195 (0.988413) | 2.549026 / 1.468490 (1.080536) | 0.967367 / 4.584777 (-3.617410) | 5.446885 / 3.745712 (1.701173) | 6.259099 / 5.269862 (0.989237) | 3.652936 / 4.565676 (-0.912740) | 0.106420 / 0.424275 (-0.317855) | 0.011293 / 0.007607 (0.003686) | 0.772026 / 0.226044 (0.545982) | 7.823986 / 2.268929 (5.555057) | 3.725328 / 55.444624 (-51.719297) | 2.851489 / 6.876477 (-4.024988) | 3.013722 / 2.142072 (0.871649) | 1.045090 / 4.805227 (-3.760137) | 0.213174 / 6.500664 (-6.287490) | 0.077104 / 0.075469 (0.001635) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.657135 / 1.841788 (-0.184652) | 24.547604 / 8.074308 (16.473296) | 19.989533 / 10.191392 (9.798141) | 0.257139 / 0.680424 (-0.423285) | 0.028448 / 0.534201 (-0.505753) | 0.490801 / 0.579283 (-0.088482) | 0.628072 / 0.434364 (0.193708) | 0.584873 / 0.540337 (0.044536) | 0.825258 / 1.386936 (-0.561678) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009258 / 0.011353 (-0.002095) | 0.005660 / 0.011008 (-0.005348) | 0.080577 / 0.038508 (0.042069) | 0.095786 / 0.023109 (0.072676) | 0.473334 / 0.275898 (0.197436) | 0.527962 / 0.323480 (0.204482) | 0.006537 / 0.007986 (-0.001449) | 0.004411 / 0.004328 (0.000083) | 0.080702 / 0.004250 (0.076452) | 0.077020 / 0.037052 (0.039968) | 0.483205 / 0.258489 (0.224716) | 0.556916 / 0.293841 (0.263076) | 0.047670 / 0.128546 (-0.080877) | 0.016647 / 0.075646 (-0.058999) | 0.090653 / 0.419271 (-0.328619) | 0.062122 / 0.043533 (0.018589) | 0.498326 / 0.255139 (0.243187) | 0.546572 / 0.283200 (0.263372) | 0.037525 / 0.141683 (-0.104157) | 1.869520 / 1.452155 (0.417365) | 1.915335 / 1.492716 (0.422619) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.248287 / 0.018006 (0.230281) | 0.611440 / 0.000490 (0.610950) | 0.004102 / 0.000200 (0.003902) | 0.000132 / 0.000054 (0.000078) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.038228 / 0.037411 (0.000817) | 0.103510 / 0.014526 (0.088984) | 0.114337 / 0.176557 (-0.062219) | 0.189662 / 0.737135 (-0.547473) | 0.119078 / 0.296338 (-0.177260) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.606622 / 0.215209 (0.391413) | 6.053900 / 2.077655 (3.976246) | 2.857972 / 1.504120 (1.353852) | 2.549756 / 1.541195 (1.008561) | 2.584557 / 1.468490 (1.116067) | 0.930431 / 4.584777 (-3.654346) | 5.524077 / 3.745712 (1.778365) | 7.858406 / 5.269862 (2.588545) | 4.890697 / 4.565676 (0.325020) | 0.095356 / 0.424275 (-0.328919) | 0.008614 / 0.007607 (0.001007) | 0.774227 / 0.226044 (0.548182) | 7.470215 / 2.268929 (5.201287) | 3.784820 / 55.444624 (-51.659805) | 3.199364 / 6.876477 (-3.677113) | 3.212002 / 2.142072 (1.069929) | 1.054104 / 4.805227 (-3.751123) | 0.226044 / 6.500664 (-6.274620) | 0.092237 / 0.075469 (0.016768) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.801054 / 1.841788 (-0.040734) | 24.220404 / 8.074308 (16.146096) | 21.652936 / 10.191392 (11.461544) | 0.247004 / 0.680424 (-0.433420) | 0.029651 / 0.534201 (-0.504550) | 0.475702 / 0.579283 (-0.103581) | 0.621121 / 0.434364 (0.186757) | 0.570489 / 0.540337 (0.030151) | 0.768840 / 1.386936 (-0.618096) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#b2fc21eda345643fb57d1d1167ebed9043310911 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009223 / 0.011353 (-0.002130) | 0.005750 / 0.011008 (-0.005258) | 0.105264 / 0.038508 (0.066756) | 0.088478 / 0.023109 (0.065369) | 0.461119 / 0.275898 (0.185221) | 0.481115 / 0.323480 (0.157636) | 0.006366 / 0.007986 (-0.001619) | 0.004515 / 0.004328 (0.000186) | 0.079296 / 0.004250 (0.075045) | 0.063483 / 0.037052 (0.026430) | 0.444490 / 0.258489 (0.186001) | 0.496474 / 0.293841 (0.202634) | 0.048568 / 0.128546 (-0.079978) | 0.013574 / 0.075646 (-0.062073) | 0.379213 / 0.419271 (-0.040059) | 0.086464 / 0.043533 (0.042932) | 0.437526 / 0.255139 (0.182387) | 0.447117 / 0.283200 (0.163917) | 0.049502 / 0.141683 (-0.092180) | 1.749146 / 1.452155 (0.296992) | 1.831082 / 1.492716 (0.338365) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.268205 / 0.018006 (0.250199) | 0.627406 / 0.000490 (0.626917) | 0.005439 / 0.000200 (0.005239) | 0.000128 / 0.000054 (0.000074) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030564 / 0.037411 (-0.006848) | 0.096365 / 0.014526 (0.081840) | 0.117484 / 0.176557 (-0.059072) | 0.189104 / 0.737135 (-0.548032) | 0.118073 / 0.296338 (-0.178266) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.618229 / 0.215209 (0.403019) | 6.437853 / 2.077655 (4.360199) | 2.789946 / 1.504120 (1.285826) | 2.339245 / 1.541195 (0.798050) | 2.588779 / 1.468490 (1.120289) | 0.921008 / 4.584777 (-3.663769) | 5.402940 / 3.745712 (1.657227) | 4.818783 / 5.269862 (-0.451078) | 3.162259 / 4.565676 (-1.403417) | 0.108501 / 0.424275 (-0.315774) | 0.009384 / 0.007607 (0.001777) | 0.766811 / 0.226044 (0.540766) | 7.624629 / 2.268929 (5.355701) | 3.442420 / 55.444624 (-52.002204) | 2.759967 / 6.876477 (-4.116510) | 3.049644 / 2.142072 (0.907572) | 1.113308 / 4.805227 (-3.691919) | 0.223923 / 6.500664 (-6.276741) | 0.079156 / 0.075469 (0.003687) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.683318 / 1.841788 (-0.158470) | 25.062141 / 8.074308 (16.987833) | 21.777131 / 10.191392 (11.585739) | 0.266939 / 0.680424 (-0.413485) | 0.029670 / 0.534201 (-0.504531) | 0.476761 / 0.579283 (-0.102522) | 0.622080 / 0.434364 (0.187716) | 0.601781 / 0.540337 (0.061443) | 0.785126 / 1.386936 (-0.601811) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.010198 / 0.011353 (-0.001155) | 0.005777 / 0.011008 (-0.005231) | 0.083003 / 0.038508 (0.044495) | 0.093411 / 0.023109 (0.070302) | 0.496178 / 0.275898 (0.220280) | 0.554670 / 0.323480 (0.231190) | 0.008351 / 0.007986 (0.000365) | 0.004678 / 0.004328 (0.000350) | 0.083631 / 0.004250 (0.079381) | 0.075538 / 0.037052 (0.038485) | 0.492410 / 0.258489 (0.233921) | 0.545209 / 0.293841 (0.251368) | 0.048365 / 0.128546 (-0.080181) | 0.014219 / 0.075646 (-0.061427) | 0.100749 / 0.419271 (-0.318523) | 0.063431 / 0.043533 (0.019898) | 0.511115 / 0.255139 (0.255976) | 0.532965 / 0.283200 (0.249765) | 0.037968 / 0.141683 (-0.103715) | 1.940268 / 1.452155 (0.488113) | 2.032934 / 1.492716 (0.540217) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.238179 / 0.018006 (0.220172) | 0.605767 / 0.000490 (0.605277) | 0.004033 / 0.000200 (0.003833) | 0.000125 / 0.000054 (0.000071) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.036436 / 0.037411 (-0.000975) | 0.108034 / 0.014526 (0.093509) | 0.118624 / 0.176557 (-0.057933) | 0.183079 / 0.737135 (-0.554056) | 0.121739 / 0.296338 (-0.174600) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.630538 / 0.215209 (0.415329) | 6.552184 / 2.077655 (4.474529) | 3.003412 / 1.504120 (1.499292) | 2.669026 / 1.541195 (1.127832) | 2.791109 / 1.468490 (1.322619) | 0.884003 / 4.584777 (-3.700774) | 5.538660 / 3.745712 (1.792947) | 5.126708 / 5.269862 (-0.143154) | 3.120825 / 4.565676 (-1.444852) | 0.101178 / 0.424275 (-0.323097) | 0.009027 / 0.007607 (0.001420) | 0.785914 / 0.226044 (0.559869) | 7.994720 / 2.268929 (5.725792) | 4.061996 / 55.444624 (-51.382629) | 3.263230 / 6.876477 (-3.613247) | 3.288622 / 2.142072 (1.146550) | 1.141867 / 4.805227 (-3.663360) | 0.255287 / 6.500664 (-6.245378) | 0.100637 / 0.075469 (0.025168) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.769821 / 1.841788 (-0.071967) | 24.994008 / 8.074308 (16.919700) | 21.765971 / 10.191392 (11.574579) | 0.268493 / 0.680424 (-0.411931) | 0.028047 / 0.534201 (-0.506154) | 0.489472 / 0.579283 (-0.089811) | 0.594809 / 0.434364 (0.160445) | 0.613578 / 0.540337 (0.073241) | 0.879360 / 1.386936 (-0.507576) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#b85b1154aef2a9ab4d558f60d91623f2cc1583c4 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006003 / 0.011353 (-0.005350) | 0.003590 / 0.011008 (-0.007418) | 0.084657 / 0.038508 (0.046149) | 0.057884 / 0.023109 (0.034775) | 0.318347 / 0.275898 (0.042449) | 0.345976 / 0.323480 (0.022496) | 0.004706 / 0.007986 (-0.003279) | 0.002921 / 0.004328 (-0.001407) | 0.061850 / 0.004250 (0.057600) | 0.050558 / 0.037052 (0.013505) | 0.320877 / 0.258489 (0.062388) | 0.356062 / 0.293841 (0.062222) | 0.027511 / 0.128546 (-0.101035) | 0.007954 / 0.075646 (-0.067693) | 0.260290 / 0.419271 (-0.158981) | 0.051207 / 0.043533 (0.007674) | 0.334423 / 0.255139 (0.079284) | 0.338575 / 0.283200 (0.055375) | 0.022330 / 0.141683 (-0.119353) | 1.445446 / 1.452155 (-0.006709) | 1.500626 / 1.492716 (0.007910) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.192440 / 0.018006 (0.174433) | 0.428455 / 0.000490 (0.427965) | 0.000318 / 0.000200 (0.000118) | 0.000056 / 0.000054 (0.000002) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.022933 / 0.037411 (-0.014478) | 0.072795 / 0.014526 (0.058269) | 0.081149 / 0.176557 (-0.095407) | 0.142941 / 0.737135 (-0.594195) | 0.082410 / 0.296338 (-0.213928) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.405220 / 0.215209 (0.190011) | 4.048585 / 2.077655 (1.970931) | 2.027908 / 1.504120 (0.523788) | 1.887828 / 1.541195 (0.346633) | 2.131780 / 1.468490 (0.663290) | 0.502847 / 4.584777 (-4.081930) | 3.069498 / 3.745712 (-0.676215) | 4.094774 / 5.269862 (-1.175088) | 2.544004 / 4.565676 (-2.021673) | 0.059540 / 0.424275 (-0.364735) | 0.006501 / 0.007607 (-0.001106) | 0.477218 / 0.226044 (0.251173) | 4.764961 / 2.268929 (2.496032) | 2.434594 / 55.444624 (-53.010030) | 2.104833 / 6.876477 (-4.771644) | 2.263059 / 2.142072 (0.120987) | 0.591755 / 4.805227 (-4.213472) | 0.131167 / 6.500664 (-6.369497) | 0.061808 / 0.075469 (-0.013661) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.345364 / 1.841788 (-0.496424) | 18.122584 / 8.074308 (10.048276) | 13.318689 / 10.191392 (3.127297) | 0.144526 / 0.680424 (-0.535898) | 0.016997 / 0.534201 (-0.517204) | 0.336036 / 0.579283 (-0.243247) | 0.359532 / 0.434364 (-0.074832) | 0.386945 / 0.540337 (-0.153392) | 0.538659 / 1.386936 (-0.848277) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006088 / 0.011353 (-0.005265) | 0.003684 / 0.011008 (-0.007324) | 0.062340 / 0.038508 (0.023832) | 0.058461 / 0.023109 (0.035352) | 0.360134 / 0.275898 (0.084236) | 0.393298 / 0.323480 (0.069818) | 0.004664 / 0.007986 (-0.003322) | 0.002909 / 0.004328 (-0.001420) | 0.062668 / 0.004250 (0.058418) | 0.050145 / 0.037052 (0.013092) | 0.361897 / 0.258489 (0.103408) | 0.402008 / 0.293841 (0.108167) | 0.027491 / 0.128546 (-0.101055) | 0.008113 / 0.075646 (-0.067534) | 0.068114 / 0.419271 (-0.351157) | 0.043303 / 0.043533 (-0.000230) | 0.360569 / 0.255139 (0.105430) | 0.387144 / 0.283200 (0.103944) | 0.020194 / 0.141683 (-0.121489) | 1.418066 / 1.452155 (-0.034089) | 1.475640 / 1.492716 (-0.017076) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.200291 / 0.018006 (0.182285) | 0.432298 / 0.000490 (0.431809) | 0.003303 / 0.000200 (0.003103) | 0.000075 / 0.000054 (0.000020) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027749 / 0.037411 (-0.009662) | 0.081890 / 0.014526 (0.067364) | 0.094319 / 0.176557 (-0.082238) | 0.148646 / 0.737135 (-0.588490) | 0.091830 / 0.296338 (-0.204509) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.433546 / 0.215209 (0.218337) | 4.326855 / 2.077655 (2.249200) | 2.230186 / 1.504120 (0.726066) | 2.052524 / 1.541195 (0.511329) | 2.117270 / 1.468490 (0.648779) | 0.500331 / 4.584777 (-4.084446) | 3.113662 / 3.745712 (-0.632050) | 2.931540 / 5.269862 (-2.338322) | 1.853615 / 4.565676 (-2.712062) | 0.058250 / 0.424275 (-0.366025) | 0.006546 / 0.007607 (-0.001061) | 0.508850 / 0.226044 (0.282806) | 5.081809 / 2.268929 (2.812880) | 2.687037 / 55.444624 (-52.757588) | 2.369317 / 6.876477 (-4.507160) | 2.383549 / 2.142072 (0.241477) | 0.587039 / 4.805227 (-4.218188) | 0.125858 / 6.500664 (-6.374806) | 0.062522 / 0.075469 (-0.012947) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.294929 / 1.841788 (-0.546858) | 18.056312 / 8.074308 (9.982004) | 13.755117 / 10.191392 (3.563725) | 0.132037 / 0.680424 (-0.548387) | 0.016866 / 0.534201 (-0.517335) | 0.339040 / 0.579283 (-0.240243) | 0.364371 / 0.434364 (-0.069993) | 0.399533 / 0.540337 (-0.140804) | 0.564524 / 1.386936 (-0.822412) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#64b811c13a7982015d7e078e3d693ce5359a05a2 \"CML watermark\")\n", "@lhoestq This bar comes from: https://github.com/huggingface/datasets/blob/b8067c0262073891180869f700ebef5ac3dc5cce/src/datasets/builder.py#L1156-L1166\r\n\r\nDo you prefer not showing it or, e.g., having `desc=\"Generating splits\"`?", "No strong opinion. Since there is a \"Generating\" progress bar already, maybe it can be \"Preparing splits\" (ref to download_and_prepare)", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006348 / 0.011353 (-0.005005) | 0.003721 / 0.011008 (-0.007287) | 0.084039 / 0.038508 (0.045531) | 0.067627 / 0.023109 (0.044517) | 0.308372 / 0.275898 (0.032474) | 0.335131 / 0.323480 (0.011652) | 0.005157 / 0.007986 (-0.002829) | 0.003266 / 0.004328 (-0.001062) | 0.065374 / 0.004250 (0.061124) | 0.055550 / 0.037052 (0.018498) | 0.314001 / 0.258489 (0.055512) | 0.350510 / 0.293841 (0.056669) | 0.030859 / 0.128546 (-0.097688) | 0.008286 / 0.075646 (-0.067361) | 0.287122 / 0.419271 (-0.132149) | 0.051494 / 0.043533 (0.007961) | 0.309868 / 0.255139 (0.054729) | 0.325845 / 0.283200 (0.042645) | 0.022622 / 0.141683 (-0.119061) | 1.468730 / 1.452155 (0.016575) | 1.547871 / 1.492716 (0.055155) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.202763 / 0.018006 (0.184757) | 0.456403 / 0.000490 (0.455914) | 0.003116 / 0.000200 (0.002916) | 0.000079 / 0.000054 (0.000024) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027297 / 0.037411 (-0.010114) | 0.081204 / 0.014526 (0.066678) | 0.094274 / 0.176557 (-0.082282) | 0.154391 / 0.737135 (-0.582744) | 0.094312 / 0.296338 (-0.202026) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.387382 / 0.215209 (0.172173) | 3.865597 / 2.077655 (1.787943) | 1.855959 / 1.504120 (0.351839) | 1.685411 / 1.541195 (0.144216) | 1.732127 / 1.468490 (0.263637) | 0.482230 / 4.584777 (-4.102547) | 3.664947 / 3.745712 (-0.080765) | 5.114379 / 5.269862 (-0.155482) | 3.102803 / 4.565676 (-1.462873) | 0.056509 / 0.424275 (-0.367766) | 0.007230 / 0.007607 (-0.000377) | 0.456788 / 0.226044 (0.230744) | 4.575831 / 2.268929 (2.306902) | 2.335249 / 55.444624 (-53.109375) | 2.003805 / 6.876477 (-4.872672) | 2.141788 / 2.142072 (-0.000285) | 0.577501 / 4.805227 (-4.227726) | 0.130264 / 6.500664 (-6.370400) | 0.058889 / 0.075469 (-0.016580) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.252673 / 1.841788 (-0.589115) | 18.676897 / 8.074308 (10.602589) | 13.988101 / 10.191392 (3.796709) | 0.151376 / 0.680424 (-0.529048) | 0.018104 / 0.534201 (-0.516097) | 0.388413 / 0.579283 (-0.190870) | 0.414841 / 0.434364 (-0.019523) | 0.456078 / 0.540337 (-0.084259) | 0.641715 / 1.386936 (-0.745221) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006315 / 0.011353 (-0.005038) | 0.003847 / 0.011008 (-0.007162) | 0.063989 / 0.038508 (0.025481) | 0.068244 / 0.023109 (0.045135) | 0.416201 / 0.275898 (0.140303) | 0.438446 / 0.323480 (0.114966) | 0.005820 / 0.007986 (-0.002166) | 0.003165 / 0.004328 (-0.001163) | 0.064143 / 0.004250 (0.059892) | 0.056529 / 0.037052 (0.019477) | 0.414916 / 0.258489 (0.156427) | 0.450771 / 0.293841 (0.156930) | 0.030611 / 0.128546 (-0.097935) | 0.008289 / 0.075646 (-0.067357) | 0.070725 / 0.419271 (-0.348546) | 0.047998 / 0.043533 (0.004465) | 0.405609 / 0.255139 (0.150470) | 0.421895 / 0.283200 (0.138696) | 0.022135 / 0.141683 (-0.119548) | 1.444238 / 1.452155 (-0.007916) | 1.515823 / 1.492716 (0.023107) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.227043 / 0.018006 (0.209037) | 0.439732 / 0.000490 (0.439242) | 0.001267 / 0.000200 (0.001067) | 0.000070 / 0.000054 (0.000016) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029082 / 0.037411 (-0.008329) | 0.086201 / 0.014526 (0.071675) | 0.098653 / 0.176557 (-0.077903) | 0.152574 / 0.737135 (-0.584561) | 0.100696 / 0.296338 (-0.195642) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.411243 / 0.215209 (0.196034) | 4.100170 / 2.077655 (2.022515) | 2.118310 / 1.504120 (0.614190) | 1.935646 / 1.541195 (0.394451) | 1.970798 / 1.468490 (0.502307) | 0.478635 / 4.584777 (-4.106142) | 3.589396 / 3.745712 (-0.156316) | 3.312462 / 5.269862 (-1.957399) | 1.963081 / 4.565676 (-2.602595) | 0.056392 / 0.424275 (-0.367883) | 0.007134 / 0.007607 (-0.000473) | 0.485131 / 0.226044 (0.259086) | 4.838946 / 2.268929 (2.570017) | 2.624550 / 55.444624 (-52.820075) | 2.223046 / 6.876477 (-4.653431) | 2.230642 / 2.142072 (0.088570) | 0.594892 / 4.805227 (-4.210335) | 0.130523 / 6.500664 (-6.370141) | 0.059585 / 0.075469 (-0.015884) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.329941 / 1.841788 (-0.511847) | 19.199057 / 8.074308 (11.124748) | 14.166009 / 10.191392 (3.974617) | 0.190595 / 0.680424 (-0.489829) | 0.018419 / 0.534201 (-0.515782) | 0.392031 / 0.579283 (-0.187252) | 0.409395 / 0.434364 (-0.024969) | 0.475930 / 0.540337 (-0.064408) | 0.654412 / 1.386936 (-0.732524) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#42fdfbd567674d075c3a9148ec3c95221eb62cfe \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.007500 / 0.011353 (-0.003853) | 0.004328 / 0.011008 (-0.006681) | 0.086718 / 0.038508 (0.048209) | 0.098638 / 0.023109 (0.075529) | 0.335308 / 0.275898 (0.059409) | 0.369163 / 0.323480 (0.045683) | 0.005733 / 0.007986 (-0.002253) | 0.003738 / 0.004328 (-0.000590) | 0.066452 / 0.004250 (0.062202) | 0.066245 / 0.037052 (0.029192) | 0.337609 / 0.258489 (0.079120) | 0.388584 / 0.293841 (0.094744) | 0.031742 / 0.128546 (-0.096804) | 0.008721 / 0.075646 (-0.066925) | 0.290820 / 0.419271 (-0.128452) | 0.053323 / 0.043533 (0.009790) | 0.329192 / 0.255139 (0.074053) | 0.350560 / 0.283200 (0.067360) | 0.025402 / 0.141683 (-0.116281) | 1.476174 / 1.452155 (0.024020) | 1.578194 / 1.492716 (0.085478) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.256160 / 0.018006 (0.238154) | 0.560315 / 0.000490 (0.559825) | 0.005287 / 0.000200 (0.005088) | 0.000094 / 0.000054 (0.000040) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029164 / 0.037411 (-0.008247) | 0.084881 / 0.014526 (0.070356) | 0.100979 / 0.176557 (-0.075577) | 0.156539 / 0.737135 (-0.580597) | 0.101510 / 0.296338 (-0.194828) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.381138 / 0.215209 (0.165929) | 3.791573 / 2.077655 (1.713918) | 1.841954 / 1.504120 (0.337834) | 1.672463 / 1.541195 (0.131268) | 1.785769 / 1.468490 (0.317279) | 0.483263 / 4.584777 (-4.101514) | 3.617391 / 3.745712 (-0.128322) | 5.607794 / 5.269862 (0.337933) | 3.359530 / 4.565676 (-1.206147) | 0.056826 / 0.424275 (-0.367449) | 0.007375 / 0.007607 (-0.000232) | 0.455853 / 0.226044 (0.229809) | 4.548965 / 2.268929 (2.280037) | 2.412716 / 55.444624 (-53.031908) | 1.991456 / 6.876477 (-4.885021) | 2.242851 / 2.142072 (0.100778) | 0.573070 / 4.805227 (-4.232157) | 0.134658 / 6.500664 (-6.366006) | 0.061539 / 0.075469 (-0.013930) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.278306 / 1.841788 (-0.563481) | 20.634317 / 8.074308 (12.560009) | 15.164246 / 10.191392 (4.972854) | 0.167487 / 0.680424 (-0.512937) | 0.019006 / 0.534201 (-0.515195) | 0.394617 / 0.579283 (-0.184666) | 0.423385 / 0.434364 (-0.010979) | 0.469968 / 0.540337 (-0.070370) | 0.630058 / 1.386936 (-0.756878) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006793 / 0.011353 (-0.004559) | 0.004260 / 0.011008 (-0.006748) | 0.065398 / 0.038508 (0.026890) | 0.077850 / 0.023109 (0.054741) | 0.371754 / 0.275898 (0.095855) | 0.400652 / 0.323480 (0.077172) | 0.005729 / 0.007986 (-0.002256) | 0.003660 / 0.004328 (-0.000669) | 0.065119 / 0.004250 (0.060869) | 0.060714 / 0.037052 (0.023661) | 0.384592 / 0.258489 (0.126103) | 0.412806 / 0.293841 (0.118965) | 0.031865 / 0.128546 (-0.096681) | 0.008807 / 0.075646 (-0.066839) | 0.071156 / 0.419271 (-0.348115) | 0.049571 / 0.043533 (0.006038) | 0.367381 / 0.255139 (0.112242) | 0.386713 / 0.283200 (0.103513) | 0.024838 / 0.141683 (-0.116845) | 1.492986 / 1.452155 (0.040831) | 1.559243 / 1.492716 (0.066526) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.269737 / 0.018006 (0.251730) | 0.565177 / 0.000490 (0.564687) | 0.000404 / 0.000200 (0.000204) | 0.000060 / 0.000054 (0.000006) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.031631 / 0.037411 (-0.005780) | 0.087289 / 0.014526 (0.072764) | 0.102798 / 0.176557 (-0.073759) | 0.158977 / 0.737135 (-0.578158) | 0.105495 / 0.296338 (-0.190843) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.425067 / 0.215209 (0.209858) | 4.243121 / 2.077655 (2.165466) | 2.234567 / 1.504120 (0.730447) | 2.070810 / 1.541195 (0.529615) | 2.176802 / 1.468490 (0.708312) | 0.484987 / 4.584777 (-4.099790) | 3.647000 / 3.745712 (-0.098712) | 3.574843 / 5.269862 (-1.695019) | 2.092581 / 4.565676 (-2.473095) | 0.057299 / 0.424275 (-0.366976) | 0.007480 / 0.007607 (-0.000128) | 0.507838 / 0.226044 (0.281794) | 5.076594 / 2.268929 (2.807666) | 2.718858 / 55.444624 (-52.725766) | 2.362793 / 6.876477 (-4.513684) | 2.451962 / 2.142072 (0.309890) | 0.581355 / 4.805227 (-4.223872) | 0.133723 / 6.500664 (-6.366941) | 0.061896 / 0.075469 (-0.013573) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.325814 / 1.841788 (-0.515974) | 20.614502 / 8.074308 (12.540194) | 14.769422 / 10.191392 (4.578029) | 0.193797 / 0.680424 (-0.486627) | 0.018379 / 0.534201 (-0.515822) | 0.394153 / 0.579283 (-0.185130) | 0.409585 / 0.434364 (-0.024779) | 0.479107 / 0.540337 (-0.061231) | 0.668397 / 1.386936 (-0.718539) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#b2d892237169bad5512c91cae453d257ebefc201 \"CML watermark\")\n", "In the end, I decided to remove the progress bar to avoid having it displayed when loading a cached dataset.", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==8.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006673 / 0.011353 (-0.004680) | 0.004162 / 0.011008 (-0.006846) | 0.084017 / 0.038508 (0.045509) | 0.079536 / 0.023109 (0.056426) | 0.313594 / 0.275898 (0.037695) | 0.349200 / 0.323480 (0.025720) | 0.005544 / 0.007986 (-0.002441) | 0.003472 / 0.004328 (-0.000857) | 0.064742 / 0.004250 (0.060491) | 0.056857 / 0.037052 (0.019805) | 0.318635 / 0.258489 (0.060146) | 0.354378 / 0.293841 (0.060537) | 0.030856 / 0.128546 (-0.097690) | 0.008759 / 0.075646 (-0.066887) | 0.287760 / 0.419271 (-0.131511) | 0.052307 / 0.043533 (0.008775) | 0.316396 / 0.255139 (0.061257) | 0.351408 / 0.283200 (0.068208) | 0.024914 / 0.141683 (-0.116769) | 1.484592 / 1.452155 (0.032437) | 1.560662 / 1.492716 (0.067945) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.280938 / 0.018006 (0.262932) | 0.580236 / 0.000490 (0.579747) | 0.003369 / 0.000200 (0.003169) | 0.000090 / 0.000054 (0.000036) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.028736 / 0.037411 (-0.008675) | 0.082916 / 0.014526 (0.068390) | 0.097761 / 0.176557 (-0.078796) | 0.153515 / 0.737135 (-0.583620) | 0.099282 / 0.296338 (-0.197057) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.401244 / 0.215209 (0.186035) | 4.019866 / 2.077655 (1.942211) | 2.029642 / 1.504120 (0.525522) | 1.849591 / 1.541195 (0.308396) | 1.946829 / 1.468490 (0.478339) | 0.479750 / 4.584777 (-4.105027) | 3.482822 / 3.745712 (-0.262890) | 3.955859 / 5.269862 (-1.314003) | 2.370747 / 4.565676 (-2.194930) | 0.056905 / 0.424275 (-0.367370) | 0.007319 / 0.007607 (-0.000288) | 0.485310 / 0.226044 (0.259266) | 4.858228 / 2.268929 (2.589299) | 2.500476 / 55.444624 (-52.944148) | 2.171156 / 6.876477 (-4.705320) | 2.427266 / 2.142072 (0.285194) | 0.570199 / 4.805227 (-4.235029) | 0.130855 / 6.500664 (-6.369809) | 0.060269 / 0.075469 (-0.015200) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.258044 / 1.841788 (-0.583743) | 20.218657 / 8.074308 (12.144349) | 13.597970 / 10.191392 (3.406578) | 0.167656 / 0.680424 (-0.512768) | 0.018137 / 0.534201 (-0.516064) | 0.395309 / 0.579283 (-0.183975) | 0.406325 / 0.434364 (-0.028039) | 0.467457 / 0.540337 (-0.072880) | 0.613636 / 1.386936 (-0.773300) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006846 / 0.011353 (-0.004507) | 0.004207 / 0.011008 (-0.006802) | 0.064525 / 0.038508 (0.026017) | 0.081329 / 0.023109 (0.058220) | 0.399838 / 0.275898 (0.123940) | 0.431305 / 0.323480 (0.107825) | 0.005859 / 0.007986 (-0.002127) | 0.003568 / 0.004328 (-0.000760) | 0.065262 / 0.004250 (0.061011) | 0.064796 / 0.037052 (0.027744) | 0.406858 / 0.258489 (0.148369) | 0.440971 / 0.293841 (0.147130) | 0.031421 / 0.128546 (-0.097125) | 0.008777 / 0.075646 (-0.066870) | 0.071418 / 0.419271 (-0.347853) | 0.049263 / 0.043533 (0.005730) | 0.384279 / 0.255139 (0.129140) | 0.410745 / 0.283200 (0.127546) | 0.024467 / 0.141683 (-0.117216) | 1.522379 / 1.452155 (0.070224) | 1.581636 / 1.492716 (0.088920) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.276161 / 0.018006 (0.258155) | 0.548842 / 0.000490 (0.548352) | 0.004523 / 0.000200 (0.004324) | 0.000098 / 0.000054 (0.000043) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030747 / 0.037411 (-0.006664) | 0.087493 / 0.014526 (0.072967) | 0.106563 / 0.176557 (-0.069993) | 0.162949 / 0.737135 (-0.574186) | 0.105303 / 0.296338 (-0.191036) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.425854 / 0.215209 (0.210645) | 4.244797 / 2.077655 (2.167142) | 2.269006 / 1.504120 (0.764886) | 2.097428 / 1.541195 (0.556234) | 2.181038 / 1.468490 (0.712548) | 0.477286 / 4.584777 (-4.107491) | 3.591452 / 3.745712 (-0.154260) | 3.481281 / 5.269862 (-1.788580) | 2.066895 / 4.565676 (-2.498782) | 0.056576 / 0.424275 (-0.367699) | 0.007409 / 0.007607 (-0.000199) | 0.498411 / 0.226044 (0.272367) | 4.994873 / 2.268929 (2.725945) | 2.749148 / 55.444624 (-52.695476) | 2.378544 / 6.876477 (-4.497932) | 2.452859 / 2.142072 (0.310786) | 0.571340 / 4.805227 (-4.233887) | 0.132174 / 6.500664 (-6.368490) | 0.061507 / 0.075469 (-0.013962) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.370773 / 1.841788 (-0.471015) | 20.493342 / 8.074308 (12.419034) | 14.809886 / 10.191392 (4.618494) | 0.175730 / 0.680424 (-0.504693) | 0.018617 / 0.534201 (-0.515583) | 0.393808 / 0.579283 (-0.185476) | 0.416419 / 0.434364 (-0.017945) | 0.477183 / 0.540337 (-0.063155) | 0.668060 / 1.386936 (-0.718876) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#2de7a2a4af5d94b0f98a7a6db94e78984af40602 \"CML watermark\")\n", "Nice one :)" ]

[]

[]

[]

[ "Hi !\r\n\r\nThis is actually supported ! but not yet in `from_pandas`.\r\nYou can use `from_dict` for now instead:\r\n```python\r\nfrom datasets import Dataset, Array2D, Features, Value\r\nimport pandas as pd\r\nimport numpy as np\r\n\r\ndataset = {\r\n 'bbox': [\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]]),\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]]),\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]]),\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]])\r\n ],\r\n 'input_ids': [1, 2, 3, 4]\r\n}\r\ndataset = Dataset.from_dict(dataset)\r\n```\r\n\r\nThis will work but to use it with the torch formatter you must specify the `Array2D` feature type in order to tell the shape:\r\n```python\r\nfrom datasets import Dataset, Array2D, Features, Value\r\nimport pandas as pd\r\nimport numpy as np\r\n\r\ndataset = {\r\n 'bbox': [\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]]),\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]]),\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]]),\r\n np.array([[1,2,3,4],[1,2,3,4],[1,2,3,4]])\r\n ],\r\n 'input_ids': [1, 2, 3, 4]\r\n}\r\ndataset = Dataset.from_dict(dataset, features=Features({\r\n \"bbox\": Array2D(shape=(3, 4), dtype=\"int64\"),\r\n \"input_ids\": Value(\"int64\")\r\n}))\r\ndataset.set_format(\"torch\")\r\nprint(dataset[0]['bbox'])\r\n# tensor([[1, 2, 3, 4],\r\n# [1, 2, 3, 4],\r\n# [1, 2, 3, 4]])\r\n```\r\nIf you don't specify the `Array2D` feature type, then the inferred type will be Sequence(Sequence(Value(\"int64\"))) and therefore the torch formatter will return list of tensors", "Thanks for the explanation. \r\nWith my original DataFrame, I did\r\n```\r\ndataset = dataset.to_dict(\"list\")\r\n```\r\nand then the rest of the transformation from dictionary works just fine." ]

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

[ "Hi! One option is use `map` with a function that overwrites the labels (`dset = dset.map(lamba _: {\"label\": 0}, features=dset.features`)). Or you can use the `remove_column` + `add_column` combination (`dset = dset.remove_columns(\"label\").add_column(\"label\", [0]*len(data)).cast(dset.features)`, but note that this approach creates an in-memory table for the added column instead of writing to disk, which could be problematic for large datasets.", "Hi! I tried your proposed solution, but it does not solve my problem unfortunately. I am working with a set of protein sequences that have been tokenized with ESM, but some sequences are longer than `max_length`, they have been truncated in the tokenization. So now I want to truncate my labels as well, but that does not work with a mapping (e.g. `dset.map` as you suggested). Specifically, what I did was the following:\r\n\r\n```\r\ndef postprocess_tokenize(tokenized_data):\r\n \"\"\"\r\n adjust label lengths if they dont match.\r\n \"\"\"\r\n if len(tokenized_data['input_ids']) < len(tokenized_data['labels']):\r\n new_labels = tokenized_data['labels'][:len(tokenized_data['input_ids'])]\r\n tokenized_data[\"labels\"] = new_labels\r\n return tokenized_data\r\n\r\ntokenized_data = tokenized_data.map(postprocess_tokenize, batched=True) # this does not adjust the labels...\r\n```\r\n\r\nAny tips on how to do this properly?\r\n\r\nMore generally, I am wondering why the DataCollator supports padding but does not support truncation? Seems odd to me.\r\n\r\nThanks in advance!" ]

[]

[]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "Hi, thanks for reporting! This issue stems from the changes introduced in https://github.com/huggingface/datasets/pull/4282 (cc @lhoestq), in which list casts are ignored if they don't change the list type (required to preserve `null` values). And `push_to_hub` does a special cast to embed external image files but doesn't change the types, hence the failure." ]

[]

[ "Awesome!\r\nLet's have this in the doc at the end :-)" ]

[]

[]

[]

[]

[]

[ "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)." ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "Thanks for reporting @sue99.\r\n\r\nUnfortunately. I'm not able to reproduce your problem:\r\n```python\r\nIn [1]: import datasets\r\n ...: from datasets import load_dataset\r\n ...: dataset = load_dataset(\"conll2003\")\r\n\r\nIn [2]: dataset\r\nOut[2]: \r\nDatasetDict({\r\n train: Dataset({\r\n features: ['id', 'tokens', 'pos_tags', 'chunk_tags', 'ner_tags'],\r\n num_rows: 14042\r\n })\r\n validation: Dataset({\r\n features: ['id', 'tokens', 'pos_tags', 'chunk_tags', 'ner_tags'],\r\n num_rows: 3251\r\n })\r\n test: Dataset({\r\n features: ['id', 'tokens', 'pos_tags', 'chunk_tags', 'ner_tags'],\r\n num_rows: 3454\r\n })\r\n})\r\n\r\nIn [3]: dataset[\"train\"][0]\r\nOut[3]: \r\n{'id': '0',\r\n 'tokens': ['EU',\r\n 'rejects',\r\n 'German',\r\n 'call',\r\n 'to',\r\n 'boycott',\r\n 'British',\r\n 'lamb',\r\n '.'],\r\n 'pos_tags': [22, 42, 16, 21, 35, 37, 16, 21, 7],\r\n 'chunk_tags': [11, 21, 11, 12, 21, 22, 11, 12, 0],\r\n 'ner_tags': [3, 0, 7, 0, 0, 0, 7, 0, 0]}\r\n```\r\n\r\nJust guessing: might be the case that you are calling `load_dataset` from a working directory that contains a local folder named `conll2003` (containing the raw data files)? If that is the case, `datasets` library gives precedence to the local folder over the dataset on the Hub. " ]

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

[ "Hi @theo-m !\r\n\r\nA few lines above this line, you'll find that the `_split_generators` method returns a list of `SplitGenerator`s objects:\r\n\r\n```python\r\ndatasets.SplitGenerator(\r\n name=datasets.Split.VALIDATION,\r\n # These kwargs will be passed to _generate_examples\r\n gen_kwargs={\r\n \"filepath\": os.path.join(data_dir, \"dev.jsonl\"),\r\n \"split\": \"dev\",\r\n },\r\n),\r\n```\r\n\r\nNotice the `gen_kwargs` argument passed to the constructor of `SplitGenerator`: this dict will be unpacked as keyword arguments to pass to the `_generat_examples` method (in this case the `filepath` and `split` arguments).\r\n\r\nLet me know if that helps!", "Oh ok I hadn't made the connection between those two, will offer a tweak to the comment and the template then - thanks!" ]

[]

[]

[]

[ "I have noticed that the master branch of your fork has diverged from the one of the repo. This is probably what causes the mess in the github diff \"Files changed\".\r\n\r\nI would suggest to re-fork the `datasets` repo and recreate a new branch and a new PR. ", "You're pretty close to having all things ready to merge !\r\nFeel free to ping me when you have a new PR", "Closing this one in favor of #1575 " ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "Hi ! I think this issue is more related to the `transformers` project. Could you open an issue on https://github.com/huggingface/transformers ?\r\n\r\nAnyway I think the issue could be that both wikipedia and bookcorpusopen have an additional \"title\" column, contrary to wikitext which only has a \"text\" column. After calling `load_dataset`, can you try doing `dataset = dataset.remove_columns(\"title\")` ?", "Removing the \"title\" column works! Thanks for your advice.\r\n\r\nMaybe I should still create an issue to `transformers' to mark this solution?" ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008227 / 0.011353 (-0.003126) | 0.004496 / 0.011008 (-0.006512) | 0.099319 / 0.038508 (0.060811) | 0.029929 / 0.023109 (0.006820) | 0.296686 / 0.275898 (0.020788) | 0.355372 / 0.323480 (0.031892) | 0.006864 / 0.007986 (-0.001122) | 0.003458 / 0.004328 (-0.000871) | 0.077234 / 0.004250 (0.072983) | 0.037072 / 0.037052 (0.000020) | 0.311675 / 0.258489 (0.053186) | 0.338965 / 0.293841 (0.045124) | 0.033562 / 0.128546 (-0.094985) | 0.011399 / 0.075646 (-0.064248) | 0.322406 / 0.419271 (-0.096865) | 0.043034 / 0.043533 (-0.000499) | 0.298083 / 0.255139 (0.042944) | 0.323661 / 0.283200 (0.040462) | 0.089380 / 0.141683 (-0.052303) | 1.479363 / 1.452155 (0.027208) | 1.518337 / 1.492716 (0.025620) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.177822 / 0.018006 (0.159816) | 0.400806 / 0.000490 (0.400317) | 0.002121 / 0.000200 (0.001921) | 0.000074 / 0.000054 (0.000019) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.021986 / 0.037411 (-0.015426) | 0.096749 / 0.014526 (0.082223) | 0.101443 / 0.176557 (-0.075113) | 0.137519 / 0.737135 (-0.599616) | 0.105558 / 0.296338 (-0.190780) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.418983 / 0.215209 (0.203774) | 4.189579 / 2.077655 (2.111924) | 1.877831 / 1.504120 (0.373711) | 1.666213 / 1.541195 (0.125019) | 1.680735 / 1.468490 (0.212245) | 0.693033 / 4.584777 (-3.891744) | 3.420553 / 3.745712 (-0.325160) | 1.819647 / 5.269862 (-3.450214) | 1.144934 / 4.565676 (-3.420743) | 0.082209 / 0.424275 (-0.342066) | 0.012433 / 0.007607 (0.004826) | 0.526781 / 0.226044 (0.300737) | 5.273689 / 2.268929 (3.004760) | 2.323468 / 55.444624 (-53.121156) | 1.960508 / 6.876477 (-4.915969) | 2.035338 / 2.142072 (-0.106735) | 0.812789 / 4.805227 (-3.992438) | 0.148429 / 6.500664 (-6.352235) | 0.064727 / 0.075469 (-0.010742) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.253218 / 1.841788 (-0.588569) | 13.303426 / 8.074308 (5.229118) | 13.651074 / 10.191392 (3.459682) | 0.135178 / 0.680424 (-0.545246) | 0.028483 / 0.534201 (-0.505717) | 0.393284 / 0.579283 (-0.185999) | 0.401957 / 0.434364 (-0.032407) | 0.457136 / 0.540337 (-0.083201) | 0.535835 / 1.386936 (-0.851101) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006335 / 0.011353 (-0.005017) | 0.004454 / 0.011008 (-0.006554) | 0.097565 / 0.038508 (0.059057) | 0.026917 / 0.023109 (0.003808) | 0.350779 / 0.275898 (0.074881) | 0.391979 / 0.323480 (0.068499) | 0.004648 / 0.007986 (-0.003337) | 0.003204 / 0.004328 (-0.001124) | 0.076987 / 0.004250 (0.072737) | 0.035257 / 0.037052 (-0.001796) | 0.347193 / 0.258489 (0.088704) | 0.391462 / 0.293841 (0.097621) | 0.031244 / 0.128546 (-0.097302) | 0.011460 / 0.075646 (-0.064186) | 0.321606 / 0.419271 (-0.097665) | 0.041218 / 0.043533 (-0.002315) | 0.341884 / 0.255139 (0.086745) | 0.374920 / 0.283200 (0.091720) | 0.086383 / 0.141683 (-0.055300) | 1.501750 / 1.452155 (0.049595) | 1.565060 / 1.492716 (0.072344) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.165447 / 0.018006 (0.147441) | 0.401885 / 0.000490 (0.401395) | 0.000975 / 0.000200 (0.000775) | 0.000070 / 0.000054 (0.000015) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024494 / 0.037411 (-0.012917) | 0.097334 / 0.014526 (0.082808) | 0.105324 / 0.176557 (-0.071232) | 0.142430 / 0.737135 (-0.594705) | 0.107249 / 0.296338 (-0.189089) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.441632 / 0.215209 (0.226423) | 4.407729 / 2.077655 (2.330074) | 2.078167 / 1.504120 (0.574047) | 1.864210 / 1.541195 (0.323015) | 1.885948 / 1.468490 (0.417458) | 0.693974 / 4.584777 (-3.890803) | 3.386837 / 3.745712 (-0.358875) | 1.840291 / 5.269862 (-3.429571) | 1.150524 / 4.565676 (-3.415153) | 0.082240 / 0.424275 (-0.342035) | 0.012488 / 0.007607 (0.004881) | 0.537589 / 0.226044 (0.311545) | 5.404007 / 2.268929 (3.135078) | 2.537467 / 55.444624 (-52.907157) | 2.190775 / 6.876477 (-4.685702) | 2.224746 / 2.142072 (0.082674) | 0.799524 / 4.805227 (-4.005703) | 0.150639 / 6.500664 (-6.350025) | 0.066473 / 0.075469 (-0.008997) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.258559 / 1.841788 (-0.583228) | 13.773583 / 8.074308 (5.699275) | 13.964322 / 10.191392 (3.772930) | 0.156295 / 0.680424 (-0.524129) | 0.016824 / 0.534201 (-0.517377) | 0.377476 / 0.579283 (-0.201807) | 0.390163 / 0.434364 (-0.044201) | 0.442541 / 0.540337 (-0.097796) | 0.529404 / 1.386936 (-0.857532) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#8f500a5c554b213aafe87293bd593920567742c3 \"CML watermark\")\n" ]

[]

[ "@lhoestq @abhishekkrthakur what happened here ?\r\n", "@lhoestq everything alright here ?", "@tanmoyio please have patience. @lhoestq has to look at 150+ PRs and it may take time. The PR looks good to me but we wait for his confirmation :) 🤗 " ]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.012442 / 0.011353 (0.001089) | 0.006274 / 0.011008 (-0.004734) | 0.128249 / 0.038508 (0.089741) | 0.040117 / 0.023109 (0.017008) | 0.383725 / 0.275898 (0.107827) | 0.510494 / 0.323480 (0.187014) | 0.009037 / 0.007986 (0.001051) | 0.008256 / 0.004328 (0.003927) | 0.105329 / 0.004250 (0.101079) | 0.046909 / 0.037052 (0.009857) | 0.401980 / 0.258489 (0.143491) | 0.461332 / 0.293841 (0.167491) | 0.065629 / 0.128546 (-0.062917) | 0.020043 / 0.075646 (-0.055604) | 0.453773 / 0.419271 (0.034501) | 0.063456 / 0.043533 (0.019923) | 0.384458 / 0.255139 (0.129319) | 0.449699 / 0.283200 (0.166499) | 0.118197 / 0.141683 (-0.023486) | 1.915080 / 1.452155 (0.462925) | 1.957132 / 1.492716 (0.464416) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.209657 / 0.018006 (0.191651) | 0.592478 / 0.000490 (0.591988) | 0.004137 / 0.000200 (0.003937) | 0.000124 / 0.000054 (0.000069) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029607 / 0.037411 (-0.007804) | 0.129559 / 0.014526 (0.115033) | 0.148326 / 0.176557 (-0.028231) | 0.190506 / 0.737135 (-0.546629) | 0.143177 / 0.296338 (-0.153162) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.626166 / 0.215209 (0.410957) | 6.612680 / 2.077655 (4.535026) | 2.432354 / 1.504120 (0.928234) | 2.051482 / 1.541195 (0.510287) | 2.055822 / 1.468490 (0.587332) | 1.210099 / 4.584777 (-3.374678) | 5.498117 / 3.745712 (1.752405) | 3.054838 / 5.269862 (-2.215024) | 2.182875 / 4.565676 (-2.382802) | 0.144518 / 0.424275 (-0.279757) | 0.014132 / 0.007607 (0.006525) | 0.801805 / 0.226044 (0.575761) | 7.911235 / 2.268929 (5.642307) | 3.372762 / 55.444624 (-52.071862) | 2.517266 / 6.876477 (-4.359210) | 2.515329 / 2.142072 (0.373256) | 1.501731 / 4.805227 (-3.303497) | 0.252569 / 6.500664 (-6.248096) | 0.080987 / 0.075469 (0.005518) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.709880 / 1.841788 (-0.131907) | 18.640340 / 8.074308 (10.566032) | 23.560908 / 10.191392 (13.369516) | 0.265680 / 0.680424 (-0.414744) | 0.046438 / 0.534201 (-0.487763) | 0.571973 / 0.579283 (-0.007310) | 0.642425 / 0.434364 (0.208061) | 0.698167 / 0.540337 (0.157830) | 0.842132 / 1.386936 (-0.544804) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009268 / 0.011353 (-0.002085) | 0.006052 / 0.011008 (-0.004956) | 0.133448 / 0.038508 (0.094939) | 0.034417 / 0.023109 (0.011308) | 0.435573 / 0.275898 (0.159675) | 0.479642 / 0.323480 (0.156162) | 0.008016 / 0.007986 (0.000030) | 0.006616 / 0.004328 (0.002288) | 0.106256 / 0.004250 (0.102005) | 0.048995 / 0.037052 (0.011942) | 0.450056 / 0.258489 (0.191567) | 0.511027 / 0.293841 (0.217187) | 0.052928 / 0.128546 (-0.075618) | 0.020824 / 0.075646 (-0.054822) | 0.450105 / 0.419271 (0.030834) | 0.062729 / 0.043533 (0.019196) | 0.438887 / 0.255139 (0.183748) | 0.468732 / 0.283200 (0.185532) | 0.116101 / 0.141683 (-0.025582) | 1.909689 / 1.452155 (0.457534) | 2.042007 / 1.492716 (0.549291) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.198265 / 0.018006 (0.180259) | 0.541799 / 0.000490 (0.541309) | 0.003938 / 0.000200 (0.003738) | 0.000116 / 0.000054 (0.000062) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.035933 / 0.037411 (-0.001478) | 0.130754 / 0.014526 (0.116229) | 0.146143 / 0.176557 (-0.030414) | 0.202042 / 0.737135 (-0.535094) | 0.155648 / 0.296338 (-0.140691) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.691123 / 0.215209 (0.475914) | 6.708370 / 2.077655 (4.630715) | 2.957120 / 1.504120 (1.453000) | 2.558350 / 1.541195 (1.017155) | 2.611271 / 1.468490 (1.142781) | 1.327355 / 4.584777 (-3.257422) | 5.755975 / 3.745712 (2.010263) | 3.295556 / 5.269862 (-1.974305) | 2.159831 / 4.565676 (-2.405845) | 0.161409 / 0.424275 (-0.262866) | 0.015470 / 0.007607 (0.007863) | 0.840611 / 0.226044 (0.614567) | 8.550064 / 2.268929 (6.281136) | 3.832013 / 55.444624 (-51.612612) | 3.032909 / 6.876477 (-3.843568) | 3.155651 / 2.142072 (1.013578) | 1.612486 / 4.805227 (-3.192741) | 0.273789 / 6.500664 (-6.226875) | 0.085618 / 0.075469 (0.010149) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.808376 / 1.841788 (-0.033412) | 18.267614 / 8.074308 (10.193306) | 21.047679 / 10.191392 (10.856286) | 0.259089 / 0.680424 (-0.421335) | 0.029211 / 0.534201 (-0.504990) | 0.556303 / 0.579283 (-0.022980) | 0.625264 / 0.434364 (0.190900) | 0.680814 / 0.540337 (0.140476) | 0.810146 / 1.386936 (-0.576790) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#20ea76c80e07acad78cf67198a4046a982feda21 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.008779 / 0.011353 (-0.002574) | 0.004644 / 0.011008 (-0.006364) | 0.099814 / 0.038508 (0.061306) | 0.029830 / 0.023109 (0.006721) | 0.299159 / 0.275898 (0.023261) | 0.354815 / 0.323480 (0.031335) | 0.006968 / 0.007986 (-0.001018) | 0.003521 / 0.004328 (-0.000808) | 0.077687 / 0.004250 (0.073437) | 0.035019 / 0.037052 (-0.002034) | 0.309548 / 0.258489 (0.051059) | 0.345228 / 0.293841 (0.051387) | 0.033644 / 0.128546 (-0.094902) | 0.011564 / 0.075646 (-0.064083) | 0.321835 / 0.419271 (-0.097437) | 0.041798 / 0.043533 (-0.001735) | 0.298190 / 0.255139 (0.043051) | 0.328874 / 0.283200 (0.045674) | 0.088175 / 0.141683 (-0.053508) | 1.481755 / 1.452155 (0.029600) | 1.503085 / 1.492716 (0.010369) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.170930 / 0.018006 (0.152924) | 0.422155 / 0.000490 (0.421666) | 0.001708 / 0.000200 (0.001509) | 0.000083 / 0.000054 (0.000028) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.022588 / 0.037411 (-0.014824) | 0.095775 / 0.014526 (0.081249) | 0.103939 / 0.176557 (-0.072618) | 0.138441 / 0.737135 (-0.598694) | 0.107896 / 0.296338 (-0.188442) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.418243 / 0.215209 (0.203034) | 4.171432 / 2.077655 (2.093777) | 1.906029 / 1.504120 (0.401909) | 1.698174 / 1.541195 (0.156979) | 1.748339 / 1.468490 (0.279849) | 0.691026 / 4.584777 (-3.893751) | 3.393354 / 3.745712 (-0.352358) | 2.722412 / 5.269862 (-2.547450) | 1.462439 / 4.565676 (-3.103238) | 0.084713 / 0.424275 (-0.339562) | 0.012131 / 0.007607 (0.004524) | 0.522153 / 0.226044 (0.296109) | 5.197916 / 2.268929 (2.928988) | 2.314270 / 55.444624 (-53.130354) | 1.986599 / 6.876477 (-4.889878) | 2.012757 / 2.142072 (-0.129315) | 0.802540 / 4.805227 (-4.002687) | 0.148673 / 6.500664 (-6.351991) | 0.065924 / 0.075469 (-0.009545) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.263790 / 1.841788 (-0.577998) | 13.874784 / 8.074308 (5.800476) | 13.842276 / 10.191392 (3.650884) | 0.149002 / 0.680424 (-0.531422) | 0.028550 / 0.534201 (-0.505651) | 0.396913 / 0.579283 (-0.182370) | 0.401543 / 0.434364 (-0.032821) | 0.473754 / 0.540337 (-0.066583) | 0.560455 / 1.386936 (-0.826481) |\n\n</details>\nPyArrow==latest\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.006724 / 0.011353 (-0.004629) | 0.004507 / 0.011008 (-0.006502) | 0.098447 / 0.038508 (0.059939) | 0.027888 / 0.023109 (0.004779) | 0.428956 / 0.275898 (0.153058) | 0.451557 / 0.323480 (0.128077) | 0.005056 / 0.007986 (-0.002929) | 0.003363 / 0.004328 (-0.000965) | 0.075990 / 0.004250 (0.071740) | 0.038688 / 0.037052 (0.001635) | 0.421550 / 0.258489 (0.163061) | 0.459480 / 0.293841 (0.165639) | 0.031408 / 0.128546 (-0.097138) | 0.011559 / 0.075646 (-0.064088) | 0.320054 / 0.419271 (-0.099217) | 0.041917 / 0.043533 (-0.001616) | 0.420878 / 0.255139 (0.165739) | 0.444813 / 0.283200 (0.161613) | 0.090409 / 0.141683 (-0.051274) | 1.490058 / 1.452155 (0.037904) | 1.645206 / 1.492716 (0.152489) |\n\n### Benchmark: benchmark_getitem\\_100B.json\n\n| metric | get_batch_of\\_1024\\_random_rows | get_batch_of\\_1024\\_rows | get_first_row | get_last_row |\n|--------|---|---|---|---|\n| new / old (diff) | 0.221105 / 0.018006 (0.203099) | 0.407537 / 0.000490 (0.407047) | 0.000410 / 0.000200 (0.000210) | 0.000059 / 0.000054 (0.000004) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024658 / 0.037411 (-0.012754) | 0.099230 / 0.014526 (0.084705) | 0.107788 / 0.176557 (-0.068769) | 0.143040 / 0.737135 (-0.594096) | 0.109440 / 0.296338 (-0.186899) |\n\n### Benchmark: benchmark_iterating.json\n\n| metric | read 5000 | read 50000 | read_batch 50000 10 | read_batch 50000 100 | read_batch 50000 1000 | read_formatted numpy 5000 | read_formatted pandas 5000 | read_formatted tensorflow 5000 | read_formatted torch 5000 | read_formatted_batch numpy 5000 10 | read_formatted_batch numpy 5000 1000 | shuffled read 5000 | shuffled read 50000 | shuffled read_batch 50000 10 | shuffled read_batch 50000 100 | shuffled read_batch 50000 1000 | shuffled read_formatted numpy 5000 | shuffled read_formatted_batch numpy 5000 10 | shuffled read_formatted_batch numpy 5000 1000 |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.453303 / 0.215209 (0.238094) | 4.520376 / 2.077655 (2.442722) | 2.133909 / 1.504120 (0.629789) | 1.926996 / 1.541195 (0.385801) | 2.019870 / 1.468490 (0.551380) | 0.707423 / 4.584777 (-3.877354) | 3.391903 / 3.745712 (-0.353809) | 1.860661 / 5.269862 (-3.409201) | 1.159940 / 4.565676 (-3.405736) | 0.083773 / 0.424275 (-0.340502) | 0.012228 / 0.007607 (0.004621) | 0.554666 / 0.226044 (0.328622) | 5.567564 / 2.268929 (3.298636) | 2.636718 / 55.444624 (-52.807907) | 2.240215 / 6.876477 (-4.636262) | 2.218951 / 2.142072 (0.076879) | 0.817167 / 4.805227 (-3.988060) | 0.151633 / 6.500664 (-6.349032) | 0.066515 / 0.075469 (-0.008954) |\n\n### Benchmark: benchmark_map_filter.json\n\n| metric | filter | map fast-tokenizer batched | map identity | map identity batched | map no-op batched | map no-op batched numpy | map no-op batched pandas | map no-op batched pytorch | map no-op batched tensorflow |\n|--------|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 1.296665 / 1.841788 (-0.545123) | 13.997898 / 8.074308 (5.923590) | 13.286607 / 10.191392 (3.095215) | 0.148906 / 0.680424 (-0.531518) | 0.016600 / 0.534201 (-0.517601) | 0.377459 / 0.579283 (-0.201824) | 0.379938 / 0.434364 (-0.054426) | 0.461628 / 0.540337 (-0.078709) | 0.550592 / 1.386936 (-0.836344) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#053f51a3e2adb762236eb29dd02791307f45f02f \"CML watermark\")\n" ]

[]

[]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

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

[]

[]

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

[ "The \"exact match\" deduplication will be possible when we resolve https://github.com/huggingface/datasets/issues/2514 (first, https://github.com/apache/arrow/issues/30950 needs to be addressed on the Arrow side). In the meantime, you can use Polars or DuckDB (e.g., via [datasets-sql](https://github.com/mariosasko/datasets_sql)).\r\n\r\nFuzzy deduplication is out-of-scope for now ([splink](https://github.com/moj-analytical-services/splink) is probably the best tool for it).", "This library can be an intermediate solution : https://github.com/ChenghaoMou/text-dedup/tree/main", "I have been using polars to remove duplicates but it would be nice to do it directly in pyarrow.\r\n\r\nFor example,\r\n\r\n1. Read dataset with pyarrow\r\n2. Use scan_pyarrow_dataset() with Polars to create a LazyFrame\r\n3. Use sort and unique to remove duplicates based on a subset of columns\r\n4. Convert to table and save data with ds.write_dataset()\r\n\r\nThere are times where that workflow makes perfect sense because I do additional transformations with Polars. Most of the time I am simply just reading dataset A and writing dataset B without duplicates though, and I wish I could use a pyarrow scanner or table directly. " ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

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

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

[]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._" ]

[]

[ "Hey @lhoestq , thanks for the feedback on this! I updated the `_generate_examples` with some comments on the process, and reduced the `dummy_data.zip` down quite a bit as well. \r\n\r\nFor the dummy data, I reduced the text to only three sentences, and aligned the corresponding entity/token/sentence annotations to that (reduced accordingly). The frames file is a strange combined format for the annotations and I found if I reduced that that would break the parser no matter what I did, so I left that as is. The difference between a reduced frames and non-reduced frames file in the compressed dummy data was only about ~4kb, so hopefully leaving this as is will be ok!" ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "Hi ! Indeed there was a breaking change in `huggingface_hub` 0.5.0 in `HfApi.create_repo`, which is called here in `datasets` by passing the org name in both the `repo_id` and the `organization` arguments:\r\n\r\nhttps://github.com/huggingface/datasets/blob/2230f7f7d7fbaf102cff356f5a8f3bd1561bea43/src/datasets/arrow_dataset.py#L3363-L3369\r\n\r\nI think we should fix that in `huggingface_hub`, will keep you posted. In the meantime please use `huggingface_hub` 0.4.0", "I'll be sending a fix for this later today on the `huggingface_hub` side.\r\n\r\nThe error would be converted to a `FutureWarning` if `datasets` uses kwargs instead of positional, for example here: \r\n\r\nhttps://github.com/huggingface/datasets/blob/2230f7f7d7fbaf102cff356f5a8f3bd1561bea43/src/datasets/arrow_dataset.py#L3363-L3369\r\n\r\nto be:\r\n\r\n``` python\r\n api.create_repo(\r\n name=dataset_name,\r\n token=token,\r\n repo_type=\"dataset\",\r\n organization=organization,\r\n private=private,\r\n )\r\n```\r\n\r\nBut `name` and `organization` are deprecated in `huggingface_hub=0.5`, and people should pass `repo_id='org/name` instead. Note that `repo_id` was introduced in 0.5 and if `datasets` wants to support older `huggingface_hub` versions (which I encourage it to do), there needs to be a helper function to do that. It can be something like:\r\n\r\n\r\n```python\r\ndef create_repo(\r\n client,\r\n name: str,\r\n token: Optional[str] = None,\r\n organization: Optional[str] = None,\r\n private: Optional[bool] = None,\r\n repo_type: Optional[str] = None,\r\n exist_ok: Optional[bool] = False,\r\n space_sdk: Optional[str] = None,\r\n) -> str:\r\n try:\r\n return client.create_repo(\r\n repo_id=f\"{organization}/{name}\",\r\n token=token,\r\n private=private,\r\n repo_type=repo_type,\r\n exist_ok=exist_ok,\r\n space_sdk=space_sdk,\r\n )\r\n except TypeError:\r\n return client.create_repo(\r\n name=name,\r\n organization=organization,\r\n token=token,\r\n private=private,\r\n repo_type=repo_type,\r\n exist_ok=exist_ok,\r\n space_sdk=space_sdk,\r\n )\r\n```\r\n\r\nin a `utils/_fixes.py` kinda file and and be used internally.\r\n\r\nI'll be sending a patch to `huggingface_hub` to convert the error reported in this issue to a `FutureWarning`.", "PR with the hotfix on the `huggingface_hub` side: https://github.com/huggingface/huggingface_hub/pull/822", "We can definitely change `push_to_hub` to use `repo_id` in `datasets` and require `huggingface_hub>=0.5.0`.\r\n\r\nLet me open a PR :)", "`huggingface_hub` 0.5.1 just got released with a fix, feel free to update `huggingface_hub` ;)" ]

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

[ "IMO, the data streaming test is good enough of a test that the dataset works correctly (assuming that we can more or less ensure that if streaming works then the non-streaming case will also work), so that for datasets that have a working dataset preview, we can remove the dummy data IMO. On the other hand, it seems like not all datasets have streaming enabled yet and for those datasets (if they are used a lot), I think it would be nice to continue testing some dummy data.\r\n\r\nI don't really have an opinion regarding the JSON metadata as I don't know enough about it.\r\n\r\n", "I don't know all the details, but generally I'd be in favor of unifying the metadata formats into YAML inside .md (and so deprecating the dataset_infos.json) \r\n\r\n(Ultimately the CI can run on \"HuggingFace Actions\" instead of on GitHub)", "The dataset_infos.json file currently has these useful infos for each dataset configuration, that I think can be moved to the dataset tags:\r\n- Size of the dataset in MB: download size, arrow file size, and total size (sum of download + arrow)\r\n- Size of each split in MB and number of examples. Again this can be moved to the dataset tags\r\n- Feature type of each column\r\n- supported task templates (it defines what columns correspond to the features and labels for example)\r\n\r\nBut it also has this, which I'm not sure if it should be in the tags or not:\r\n- Checksums of the downloaded files for integrity verifications\r\n\r\nSo ultimately this file could probably be deprecated in favor of having the infos in the tags.\r\n\r\n> Also note that for generating both (dataset_infos.json file and dummy data), the entire dataset needs being downloaded. This can be an issue for huge datasets (like WIT, with 400 GB of data).\r\n\r\nTo get the exact number of examples and size in MB of the dataset, one needs to download and generate it completely. IMO these infos are very important when someone considers using a dataset. Though using streaming we could do some extrapolation to have approximate values instead.\r\n\r\nFor the integrity verifications we also need the number of examples and the checksums of the downloaded files, so it requires the dataset to be fully downloaded once. This can be optional though.\r\n\r\n> IMO, the data streaming test is good enough of a test that the dataset works correctly (assuming that we can more or less ensure that if streaming works then the non-streaming case will also work)\r\n\r\nI agree with this. Usually if a dataset works in streaming mode, then it works in non-streaming mode (the other way around is not true though).\r\n\r\n> On the other hand, it seems like not all datasets have streaming enabled yet and for those datasets (if they are used a lot), I think it would be nice to continue testing some dummy data.\r\n\r\nYes indeed, or at least make sure that it was tested on the true data.", "(note that if we wanted to display sizes, etc we could also pretty easily parse the `dataset_infos.json` on the hub side)", "I agree that we can move the relevant parts of `dataset_infos.json` to the YAML tags.\r\n\r\n> On the other hand, it seems like not all datasets have streaming enabled yet and for those datasets (if they are used a lot), I think it would be nice to continue testing some dummy data. <\r\n> > Yes indeed, or at least make sure that it was tested on the true data.\r\n\r\nI like the idea of testing streaming and falling back to the dummy data test if streaming does not work. Generating dummy data can be very tedious, so this would be a nice incentive for the contributors to make their datasets streamable. ", "CC: @severo ", "About dummy data, please see e.g. this PR: https://github.com/huggingface/datasets/pull/3692/commits/62368daac0672041524a471386d5e78005cf357a\r\n- I updated the previous dummy data: I just had to rename the file and its directory\r\n - the dummy data zip contains only a single file: `pubmed22n0001.xml.gz`\r\n\r\nThen I discover it fails: https://app.circleci.com/pipelines/github/huggingface/datasets/9800/workflows/173a4433-8feb-4fc6-ab9e-59762084e3e1/jobs/60437\r\n```\r\nNo such file or directory: '.../dummy_data/pubmed22n0002.xml.gz'\r\n```\r\n- it needs dummy data for all the 1114 files: \r\n `_URLs = [f\"ftp://ftp.ncbi.nlm.nih.gov/pubmed/baseline/pubmed22n{i:04d}.xml.gz\" for i in range(1, 1115)]`\r\n- this confirms me that it never passed the test: these dummy data files were not present before my PR\r\n- therefore, is it really useful the data test if we just ignore it when it does not pass?\r\n\r\nIn relation with JSON metadata, I'm generating the file for `pubmed` (see above) in a GCP instance: it's running for more than 3 hours and only 9 million examples generated so far (before my PR, it had 32 million, now it has more).", "I mention in https://github.com/huggingface/datasets-server/wiki/Preliminary-design that the future \"datasets server\" could be in charge of generating both the dummy data and the dataset-info.json file if required (or their equivalent).", "Hi ! I think dummy data generation is out of scope for the datasets server, since it's about generating the original data files.\r\n\r\nThat would be amazing to have it generate the dataset_infos.json though !", "From some offline discussion with @mariosasko and especially for vision datasets, we'll probably not require dummy data anymore and use streaming instead :) This will make adding a new dataset much easier.\r\nThis should also make sure that streaming works as expected directly in the CI, without having to check the dataset viewer once the PR is merged", "OK. I removed the \"dummy data\" item from the services of the dataset server", "It seems that migration from dataset-info.json to dataset card YAML has been acted.\r\n\r\nProbably it's a good idea, but I didn't find the pros and cons of this decision, so I put some I could think of:\r\n\r\npros:\r\n- only one file to parse, share, sync\r\n- it gives a hint to the users that if you write your dataset card, you should also specify the metadata\r\n\r\ncons:\r\n- the metadata header might be very long, before reaching the start of the README/dataset card. It might be surprising when you edit the file because the metadata is not shown on top when the dataset card is rendered (dataset page). It also somewhat prevents including large strings like the checksums\r\n- YAML vs JSON: not sure which one is easier for users to fill and maintain\r\n- two concepts are mixed in the same file (metadata and documentation). This means that if you're interested only in one of them, you still have to know how to parse the whole file.\r\n- [low priority] besides the JSON file, we might want to support yaml or toml file if the user prefers (as [prettier](https://prettier.io/docs/en/configuration.html) and others do for their config files, for example). Inside the md, I understand that only YAML is allowed", "> the metadata header might be very long, before reaching the start of the README/dataset card. It might be surprising when you edit the file because the metadata is not shown on top when the dataset card is rendered (dataset page). It also somewhat prevents including large strings like the checksums\r\n\r\nNote that we could simply not have the checksums in the YAML metadata at all, or maybe at one point have a pointer to another file instead.\r\n\r\nWe can also choose to hide (collapse) certain sections in the YAML by default when we open the dataset card editor.\r\n\r\n> two concepts are mixed in the same file (metadata and documentation). This means that if you're interested only in one of them, you still have to know how to parse the whole file.\r\n\r\nI think it's fine for now. Later if we really end up with too many YAML sections we can see if we need to tweak the API endpoints or the `datasets`/`huggingface_hub` tools\r\n\r\n> YAML vs JSON: not sure which one is easier for users to fill and maintain\r\n\r\nRegarding YAML vs JSON: I think YAML is easier to write by hand, and I also think that it's better for consistency - i.e. we're using more and more YAML to configure models/datasets/spaces", "I didn't know the decision was already taken. Good to know. 😅", "> the metadata header might be very long, before reaching the start of the README/dataset card. It might be surprising when you edit the file because the metadata is not shown on top when the dataset card is rendered (dataset page). It also somewhat prevents including large strings like the checksums\r\n\r\nWe can definitely work on this on the hub side to make the UX better", "Tensorflow Datasets catalog includes a community catalog where you can find and use HF datasets (see [here](https://www.tensorflow.org/datasets/community_catalog/huggingface)).\r\n\r\nFYI I noticed today that they are using the exported dataset_infos.json files from github to get the metadata (see their code [here](https://github.com/tensorflow/datasets/blob/a482f01c036a10496f5e22e69a2ef81b707cc418/tensorflow_datasets/scripts/documentation/build_community_catalog.py#L261))", "Metadata is now stored as YAML, and dummy data is deprecated, so I think we can close this issue." ]

[]

[ "Nice !\r\n\r\nAlso it looks like we can have a train_test_split method for free:\r\n```python\r\ntrain_indices, test_indices = train_test_split(range(len(dataset)))\r\ntrain = dataset.sort(indices=train_indices)\r\ntest = dataset.sort(indices=test_indices)\r\n```\r\n\r\nand a shuffling method for free:\r\n```python\r\nshuffled_indices = shuffle(range(len(dataset)))\r\nshuffled_dataset = dataset.sort(indices=shuffled_indices)\r\n```\r\n\r\nMaybe we can have a specific API for train_test_split and shuffle. They are two features asked quite often (see #147, #166)", "Ok, I think this one is ready to merge.\r\n\r\n@patrickvonplaten @jplu @mariamabarham @joeddav @n1t0 @julien-c you may want to give it a look, it adds a bunch of methods to reorder/split/select rows in a dataset:\r\n- `dataset.select(indices)`: Create a new dataset with rows selected following the list/array of indices (which can have a different size than the dataset and contain duplicated indices, the only constrain is that all the integers in the list must be smaller than the dataset size, otherwise we're indexing outside the dataset...)\r\n- `dataset.sort(column_name)`: sort a dataset according to a column (has to be a column with a numpy compatible type)\r\n- `dataset.shuffle(seed)`: shuffle a dataset rows\r\n- `dataset.train_test_split(test_size, train_size)`: Return a dictionary with two random train and test subsets (`train` and `test` ``Dataset`` splits)\r\n\r\nAll these methods are **not** in-place which means they return new ``Dataset``, which is the default behavior in the library.", "> Might be a solution to put 0.25 and 0.75 as default values for respectively `test_size` and `train_size`. WDYT?\r\n\r\nAccording to sklearn documentation, it is indeed set to 0.25 and 0.75 if both `test_size` and `train_size` are None.\r\nLet me add it.", "I think we're good to go now :) @joeddav @thomwolf @jplu " ]

[ { "color": "E5583E", "default": false, "description": "Related to the dataset viewer on huggingface.co", "id": 3470211881, "name": "dataset-viewer", "node_id": "LA_kwDODunzps7O1zsp", "url": "https://api.github.com/repos/huggingface/datasets/labels/dataset-viewer" } ]

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

[]

[]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]

[ "The `Trainer` support `IterableDataset`, not just datasets." ]

[ { "color": "d73a4a", "default": true, "description": "Something isn't working", "id": 1935892857, "name": "bug", "node_id": "MDU6TGFiZWwxOTM1ODkyODU3", "url": "https://api.github.com/repos/huggingface/datasets/labels/bug" } ]