Datasets:

---

Francesco-A

/

github-issues_huggingface-datasets

like 0

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

[ { "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 @aleSuglia, thanks for reporting.\r\n\r\nWe are having a look at it. \r\n\r\nWe transfer this issue to the Community tab of the corresponding Hub dataset: https://huggingface.co/datasets/Heriot-WattUniversity/dialog-babi/discussions" ]

[ { "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" } ]

[ "Never mind. It is because of the caching of datasets..." ]

[]

[ "Hi !\r\nThanks for reporting :) \r\nIndeed this is an issue on the `datasets` side.\r\nI'm creating a PR", "Thanks @lhoestq !" ]

[]

[]

[]

[]

[ { "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" } ]

[ "Thanks, looking at it!", "Finally fixed! We updated the dataset viewer and it fixed the issue.\r\n\r\nhttps://huggingface.co/datasets/universal_dependencies/viewer/aqz_tudet/train\r\n\r\n<img width=\"1561\" alt=\"Capture d’écran 2022-09-07 à 13 29 18\" src=\"https://user-images.githubusercontent.com/1676121/188867795-4f7dd438-d4f2-46cd-8a92-20a37fb2d6bc.png\">\r\n" ]

[]

[ "Hi ! I'm not sure we should make this split decision arbitrarily on our side. Users can split it afterwards to whatever they want using `dataset.train_test_split` for example.\r\nMoreover it looks like there's already papers that use this dataset and propose their own splits ([here](http://xanthippi.ceid.upatras.gr/HealthSign/resources/Publications/sitis_paper_25_10.pdf) 80-20) \r\nWhat do you think ?", "I was not aware of the `train_test_split` method, thanks!\r\nSoe ven though it contributes to reproducibility, no need to do this split then." ]

[ { "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" } ]

[ "This will be fixed by https://github.com/huggingface/datasets/pull/3221" ]

[ { "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" } ]

[ "usually you can replace `download` in your dataset script with `download_and_prepare()` - could you share the code for your dataset here? :-) ", "I have an initial version here: https://github.com/EntilZha/nlp/tree/master/datasets/qanta Thats pretty close to what I'll do as a PR, but still want to do some more sanity checks/tests (just got tests passing).\r\n\r\nI figured out how to get all tests passing by adding a download command and some finagling with the data zip https://github.com/EntilZha/nlp/blob/master/tests/utils.py#L127\r\n\r\n", "I'm quite positive that you can just replace the `dl_manager.download()` statements here: https://github.com/EntilZha/nlp/blob/4d46443b65f1f756921db8275594e6af008a1de7/datasets/qanta/qanta.py#L194 with `dl_manager.download_and_extract()` even though you don't extract anything. I would prefer to avoid adding more functions to the MockDataLoadManager and keep it as simple as possible (It's already to complex now IMO). \r\n\r\nCould you check if you can replace the `download()` function? ", "I might be doing something wrong, but swapping those two gives this error:\r\n```\r\n> with open(path) as f:\r\nE IsADirectoryError: [Errno 21] Is a directory: 'datasets/qanta/dummy/mode=first,char_skip=25/2018.4.18/dummy_data-zip-extracted/dummy_data'\r\n\r\nsrc/nlp/datasets/qanta/3d965403133687b819905ead4b69af7bcee365865279b2f797c79f809b4490c3/qanta.py:280: IsADirectoryError\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n```\r\n\r\nSo it seems like the directory name is getting passed. Is this not functioning as expected, or is there some caching happening maybe? I deleted the dummy files and re-ran the import script with no changes. I'm digging a bit in with a debugger, but no clear reason yet", "From what I can tell here: https://github.com/huggingface/nlp/blob/master/tests/utils.py#L115\r\n\r\n1. `data_url` is the correct http link\r\n2. `path_to_dummy_data` is a directory, which is causing the issue\r\n\r\nThat path comes from `download_dummy_data`, which I think assumes that the data comes from the zip file, but isn't aware of individual files. So it seems like it data manager needs to be aware if the url its getting is for a file or a zip/directory, and pass this information along. This might happen in `download_dummy_data`, but probably better to happen in `download_and_extract`? Maybe a simple check to see if `os.path.basename` returns the dummy data zip filename, if not then join paths with the basename of the url?", "I think the dataset script works correctly. Just the dummy data structure seems to be wrong. I will soon add more commands that should make the create of the dummy data easier.\r\n\r\nI'd recommend that you won't concentrate too much on the dummy data.\r\nIf you manage to load the dataset correctly via:\r\n\r\n```python \r\n# use local path to qanta\r\nnlp.load_dataset(\"./datasets/qanta\")\r\n```\r\n\r\nthen feel free to open a PR and we will look into the dummy data problem together :-) \r\n\r\nAlso please make sure that the Version is in the format 1.0.0 (three numbers separated by two points) - not a date. ", "The script loading seems to work fine so I'll work on getting a PR open after a few sanity checks on the data.\r\n\r\nOn version, we currently have it versioned with YYYY.MM.DD scheme so it would be nice to not change that, but will it cause issues?", "> The script loading seems to work fine so I'll work on getting a PR open after a few sanity checks on the data.\r\n> \r\n> On version, we currently have it versioned with YYYY.MM.DD scheme so it would be nice to not change that, but will it cause issues?\r\n\r\nIt would cause issues for sure for the tests....not sure if it would also cause issues otherwise.\r\n\r\nI would prefer to keep the same version style as we have for other models. You could for example simply add version 1.0.0 and add a comment with the date you currently use for the versioning.\r\n\r\n What is your opinion regarding the version here @lhoestq @mariamabarham @thomwolf ? ", "Maybe use the YYYY.MM.DD as the config name ? That's what we are doing for wikipedia", "> Maybe use the YYYY.MM.DD as the config name ? That's what we are doing for wikipedia\r\n\r\nI'm not sure if this will work because the name should be unique and it seems that he has multiple config name in his data with the same version.\r\nAs @patrickvonplaten suggested, I think you can add a comment about the version in the data description.", "Actually maybe our versioning format (inherited from tfds) is too strong for what we use it for?\r\nWe could allow any string maybe?\r\n\r\nI see it more and more like an identifier for the user that we will back with a serious hashing/versioning system.- so we could let the user quite free on it.", "I'm good with either putting it in description, adding it to the config, or loosening version formatting. I mostly don't have a full conceptual grasp of what each identifier ends up meaning in the datasets code so hard to evaluate the best approach.\r\n\r\nFor background, the multiple formats is a consequence of:\r\n\r\n1. Each example is one multi-sentence trivia question\r\n2. For training, its better to treat each sentence as an example\r\n3. For evaluation, should test on: (1) first sentence, (2) full question, and (3) partial questions (does the model get the question right having seen the first half)\r\n\r\nWe use the date format for version since: (1) we expect some degree of updates since new questions come in every year and (2) the timestamp itself matches the Wikipedia dump that it is dependent on (so similar to the Wikipedia dataset).\r\n\r\nperhaps this is better discussed in https://github.com/huggingface/nlp/pull/169 or update title?" ]

[ { "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" } ]

[ "<s>I think this is because the tokenizer is stateful and because the order in which the splits are processed is not deterministic. Because of that, the hash of the tokenizer may change for certain splits, which causes issues with caching.\r\n\r\nTo fix this we can try making the order of the splits deterministic for map.</s>", "Actually this is not because of the order of the splits, but most likely because the tokenizer used to process the second split is in a state that has been modified by the first split.\r\n\r\nTherefore after reloading the first split from the cache, then the second split can't be reloaded since the tokenizer hasn't seen the first split (and therefore is considered a different tokenizer).\r\n\r\nThis is a bit trickier to fix, we can explore fixing this next week maybe", "Sorry didn't have the bandwidth to take care of this yet - will re-assign when I'm diving into it again !", "I had this issue with `run_speech_recognition_ctc.py` for wa2vec2.0 fine-tuning. I made a small change and the hash for the function (which includes tokenisation) is now the same before and after pre-porocessing. With the hash being the same, the caching works as intended.\r\n\r\nBefore:\r\n```\r\n def prepare_dataset(batch):\r\n # load audio\r\n sample = batch[audio_column_name]\r\n\r\n inputs = feature_extractor(sample[\"array\"], sampling_rate=sample[\"sampling_rate\"])\r\n batch[\"input_values\"] = inputs.input_values[0]\r\n batch[\"input_length\"] = len(batch[\"input_values\"])\r\n\r\n # encode targets\r\n additional_kwargs = {}\r\n if phoneme_language is not None:\r\n additional_kwargs[\"phonemizer_lang\"] = phoneme_language\r\n\r\n batch[\"labels\"] = tokenizer(batch[\"target_text\"], **additional_kwargs).input_ids\r\n\r\n return batch\r\n\r\n with training_args.main_process_first(desc=\"dataset map preprocessing\"):\r\n vectorized_datasets = raw_datasets.map(\r\n prepare_dataset,\r\n remove_columns=next(iter(raw_datasets.values())).column_names,\r\n num_proc=num_workers,\r\n desc=\"preprocess datasets\",\r\n )\r\n```\r\nAfter:\r\n```\r\n def prepare_dataset(batch, feature_extractor, tokenizer):\r\n # load audio\r\n sample = batch[audio_column_name]\r\n\r\n inputs = feature_extractor(sample[\"array\"], sampling_rate=sample[\"sampling_rate\"])\r\n batch[\"input_values\"] = inputs.input_values[0]\r\n batch[\"input_length\"] = len(batch[\"input_values\"])\r\n\r\n # encode targets\r\n additional_kwargs = {}\r\n if phoneme_language is not None:\r\n additional_kwargs[\"phonemizer_lang\"] = phoneme_language\r\n\r\n batch[\"labels\"] = tokenizer(batch[\"target_text\"], **additional_kwargs).input_ids\r\n\r\n return batch\r\n\r\n pd = lambda batch: prepare_dataset(batch, feature_extractor, tokenizer)\r\n\r\n with training_args.main_process_first(desc=\"dataset map preprocessing\"):\r\n vectorized_datasets = raw_datasets.map(\r\n pd,\r\n remove_columns=next(iter(raw_datasets.values())).column_names,\r\n num_proc=num_workers,\r\n desc=\"preprocess datasets\",\r\n )\r\n```", "Not sure why the second one would work and not the first one - they're basically the same with respect to hashing. In both cases the function is hashed recursively, and therefore the feature_extractor and the tokenizer are hashed the same way.\r\n\r\nWith which tokenizer or feature extractor are you experiencing this behavior ?\r\n\r\nDo you also experience this ?\r\n> Tokenization for some subsets are repeated at the 2nd and 3rd run. Starting from the 4th run, everything are loaded from cache.", "Thanks ! Hopefully this can be useful to others, and also to better understand and improve hashing/caching ", "`tokenizer.save_pretrained(training_args.output_dir)` produces a different tokenizer hash when loaded on restart of the script. When I was debugging before I was terminating the script prior to this command, then rerunning. \r\n\r\nI compared the tokenizer items on the first and second runs, there are two different items:\r\n1st:\r\n```\r\n('_additional_special_tokens', [AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True)])\r\n\r\n...\r\n\r\n('tokens_trie', <transformers.tokenization_utils.Trie object at 0x7f4d6d0ddb38>)\r\n```\r\n\r\n2nd:\r\n```\r\n('_additional_special_tokens', [AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"<s>\", rstrip=False, lstrip=False, single_word=False, normalized=True), AddedToken(\"</s>\", rstrip=False, lstrip=False, single_word=False, normalized=True)])\r\n\r\n...\r\n\r\n('tokens_trie', <transformers.tokenization_utils.Trie object at 0x7efc23dcce80>)\r\n```\r\n\r\n On every run of this the special tokens are being added on, and the hash is different on the `tokens_trie`. The increase in the special tokens category could be cleaned, but not sure about the hash for the `tokens_trie`. What might work is that the call for the tokenizer encoding can be translated into a function that strips any unnecessary information out, but that's a guess.\r\n", "Thanks for investigating ! Does that mean that `save_pretrained`() produces non-deterministic tokenizers on disk ? Or is it `from_pretrained()` which is not deterministic given the same files on disk ?\r\n\r\nI think one way to fix this would be to make save/from_pretrained deterministic, or make the pickling of `transformers.tokenization_utils.Trie` objects deterministic (this could be implemented in `transformers`, but maybe let's discuss in an issue in `transformers` before opening a PR)", "Late to the party but everything should be deterministic (afaik at least).\r\n\r\nBut `Trie` is a simple class object, so afaik it's hash function is linked to its `id(self)` so basically where it's stored in memory, so super highly non deterministic. Could that be the issue ?", "> But Trie is a simple class object, so afaik it's hash function is linked to its id(self) so basically where it's stored in memory, so super highly non deterministic. Could that be the issue ?\r\n\r\nWe're computing the hash of the pickle dump of the class so it should be fine, as long as the pickle dump is deterministic", "I've ported wav2vec2.0 fine-tuning into Optimum-Graphcore which is where I found the issue. The majority of the script was copied from the Transformers version to keep it similar, [here is the tokenizer loading section from the source](https://github.com/huggingface/transformers/blob/f0982682bd6fd0b438dda79ec45f3a8fac83a985/examples/pytorch/speech-recognition/run_speech_recognition_ctc.py#L531).\r\n\r\nIn the last comment I have two loaded tokenizers, one from run 'N' of the script and one from 'N+1'. I think what's happening is that when you add special tokens (e.g. PAD and UNK) another AddedToken object is appended when tokenizer is saved regardless of whether special tokens are there already. \r\n\r\nIf there is a AddedTokens cleanup at load/save this could solve the issue, but then is Trie going to cause hash to be different? I'm not sure. ", "Which Python version are you using ?\r\n\r\nThe trie is basically a big dict of dics, so deterministic nature depends on python version:\r\nhttps://stackoverflow.com/questions/2053021/is-the-order-of-a-python-dictionary-guaranteed-over-iterations\r\n\r\nMaybe the investigation is actually not finding the right culprit though (the memory id is changed, but `datasets` is not using that to compare, so maybe we need to be looking within `datasets` so see where the comparison fails)", "Similar issue found on `BartTokenizer`. You can bypass the bug by loading a fresh new tokenizer everytime.\r\n\r\n```\r\n dataset = dataset.map(lambda x: tokenize_func(x, BartTokenizer.from_pretrained(xxx)),\r\n num_proc=num_proc, desc='Tokenize')\r\n```" ]

[]

[ "Thanks for approving @lhoestq!\r\nWill apply these changes for the other datasets I've added too." ]

[]

[ "This was fixed in #740 \r\nCould you try to update `datasets` and try again ?", "This has been fixed in datasets 1.1.3" ]

[ { "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": "fef2c0", "default": false, "description": "", "id": 3287858981, "name": "streaming", "node_id": "MDU6TGFiZWwzMjg3ODU4OTgx", "url": "https://api.github.com/repos/huggingface/datasets/labels/streaming" } ]

[]

[ { "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" }, { "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" }, { "color": "DF8D62", "default": false, "description": "", "id": 4614514401, "name": "hacktoberfest", "node_id": "LA_kwDODunzps8AAAABEwvm4Q", "url": "https://api.github.com/repos/huggingface/datasets/labels/hacktoberfest" } ]

[ "Hi, thanks for reporting! This line should be replaced with \r\n```python\r\ndataset = dataset.map(lambda examples: tokenizer(examples[\"text\"], return_tensors=\"np\"), batched=True)\r\n```\r\nfor it to work (the `return_tensors` part inside the `tokenizer` call).", "Can I work on this?", "Fixed in https://github.com/huggingface/datasets/pull/5095" ]

[]

[]

[]

[ "<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) | 0.006616 / 0.011353 (-0.004737) | 0.003915 / 0.011008 (-0.007093) | 0.083271 / 0.038508 (0.044763) | 0.072595 / 0.023109 (0.049485) | 0.307224 / 0.275898 (0.031326) | 0.337244 / 0.323480 (0.013764) | 0.005296 / 0.007986 (-0.002690) | 0.003325 / 0.004328 (-0.001003) | 0.064589 / 0.004250 (0.060339) | 0.056369 / 0.037052 (0.019316) | 0.310829 / 0.258489 (0.052340) | 0.345563 / 0.293841 (0.051722) | 0.030551 / 0.128546 (-0.097995) | 0.008519 / 0.075646 (-0.067127) | 0.286368 / 0.419271 (-0.132903) | 0.052498 / 0.043533 (0.008966) | 0.308735 / 0.255139 (0.053596) | 0.329234 / 0.283200 (0.046034) | 0.022588 / 0.141683 (-0.119095) | 1.453135 / 1.452155 (0.000981) | 1.525956 / 1.492716 (0.033239) |\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) | 0.199417 / 0.018006 (0.181410) | 0.454621 / 0.000490 (0.454131) | 0.004928 / 0.000200 (0.004728) | 0.000079 / 0.000054 (0.000025) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.028436 / 0.037411 (-0.008975) | 0.083722 / 0.014526 (0.069196) | 0.095162 / 0.176557 (-0.081395) | 0.153434 / 0.737135 (-0.583702) | 0.099480 / 0.296338 (-0.196859) |\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) | 0.384647 / 0.215209 (0.169438) | 3.838406 / 2.077655 (1.760751) | 1.891267 / 1.504120 (0.387148) | 1.751432 / 1.541195 (0.210238) | 1.737443 / 1.468490 (0.268953) | 0.487758 / 4.584777 (-4.097019) | 3.635925 / 3.745712 (-0.109787) | 5.208718 / 5.269862 (-0.061144) | 3.029374 / 4.565676 (-1.536302) | 0.057613 / 0.424275 (-0.366662) | 0.007177 / 0.007607 (-0.000430) | 0.455596 / 0.226044 (0.229552) | 4.559969 / 2.268929 (2.291040) | 2.325321 / 55.444624 (-53.119303) | 2.034924 / 6.876477 (-4.841552) | 2.163869 / 2.142072 (0.021796) | 0.583477 / 4.805227 (-4.221750) | 0.132870 / 6.500664 (-6.367795) | 0.059618 / 0.075469 (-0.015851) |\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) | 1.263751 / 1.841788 (-0.578037) | 19.740004 / 8.074308 (11.665696) | 14.410980 / 10.191392 (4.219588) | 0.170367 / 0.680424 (-0.510057) | 0.018225 / 0.534201 (-0.515976) | 0.390101 / 0.579283 (-0.189182) | 0.404298 / 0.434364 (-0.030066) | 0.455295 / 0.540337 (-0.085043) | 0.621179 / 1.386936 (-0.765757) |\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) | 0.006580 / 0.011353 (-0.004773) | 0.004078 / 0.011008 (-0.006930) | 0.065842 / 0.038508 (0.027334) | 0.074494 / 0.023109 (0.051385) | 0.403644 / 0.275898 (0.127746) | 0.430204 / 0.323480 (0.106724) | 0.005343 / 0.007986 (-0.002643) | 0.003366 / 0.004328 (-0.000963) | 0.064858 / 0.004250 (0.060607) | 0.056252 / 0.037052 (0.019200) | 0.412556 / 0.258489 (0.154067) | 0.434099 / 0.293841 (0.140258) | 0.031518 / 0.128546 (-0.097028) | 0.008543 / 0.075646 (-0.067104) | 0.071658 / 0.419271 (-0.347613) | 0.049962 / 0.043533 (0.006430) | 0.398511 / 0.255139 (0.143372) | 0.415908 / 0.283200 (0.132708) | 0.025011 / 0.141683 (-0.116672) | 1.492350 / 1.452155 (0.040195) | 1.552996 / 1.492716 (0.060280) |\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) | 0.204971 / 0.018006 (0.186964) | 0.439965 / 0.000490 (0.439475) | 0.002071 / 0.000200 (0.001872) | 0.000084 / 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) | 0.031673 / 0.037411 (-0.005738) | 0.087529 / 0.014526 (0.073004) | 0.099882 / 0.176557 (-0.076675) | 0.156994 / 0.737135 (-0.580141) | 0.101421 / 0.296338 (-0.194918) |\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) | 0.407480 / 0.215209 (0.192271) | 4.069123 / 2.077655 (1.991468) | 2.081288 / 1.504120 (0.577169) | 1.920367 / 1.541195 (0.379172) | 1.981053 / 1.468490 (0.512563) | 0.481995 / 4.584777 (-4.102782) | 3.546486 / 3.745712 (-0.199226) | 5.133150 / 5.269862 (-0.136712) | 3.056444 / 4.565676 (-1.509232) | 0.056650 / 0.424275 (-0.367625) | 0.007746 / 0.007607 (0.000139) | 0.490891 / 0.226044 (0.264847) | 4.902160 / 2.268929 (2.633232) | 2.564726 / 55.444624 (-52.879899) | 2.234988 / 6.876477 (-4.641489) | 2.387656 / 2.142072 (0.245583) | 0.576315 / 4.805227 (-4.228912) | 0.132065 / 6.500664 (-6.368599) | 0.060728 / 0.075469 (-0.014741) |\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) | 1.370568 / 1.841788 (-0.471220) | 19.883159 / 8.074308 (11.808851) | 14.442066 / 10.191392 (4.250674) | 0.150119 / 0.680424 (-0.530305) | 0.018359 / 0.534201 (-0.515842) | 0.394128 / 0.579283 (-0.185155) | 0.411697 / 0.434364 (-0.022667) | 0.460580 / 0.540337 (-0.079757) | 0.608490 / 1.386936 (-0.778446) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#035d0cf842b82b14059999baa78e8d158dfbed12 \"CML watermark\")\n", "_The documentation is not available anymore as the PR was closed or merged._", "merging now if you don't mind - this way I can make a patch release" ]

[]

[ "This is the test set given by the Glue benchmark. The labels are not provided, and therefore set to -1.", "Ah, yeah. Why it didn’t occur to me. 😂\nThank you for your comment." ]

[ { "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! `git bisect` points to #2986 as the PR that introduced the bug. Since then, there have been some additional changes to the loading logic, and in the current state, `force_local_path` (set via `local_path`) forbids pulling a script from the internet instead of downloading it: https://github.com/huggingface/datasets/blob/cfae0545b2ba05452e16136cacc7d370b4b186a1/src/datasets/inspect.py#L89-L91\r\n\r\ncc @lhoestq: `force_local_path` is only used in `inspect_dataset` and `inspect_metric`. Is it OK if we revert the behavior to match the old one?", "Good catch ! Yea I think it's fine :)" ]

[ { "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,\r\n\r\nyou can suppress progress bars by patching logging as follows:\r\n```python\r\nimport datasets\r\nimport logging\r\ndatasets.logging.get_verbosity = lambda: logging.NOTSET\r\n# map call ...\r\n```\r\nEDIT: now you have to use `disable_progress_bar `", "Thank you, it worked :)", "See https://github.com/huggingface/datasets/issues/2528 for reference", "Note also that you can disable the progress bar with\r\n\r\n```python\r\nfrom datasets.utils import disable_progress_bar\r\ndisable_progress_bar()\r\n```\r\n\r\nSee https://github.com/huggingface/datasets/blob/8814b393984c1c2e1800ba370de2a9f7c8644908/src/datasets/utils/tqdm_utils.py#L84", "Now the library officially recommends `set_progress_bar_enabled(False)`\r\n\r\n```py\r\nfrom datasets.utils import set_progress_bar_enabled\r\n\r\nset_progress_bar_enabled(False)\r\n```\r\n\r\nsource:\r\n\r\nhttps://github.com/huggingface/datasets/blob/1fd47120ace13626c528367787ffa13e1a26e6c0/src/datasets/utils/tqdm_utils.py#L83-L88\r\n\r\n", "From https://github.com/huggingface/datasets/pull/3897, `disable_progress_bar` is the function you should use", "Now ``disable_progress_bar`` function is in ``datasets/src/datasets/utils/logging.py``.\r\nhttps://github.com/huggingface/datasets/blob/aa555a299ad73c65e3f997a764e9d211675ab05d/src/datasets/utils/logging.py#L233-L236\r\nAnd the method mentioned in https://github.com/huggingface/datasets/issues/2651#issuecomment-880270774 is not working now." ]

[]

[ "May eventually be solved in #5883 ", "#self-assign" ]

[ { "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!\r\n\r\nYou can also use names other than \"train\", \"validation\" and \"test\". As an example, check the [script](https://huggingface.co/datasets/mozilla-foundation/common_voice_11_0/blob/e095840f23f3dffc1056c078c2f9320dad9ca74d/common_voice_11_0.py#L139) of the Common Voice 11 dataset. " ]

[]

[]

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

[ "See notebook by @geohci: https://public.paws.wmcloud.org/User:Isaac_(WMF)/HuggingFace%20Wikisource%20Processing.ipynb" ]

[]

[ "Hi ! `ds.sort()` does sort the full dataset, not just one column:\r\n```python\r\nfrom datasets import *\r\n\r\nds = Dataset.from_dict({\"foo\": [3, 2, 1], \"bar\": [\"c\", \"b\", \"a\"]})\r\nprint(d.sort(\"foo\").to_pandas()\r\n# foo bar\r\n# 0 1 a\r\n# 1 2 b\r\n# 2 3 c\r\n```\r\n\r\nWhat made you think it was not the case ? Did you experience a situation where it was only sorting one column ?", "Hi! thank you for your quick reply!\r\nI wanted to sort the `cnn_dailymail` dataset by the length of the labels (num of characters). I added a new column to the dataset (`ds.add_column`) with the lengths and then sorted by this new column. Only the new length column was sorted, the reset left in their original order. ", "That's unexpected, can you share the code you used to get this ?" ]

[ { "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" } ]

[ "Even more complicating - \r\n\r\nAs I see it, datasets can have \"addons\".\r\nFor example, the WebNLG dataset is a dataset for data-to-text. However, a work of mine and other works enriched this dataset with text plans / underlying text structures. In that case, I see a need to load the dataset \"WebNLG\" with \"plans\" addon.\r\n\r\nSame for sign language - if there is a dataset of videos, one addon can be to run OpenPose, another to run ARKit4 pose estimation, and another to run PoseNet, or even just a video embedding addon. (which are expensive to run individually for everyone who wants to use these data)\r\n\r\nThis is something I dabbled with my own implementation to a [research datasets library](https://github.com/AmitMY/meta-scholar/) and I love to get the discussion going on these topics.", "This is a really cool idea !\r\nThe example for data objects you gave for the RWTH-PHOENIX-Weather 2014 T dataset can totally fit inside the library.\r\n\r\nFor your point about formats like `ilex`, `eaf`, or `srt`, it is possible to use any library in your dataset script.\r\nHowever most user probably won't need these libraries, as most datasets don't need them, and therefore it's unlikely that we will have them in the minimum requirements to use `nlp` (we want to keep it as light-weight as possible). If a user wants to load your dataset and doesn't have the libraries you need, an error is raised asking the user to install them.\r\n\r\nMore generally, we plan to have something like a `requirements.txt` per dataset. This could also be a place for addons as you said. What do you think ?", "Thanks, Quentin, I think a `requirements.txt` per dataset will be a good thing.\r\nI will work on adding this dataset next week, and once we sort all of the kinks, I'll add more." ]

[]

[ "Feel free to merge once the CI is all green ;)" ]

[]

[ "> Nice thanks !\r\n> \r\n> After your changes feel free to mark this PR as \"ready for review\" ;)\r\n\r\nMarking PR ready for review.\r\n\r\n@lhoestq Let me know if there is anything else required or if we are good to go ahead and merge.", "_The documentation is not available anymore as the PR was closed or merged._" ]

[]

[]

[]

[ "It not always practical to use nested `Sequence`. If you have troubles with sequence you can use lists instead. \r\n\r\nFor example\r\n```python\r\n\r\nfeatures=datasets.Features(\r\n {\r\n \"dial_id\": datasets.Value(\"string\"),\r\n \"doc_id\": datasets.Value(\"string\"),\r\n \"domain\": datasets.Value(\"string\"),\r\n \"turns\": [\r\n {\r\n \"turn_id\": datasets.Value(\"int32\"),\r\n \"role\": datasets.Value(\"string\"),\r\n \"da\": datasets.Value(\"string\"),\r\n \"reference\": [\r\n {\r\n \"keys\" : datasets.Value(\"string\"),\r\n \"values\": datasets.Value(\"string\"), \r\n }\r\n\r\n ],\r\n \"utterance\": datasets.Value(\"string\"),\r\n }\r\n ],\r\n }\r\n),\r\n```\r\n\r\nthis way `turns` will be a list of dict, and the \"reference\" key of `turns` will be a list of dict as well", "No problem thanks for all your help getting this to the final stages! Added .gitignore, removed .lock and applied the changes you asked for." ]

[]

[ "Duplicated issue." ]

[ { "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" } ]

[]

[]

[ "We can probably scan the dataset scripts with a regexpr to try to identify this pattern cc @patrickvonplaten maybe", "@mariamabarham This script should work. I tested it for a couple of datasets. There might be exceptions where the script breaks - did not test everything.\r\n\r\n```python\r\n#!/usr/bin/env python3\r\n\r\nfrom nlp import prepare_module, DownloadConfig, import_main_class, hf_api\r\nimport tempfile\r\n\r\n\r\ndef scan_for_nested_unnecessary_dict(dataset_name):\r\n\r\n def load_builder_class(dataset_name):\r\n module_path = prepare_module(dataset_name, download_config=DownloadConfig(force_download=True))\r\n return import_main_class(module_path)\r\n\r\n def load_configs(dataset_name):\r\n builder_cls = load_builder_class(dataset_name)\r\n if len(builder_cls.BUILDER_CONFIGS) == 0:\r\n return [None]\r\n return builder_cls.BUILDER_CONFIGS\r\n\r\n def scan_features_for_nested_dict(features):\r\n is_sequence = False\r\n if hasattr(features, \"_type\"):\r\n if features._type != 'Sequence':\r\n return False\r\n else:\r\n is_sequence = True\r\n features = features.feature\r\n\r\n if isinstance(features, list):\r\n for value in features:\r\n if scan_features_for_nested_dict(value):\r\n return True\r\n return False\r\n\r\n elif isinstance(features, dict):\r\n for key, value in features.items():\r\n if is_sequence and len(features.keys()) == 1 and hasattr(features[key], \"_type\") and features[key]._type != \"Sequence\":\r\n return True\r\n if scan_features_for_nested_dict(value):\r\n return True\r\n return False\r\n elif hasattr(features, \"_type\"):\r\n return False\r\n else:\r\n raise ValueError(f\"{features} should be either a list, a dict or a feature\")\r\n\r\n configs = load_configs(dataset_name)\r\n\r\n for config in configs:\r\n with tempfile.TemporaryDirectory() as processed_temp_dir:\r\n # create config and dataset\r\n dataset_builder_cls = load_builder_class(dataset_name)\r\n name = config.name if config is not None else None\r\n dataset_builder = dataset_builder_cls(name=name, cache_dir=processed_temp_dir)\r\n\r\n is_nested_dict_in_dataset = scan_features_for_nested_dict(dataset_builder._info().features)\r\n if is_nested_dict_in_dataset:\r\n print(f\"{dataset_name} with {name} needs refactoring\")\r\n\r\n\r\nif __name__ == \"__main__\":\r\n scan_for_nested_unnecessary_dict(\"race\") # prints True\r\n scan_for_nested_unnecessary_dict(\"mlqa\") # prints True\r\n scan_for_nested_unnecessary_dict(\"squad\") # prints Nothing\r\n\r\n # ran the following lines for 1min and seems to work -> didn't check for all datasets though\r\n# api = hf_api.HfApi()\r\n# all_datasets = [x.id for x in api.dataset_list(with_community_datasets=False)]\r\n# for dataset in all_datasets:\r\n# scan_for_nested_unnecessary_dict(dataset)\r\n```", "> @mariamabarham This script should work. I tested it for a couple of datasets. There might be exceptions where the script breaks - did not test everything.\r\n> \r\n> ```python\r\n> #!/usr/bin/env python3\r\n> \r\n> from nlp import prepare_module, DownloadConfig, import_main_class, hf_api\r\n> import tempfile\r\n> \r\n> \r\n> def scan_for_nested_unnecessary_dict(dataset_name):\r\n> \r\n> def load_builder_class(dataset_name):\r\n> module_path = prepare_module(dataset_name, download_config=DownloadConfig(force_download=True))\r\n> return import_main_class(module_path)\r\n> \r\n> def load_configs(dataset_name):\r\n> builder_cls = load_builder_class(dataset_name)\r\n> if len(builder_cls.BUILDER_CONFIGS) == 0:\r\n> return [None]\r\n> return builder_cls.BUILDER_CONFIGS\r\n> \r\n> def scan_features_for_nested_dict(features):\r\n> is_sequence = False\r\n> if hasattr(features, \"_type\"):\r\n> if features._type != 'Sequence':\r\n> return False\r\n> else:\r\n> is_sequence = True\r\n> features = features.feature\r\n> \r\n> if isinstance(features, list):\r\n> for value in features:\r\n> if scan_features_for_nested_dict(value):\r\n> return True\r\n> return False\r\n> \r\n> elif isinstance(features, dict):\r\n> for key, value in features.items():\r\n> if is_sequence and len(features.keys()) == 1 and hasattr(features[key], \"_type\") and features[key]._type != \"Sequence\":\r\n> return True\r\n> if scan_features_for_nested_dict(value):\r\n> return True\r\n> return False\r\n> else:\r\n> raise ValueError(f\"{features} should be either a list of a dict\")\r\n> \r\n> configs = load_configs(dataset_name)\r\n> \r\n> for config in configs:\r\n> with tempfile.TemporaryDirectory() as processed_temp_dir:\r\n> # create config and dataset\r\n> dataset_builder_cls = load_builder_class(dataset_name)\r\n> name = config.name if config is not None else None\r\n> dataset_builder = dataset_builder_cls(name=name, cache_dir=processed_temp_dir)\r\n> \r\n> is_nested_dict_in_dataset = scan_features_for_nested_dict(dataset_builder._info().features)\r\n> if is_nested_dict_in_dataset:\r\n> print(f\"{dataset_name} with {name} needs refactoring\")\r\n> \r\n> \r\n> if __name__ == \"__main__\":\r\n> scan_for_nested_unnecessary_dict(\"race\") # prints True\r\n> scan_for_nested_unnecessary_dict(\"mlqa\") # prints True\r\n> scan_for_nested_unnecessary_dict(\"squad\") # prints Nothing\r\n> \r\n> # ran the following lines for 1min and seems to work -> didn't check for all datasets though\r\n> # api = hf_api.HfApi()\r\n> # all_datasets = [x.id for x in api.dataset_list(with_community_datasets=False)]\r\n> # for dataset in all_datasets:\r\n> # scan_for_nested_unnecessary_dict(dataset)\r\n> ```\r\n\r\nGreat, I will try it", "I'm not sure the work on this PR was finished @lhoestq cc @mariamabarham @patrickvonplaten ", "Sorry for that, apparently there are other datasets that could have unnecessary nested dicts.\r\nWe can have another PR to scan and fix the other datasets.\r\n" ]

[]

[]

[]

[]

[]

[]

[]

[ "Update:\r\n\r\nThe imdb download completed after a long time (about 45 mins). Ofcourse once download loading was instantaneous. Also, the loaded object was of type `arrow_dataset`. \r\n\r\nThe urls for glue still doesn't work though.", "Thanks for the report, I'll give a look!", "I am also seeing a similar error when running the following:\r\n\r\n```\r\nimport nlp\r\ndataset = load_dataset('cola')\r\n```\r\nError:\r\n```\r\nTraceback (most recent call last):\r\n File \"<stdin>\", line 1, in <module>\r\n File \"/home/js11133/.conda/envs/jiant/lib/python3.8/site-packages/nlp/load.py\", line 509, in load_dataset\r\n module_path = prepare_module(path, download_config=download_config, dataset=True)\r\n File \"/home/js11133/.conda/envs/jiant/lib/python3.8/site-packages/nlp/load.py\", line 248, in prepare_module\r\n local_path = cached_path(file_path, download_config=download_config)\r\n File \"/home/js11133/.conda/envs/jiant/lib/python3.8/site-packages/nlp/utils/file_utils.py\", line 191, in cached_path\r\n output_path = get_from_cache(\r\n File \"/home/js11133/.conda/envs/jiant/lib/python3.8/site-packages/nlp/utils/file_utils.py\", line 356, in get_from_cache\r\n raise ConnectionError(\"Couldn't reach {}\".format(url))\r\nConnectionError: Couldn't reach https://s3.amazonaws.com/datasets.huggingface.co/nlp/datasets/cola/cola.py\r\n```", "@jeswan `\"cola\"` is not a valid dataset identifier (you can check the up-to-date list on https://huggingface.co/datasets) but you can find cola inside glue.", "Ah right. Thanks!", "Hi. Closing this one since #626 updated the glue urls.\r\n\r\n> 1. Why is it still blocking? Is it still downloading?\r\n\r\nAfter downloading it generates the arrow file by iterating through the examples.\r\nThe number of examples processed by second is shown during the processing (not sure why it was not the case for you)\r\n\r\n> 2. I specified split as train, so why is the test folder being populated?\r\n\r\nIt downloads every split\r\n\r\n\r\n\r\n" ]

[ { "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": "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" } ]

[ "Assigned to myself and will get to it in the next week, but if someone finds this issue annoying and wants to submit a PR before I do, just ping me here and I'll reassign :). ", "I would like to be assigned to this issue, @nateraw . #self-assign" ]

[]

[]

[]

[ "Looks all good :)\r\nI rebased from master and added a test for parallel `map_nested`", "you're da best" ]

[]

[ "Thanks for reporting, @lsb. \r\n\r\nWe are investigating it.\r\n\r\nOn the other hand, please note that in the next `datasets` release, the index will not be created by default (see #5583). If you would like to have it, you will need to explicitly pass `index=True`. ", "I think this is low enough priority for me to close this as Won't Fix. If I need any primary keys I can generate them beforehand. Feel free to reopen." ]

[]

[]

[]

[]

[]

[ "@lhoestq, lütfen bu PR'ı gözden geçirebilir misiniz?", "> Thanks ! This looks all good now :)\r\n\r\nThanks" ]

[]

[ "Hi! This behavior stems from these lines:\r\n\r\nhttps://github.com/huggingface/datasets/blob/b065547654efa0ec633cf373ac1512884c68b2e1/src/datasets/formatting/np_formatter.py#L45-L46\r\n\r\nI agree we should preserve the original type whenever possible and downcast explicitly with a warning.\r\n\r\n@lhoestq Do you remember why we need this \"default dtype\" logic in our formatters?", "I was also wondering why the default type logic is needed. Me just deleting it is probably too naive of a solution.", "Hmm I think the idea was to end up with the usual default precision for deep learning models - no matter how the data was stored or where it comes from.\r\n\r\nFor example in NLP we store tokens using an optimized low precision to save disk space, but when we set the format to `torch` we actually need to get `int64`. Although the need for a default for integers also comes from numpy not returning the same integer precision depending on your machine. Finally I guess we added a default for floats as well for consistency.\r\n\r\nI'm a bit embarrassed by this though, as a user I'd have expected to get the same precision indeed as well and get a zero copy view.", "Will you fix this or should I open a PR?", "Unfortunately removing it for integers is a breaking change for most `transformers` + `datasets` users for NLP (which is a common case). Removing it for floats is a breaking change for `transformers` + `datasets` for ASR as well. And it also is a breaking change for the other users relying on this behavior.\r\n\r\nTherefore I think that the only short term solution is for the user to provide `dtype=` manually and document better this behavior. We could also extend `dtype` to accept a value that means \"return the same dtype as the underlying storage\" and make it easier to do zero copy.", "@lhoestq It should be fine to remove this conversion in Datasets 3.0, no? For now, we can warn the user (with a log message) about the future change when the default type is changed.", "Let's see with the transformers team if it sounds reasonable ? We'd have to fix multiple example scripts though.\r\n\r\nIf it's not ok we can also explore keeping this behavior only for tokens and audio data.", "IMO being coupled with Transformers can lead to unexpected behavior when one tries to use our lib without pairing it with Transformers, so I think it's still important to \"fix\" this, even if it means we will need to update Transformers' example scripts afterward.\r\n", "Ideally let's update the `transformers` example scripts before the change :P", "For others that run into the same issue: A temporary workaround for me is this:\r\n```python\r\ndef numpy_transform(batch):\r\n return {key: np.asarray(val) for key, val in batch.items()}\r\n\r\ndataset = dataset.with_transform(numpy_transform)\r\n```" ]

[]

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

[]

[ "> Hi @tianjianjiang, thanks for the fix.\r\n> The links should also be updated in the `dataset_infos.json` file.\r\n> The failing tests are due to the missing tag in the header of the `README.md` file:\r\n\r\nHi @albertvillanova, thank you for the info! My apologies for the messy PR.\r\n" ]

[]

[]

[]

[ "Hi ! Thanks for the fix :)\r\n\r\nIf I understand correctly, `OptionalSequence` doesn't have an associated arrow type that we know in advance unlike the other feature types, because it depends on the type of the examples.\r\n\r\nFor example, I tested this and it raises an error:\r\n```python\r\nimport datasets as ds\r\nimport pyarrow as pa\r\n\r\nfeatures = ds.Features({\"a\": ds.features.OptionalSequence(ds.Value(\"int32\"))})\r\nbatch = {\"a\": [[0]]}\r\n\r\nwriter = ds.ArrowWriter(features=features, stream=pa.BufferOutputStream())\r\nwriter.write_batch(batch)\r\n# ArrowInvalid: Could not convert [0] with type list: tried to convert to int\r\n```\r\nThis error happens because `features.type` is `StructType(struct<a: int32>)`.\r\n\r\nAnother way to add support for multilabel would be to have several configurations for these metrics. By default it would set the features without sequences, and for the multi label configuration it would use features with sequences. Let me know what you think", "Hi @lhoestq, thanks for your feedback :)\r\n\r\nDefinitely, your suggested approach is simpler. I am going to refactor all my PR unless we could envision some other use cases where an OptionalSequence might be convenient, but for now I can't think of any...", "@albertvillanova @lhoestq I couldnt find the related docs in F1 card: https://huggingface.co/spaces/evaluate-metric/f1\r\n\r\nHow do I perform multilabel F1 evaluation using evaluate package?", "I was going to transfer your question to the `evaluate` GitHub repository, but I saw you have already done it (and even opened a PR):\r\n- https://github.com/huggingface/evaluate/issues/219\r\n- https://github.com/huggingface/evaluate/pull/221\r\n\r\nThanks, @fcakyon. ", "Sorry to bomb you on multiple channels :sweat_smile: @albertvillanova, I have solved my problems, and opened a PR so that others also don't get confused :+1: " ]

[]

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

[]

[ "Hi ! Can you \r\n```python\r\nimport platform\r\nprint(platform.python_version())\r\n```\r\nto see that it returns ?", "Hi,\n\n3.8.13\n\nGet Outlook for Android<https://aka.ms/AAb9ysg>\n________________________________\nFrom: Quentin Lhoest ***@***.***>\nSent: Tuesday, November 22, 2022 2:37:02 PM\nTo: huggingface/datasets ***@***.***>\nCc: feketedavid1012 ***@***.***>; Author ***@***.***>\nSubject: Re: [huggingface/datasets] Import error (Issue #5280)\n\n\nHi ! Can you\n\nimport platform\nprint(platform.python_version())\n\nto see that it returns ?\n\n—\nReply to this email directly, view it on GitHub<https://github.com/huggingface/datasets/issues/5280#issuecomment-1323691385>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AJW7F5YGG32W6WABYC25NJTWJTD75ANCNFSM6AAAAAASHZJ2AU>.\nYou are receiving this because you authored the thread.Message ID: ***@***.***>\n", "Then it should work as expected if you use the same python when using `datasets`\r\n\r\nPlease make sure you're running your code in the right environment", "It's the right environment. But in if statement I have\n\"3.8.13\" < 3.7\nAnd in the error message is Python>=3.7 which is true in my case (3.8.13 is greater then 3.7), so I don't understand my python should be below the 3.7 which case the if statement is right, but the message is wrong, or above 3.7 which case if statement is wrong, error message is right.\n\nGet Outlook for Android<https://aka.ms/AAb9ysg>\n________________________________\nFrom: Quentin Lhoest ***@***.***>\nSent: Tuesday, November 22, 2022 2:41:43 PM\nTo: huggingface/datasets ***@***.***>\nCc: feketedavid1012 ***@***.***>; Author ***@***.***>\nSubject: Re: [huggingface/datasets] Import error (Issue #5280)\n\n\nThen it should work as expected if you use the same python when using datasets\n\nPlease make sure you're running your code in the right environment\n\n—\nReply to this email directly, view it on GitHub<https://github.com/huggingface/datasets/issues/5280#issuecomment-1323697094>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/AJW7F54JURTAJJWWDO2QGI3WJTERPANCNFSM6AAAAAASHZJ2AU>.\nYou are receiving this because you authored the thread.Message ID: ***@***.***>\n", "If you're having an error then you're not running your code in the right environment." ]

[]

[ "Build fails but this change should not be the reason...", "rebased on master" ]

[]

[]

[]

[]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "Good catch ! Sorry I forgot (again) about windows paths when writing this x)" ]

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

[ "Datasets are not tracked in this repository anymore." ]

[]

[ "Close #1872" ]

[]

[ "Works smoothly with the `transformers.Trainer` class now, thank you!" ]

[]

[ "_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) | 0.008272 / 0.011353 (-0.003081) | 0.004494 / 0.011008 (-0.006515) | 0.100764 / 0.038508 (0.062256) | 0.028741 / 0.023109 (0.005632) | 0.309020 / 0.275898 (0.033122) | 0.354184 / 0.323480 (0.030704) | 0.007455 / 0.007986 (-0.000531) | 0.003377 / 0.004328 (-0.000951) | 0.078472 / 0.004250 (0.074222) | 0.034719 / 0.037052 (-0.002333) | 0.312787 / 0.258489 (0.054298) | 0.342878 / 0.293841 (0.049037) | 0.033326 / 0.128546 (-0.095221) | 0.011519 / 0.075646 (-0.064127) | 0.323556 / 0.419271 (-0.095716) | 0.039929 / 0.043533 (-0.003604) | 0.304627 / 0.255139 (0.049488) | 0.322876 / 0.283200 (0.039677) | 0.086410 / 0.141683 (-0.055273) | 1.502607 / 1.452155 (0.050453) | 1.577953 / 1.492716 (0.085237) |\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) | 0.192861 / 0.018006 (0.174855) | 0.406008 / 0.000490 (0.405519) | 0.001075 / 0.000200 (0.000875) | 0.000071 / 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) | 0.023351 / 0.037411 (-0.014060) | 0.096086 / 0.014526 (0.081561) | 0.104641 / 0.176557 (-0.071915) | 0.141940 / 0.737135 (-0.595195) | 0.109266 / 0.296338 (-0.187073) |\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) | 0.416496 / 0.215209 (0.201287) | 4.161581 / 2.077655 (2.083926) | 1.815357 / 1.504120 (0.311238) | 1.609536 / 1.541195 (0.068341) | 1.654105 / 1.468490 (0.185615) | 0.693947 / 4.584777 (-3.890830) | 3.349029 / 3.745712 (-0.396683) | 1.883968 / 5.269862 (-3.385893) | 1.287988 / 4.565676 (-3.277688) | 0.081765 / 0.424275 (-0.342511) | 0.012373 / 0.007607 (0.004766) | 0.517186 / 0.226044 (0.291142) | 5.200892 / 2.268929 (2.931964) | 2.247414 / 55.444624 (-53.197211) | 1.910601 / 6.876477 (-4.965876) | 1.965407 / 2.142072 (-0.176666) | 0.814386 / 4.805227 (-3.990841) | 0.149295 / 6.500664 (-6.351369) | 0.064667 / 0.075469 (-0.010802) |\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) | 1.247258 / 1.841788 (-0.594530) | 13.837355 / 8.074308 (5.763047) | 13.850454 / 10.191392 (3.659062) | 0.136078 / 0.680424 (-0.544346) | 0.028322 / 0.534201 (-0.505878) | 0.391394 / 0.579283 (-0.187889) | 0.407494 / 0.434364 (-0.026870) | 0.473784 / 0.540337 (-0.066554) | 0.562953 / 1.386936 (-0.823983) |\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) | 0.006559 / 0.011353 (-0.004794) | 0.004546 / 0.011008 (-0.006462) | 0.099527 / 0.038508 (0.061019) | 0.027428 / 0.023109 (0.004319) | 0.344276 / 0.275898 (0.068377) | 0.377897 / 0.323480 (0.054417) | 0.004913 / 0.007986 (-0.003072) | 0.003338 / 0.004328 (-0.000990) | 0.077589 / 0.004250 (0.073339) | 0.038819 / 0.037052 (0.001766) | 0.343165 / 0.258489 (0.084676) | 0.386228 / 0.293841 (0.092387) | 0.031753 / 0.128546 (-0.096794) | 0.011756 / 0.075646 (-0.063890) | 0.322537 / 0.419271 (-0.096735) | 0.049865 / 0.043533 (0.006332) | 0.340493 / 0.255139 (0.085354) | 0.372179 / 0.283200 (0.088980) | 0.099669 / 0.141683 (-0.042013) | 1.487841 / 1.452155 (0.035686) | 1.527400 / 1.492716 (0.034683) |\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) | 0.180782 / 0.018006 (0.162776) | 0.393494 / 0.000490 (0.393004) | 0.003004 / 0.000200 (0.002804) | 0.000076 / 0.000054 (0.000022) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024997 / 0.037411 (-0.012415) | 0.098232 / 0.014526 (0.083707) | 0.107869 / 0.176557 (-0.068688) | 0.141042 / 0.737135 (-0.596093) | 0.109551 / 0.296338 (-0.186787) |\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) | 0.477115 / 0.215209 (0.261906) | 4.783928 / 2.077655 (2.706273) | 2.435725 / 1.504120 (0.931605) | 2.233111 / 1.541195 (0.691916) | 2.341097 / 1.468490 (0.872607) | 0.694304 / 4.584777 (-3.890473) | 3.345687 / 3.745712 (-0.400025) | 1.886932 / 5.269862 (-3.382929) | 1.155585 / 4.565676 (-3.410092) | 0.082867 / 0.424275 (-0.341408) | 0.012420 / 0.007607 (0.004813) | 0.576575 / 0.226044 (0.350530) | 5.777691 / 2.268929 (3.508762) | 2.882219 / 55.444624 (-52.562405) | 2.543613 / 6.876477 (-4.332864) | 2.578939 / 2.142072 (0.436866) | 0.803143 / 4.805227 (-4.002084) | 0.151929 / 6.500664 (-6.348735) | 0.067777 / 0.075469 (-0.007693) |\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) | 1.282711 / 1.841788 (-0.559077) | 13.942771 / 8.074308 (5.868463) | 13.376206 / 10.191392 (3.184814) | 0.152916 / 0.680424 (-0.527508) | 0.016619 / 0.534201 (-0.517582) | 0.375141 / 0.579283 (-0.204142) | 0.381660 / 0.434364 (-0.052704) | 0.465090 / 0.540337 (-0.075247) | 0.555068 / 1.386936 (-0.831868) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#10a6a638e0feb955f7b607b4433ee715c30acccf \"CML watermark\")\n" ]

[]

[ "_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) | 0.006665 / 0.011353 (-0.004688) | 0.004376 / 0.011008 (-0.006633) | 0.085529 / 0.038508 (0.047021) | 0.076372 / 0.023109 (0.053263) | 0.310019 / 0.275898 (0.034121) | 0.341404 / 0.323480 (0.017924) | 0.005666 / 0.007986 (-0.002320) | 0.003763 / 0.004328 (-0.000566) | 0.064678 / 0.004250 (0.060427) | 0.059283 / 0.037052 (0.022231) | 0.316194 / 0.258489 (0.057704) | 0.349397 / 0.293841 (0.055557) | 0.031199 / 0.128546 (-0.097347) | 0.008724 / 0.075646 (-0.066923) | 0.300236 / 0.419271 (-0.119035) | 0.068872 / 0.043533 (0.025339) | 0.308521 / 0.255139 (0.053382) | 0.331292 / 0.283200 (0.048092) | 0.028236 / 0.141683 (-0.113447) | 1.501365 / 1.452155 (0.049211) | 1.554334 / 1.492716 (0.061618) |\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) | 0.238291 / 0.018006 (0.220285) | 0.565069 / 0.000490 (0.564580) | 0.001626 / 0.000200 (0.001426) | 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) | 0.029777 / 0.037411 (-0.007634) | 0.082873 / 0.014526 (0.068347) | 0.099619 / 0.176557 (-0.076937) | 0.156572 / 0.737135 (-0.580563) | 0.099887 / 0.296338 (-0.196452) |\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) | 0.401017 / 0.215209 (0.185808) | 3.827192 / 2.077655 (1.749537) | 1.861554 / 1.504120 (0.357434) | 1.699869 / 1.541195 (0.158674) | 1.720043 / 1.468490 (0.251553) | 0.486757 / 4.584777 (-4.098020) | 3.638125 / 3.745712 (-0.107587) | 5.844959 / 5.269862 (0.575097) | 3.454901 / 4.565676 (-1.110775) | 0.057650 / 0.424275 (-0.366625) | 0.007341 / 0.007607 (-0.000266) | 0.462698 / 0.226044 (0.236654) | 4.633472 / 2.268929 (2.364544) | 2.287607 / 55.444624 (-53.157017) | 2.057318 / 6.876477 (-4.819159) | 2.203657 / 2.142072 (0.061584) | 0.598136 / 4.805227 (-4.207091) | 0.134012 / 6.500664 (-6.366653) | 0.060824 / 0.075469 (-0.014645) |\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) | 1.277752 / 1.841788 (-0.564036) | 20.013398 / 8.074308 (11.939089) | 14.372993 / 10.191392 (4.181601) | 0.169991 / 0.680424 (-0.510433) | 0.018344 / 0.534201 (-0.515857) | 0.396985 / 0.579283 (-0.182299) | 0.416289 / 0.434364 (-0.018075) | 0.458658 / 0.540337 (-0.081680) | 0.692980 / 1.386936 (-0.693956) |\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) | 0.006689 / 0.011353 (-0.004664) | 0.004393 / 0.011008 (-0.006615) | 0.064069 / 0.038508 (0.025561) | 0.080717 / 0.023109 (0.057607) | 0.370090 / 0.275898 (0.094191) | 0.400432 / 0.323480 (0.076952) | 0.005613 / 0.007986 (-0.002372) | 0.003641 / 0.004328 (-0.000687) | 0.064771 / 0.004250 (0.060520) | 0.057555 / 0.037052 (0.020502) | 0.392156 / 0.258489 (0.133667) | 0.409842 / 0.293841 (0.116001) | 0.031500 / 0.128546 (-0.097047) | 0.008786 / 0.075646 (-0.066860) | 0.070342 / 0.419271 (-0.348929) | 0.048646 / 0.043533 (0.005113) | 0.360914 / 0.255139 (0.105775) | 0.387626 / 0.283200 (0.104426) | 0.022787 / 0.141683 (-0.118896) | 1.508915 / 1.452155 (0.056761) | 1.539719 / 1.492716 (0.047002) |\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) | 0.257985 / 0.018006 (0.239979) | 0.550990 / 0.000490 (0.550501) | 0.000407 / 0.000200 (0.000207) | 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) | 0.030183 / 0.037411 (-0.007228) | 0.086882 / 0.014526 (0.072356) | 0.102382 / 0.176557 (-0.074175) | 0.154745 / 0.737135 (-0.582390) | 0.104008 / 0.296338 (-0.192331) |\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) | 0.426284 / 0.215209 (0.211075) | 4.240812 / 2.077655 (2.163158) | 2.261240 / 1.504120 (0.757120) | 2.085905 / 1.541195 (0.544710) | 2.160374 / 1.468490 (0.691883) | 0.481126 / 4.584777 (-4.103651) | 3.516234 / 3.745712 (-0.229478) | 3.325322 / 5.269862 (-1.944539) | 2.043307 / 4.565676 (-2.522369) | 0.056663 / 0.424275 (-0.367612) | 0.007786 / 0.007607 (0.000179) | 0.497614 / 0.226044 (0.271570) | 4.974529 / 2.268929 (2.705600) | 2.700018 / 55.444624 (-52.744606) | 2.393778 / 6.876477 (-4.482699) | 2.628202 / 2.142072 (0.486130) | 0.594316 / 4.805227 (-4.210911) | 0.147092 / 6.500664 (-6.353572) | 0.062207 / 0.075469 (-0.013262) |\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) | 1.315676 / 1.841788 (-0.526112) | 20.749251 / 8.074308 (12.674943) | 14.371553 / 10.191392 (4.180160) | 0.170249 / 0.680424 (-0.510175) | 0.018478 / 0.534201 (-0.515722) | 0.395710 / 0.579283 (-0.183573) | 0.409706 / 0.434364 (-0.024658) | 0.463454 / 0.540337 (-0.076884) | 0.615657 / 1.386936 (-0.771279) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#c5a752d8e8ca0a6ed118b024ba03c1b4a2881177 \"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) | 0.007224 / 0.011353 (-0.004129) | 0.004506 / 0.011008 (-0.006503) | 0.096729 / 0.038508 (0.058221) | 0.082394 / 0.023109 (0.059284) | 0.390954 / 0.275898 (0.115056) | 0.416647 / 0.323480 (0.093167) | 0.005894 / 0.007986 (-0.002092) | 0.003756 / 0.004328 (-0.000572) | 0.075800 / 0.004250 (0.071549) | 0.062683 / 0.037052 (0.025631) | 0.398959 / 0.258489 (0.140470) | 0.436624 / 0.293841 (0.142783) | 0.034650 / 0.128546 (-0.093896) | 0.009655 / 0.075646 (-0.065991) | 0.315761 / 0.419271 (-0.103511) | 0.060957 / 0.043533 (0.017424) | 0.385649 / 0.255139 (0.130510) | 0.394022 / 0.283200 (0.110822) | 0.024601 / 0.141683 (-0.117082) | 1.729586 / 1.452155 (0.277431) | 1.724153 / 1.492716 (0.231437) |\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) | 0.207070 / 0.018006 (0.189063) | 0.466502 / 0.000490 (0.466012) | 0.010739 / 0.000200 (0.010540) | 0.000214 / 0.000054 (0.000160) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.031633 / 0.037411 (-0.005779) | 0.095345 / 0.014526 (0.080819) | 0.105399 / 0.176557 (-0.071157) | 0.174173 / 0.737135 (-0.562962) | 0.104207 / 0.296338 (-0.192132) |\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) | 0.435312 / 0.215209 (0.220103) | 4.265600 / 2.077655 (2.187946) | 2.056500 / 1.504120 (0.552380) | 1.848023 / 1.541195 (0.306828) | 1.946156 / 1.468490 (0.477666) | 0.557788 / 4.584777 (-4.026989) | 4.070289 / 3.745712 (0.324577) | 3.608027 / 5.269862 (-1.661835) | 2.214556 / 4.565676 (-2.351121) | 0.062623 / 0.424275 (-0.361652) | 0.008083 / 0.007607 (0.000476) | 0.491782 / 0.226044 (0.265738) | 4.989963 / 2.268929 (2.721035) | 2.575867 / 55.444624 (-52.868757) | 2.208045 / 6.876477 (-4.668431) | 2.364184 / 2.142072 (0.222112) | 0.633925 / 4.805227 (-4.171302) | 0.144323 / 6.500664 (-6.356341) | 0.067505 / 0.075469 (-0.007965) |\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) | 1.467219 / 1.841788 (-0.374569) | 22.334967 / 8.074308 (14.260659) | 15.715747 / 10.191392 (5.524355) | 0.175443 / 0.680424 (-0.504980) | 0.026165 / 0.534201 (-0.508036) | 0.490675 / 0.579283 (-0.088608) | 0.509211 / 0.434364 (0.074847) | 0.586303 / 0.540337 (0.045965) | 0.785052 / 1.386936 (-0.601884) |\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) | 0.007893 / 0.011353 (-0.003460) | 0.004577 / 0.011008 (-0.006431) | 0.075781 / 0.038508 (0.037273) | 0.095492 / 0.023109 (0.072382) | 0.433259 / 0.275898 (0.157361) | 0.469386 / 0.323480 (0.145906) | 0.006317 / 0.007986 (-0.001669) | 0.003708 / 0.004328 (-0.000621) | 0.074417 / 0.004250 (0.070167) | 0.068605 / 0.037052 (0.031552) | 0.448701 / 0.258489 (0.190212) | 0.469131 / 0.293841 (0.175290) | 0.036647 / 0.128546 (-0.091899) | 0.010077 / 0.075646 (-0.065570) | 0.082457 / 0.419271 (-0.336815) | 0.063255 / 0.043533 (0.019722) | 0.428144 / 0.255139 (0.173005) | 0.451872 / 0.283200 (0.168672) | 0.033953 / 0.141683 (-0.107730) | 1.781752 / 1.452155 (0.329597) | 1.869014 / 1.492716 (0.376297) |\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) | 0.223596 / 0.018006 (0.205590) | 0.470307 / 0.000490 (0.469818) | 0.005059 / 0.000200 (0.004859) | 0.000104 / 0.000054 (0.000049) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.038804 / 0.037411 (0.001393) | 0.117879 / 0.014526 (0.103353) | 0.140701 / 0.176557 (-0.035855) | 0.194672 / 0.737135 (-0.542463) | 0.132806 / 0.296338 (-0.163533) |\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) | 0.510109 / 0.215209 (0.294900) | 4.729457 / 2.077655 (2.651803) | 2.512113 / 1.504120 (1.007993) | 2.302553 / 1.541195 (0.761358) | 2.420462 / 1.468490 (0.951972) | 0.531682 / 4.584777 (-4.053095) | 4.061208 / 3.745712 (0.315496) | 3.588542 / 5.269862 (-1.681320) | 2.203187 / 4.565676 (-2.362489) | 0.065791 / 0.424275 (-0.358484) | 0.008839 / 0.007607 (0.001232) | 0.562041 / 0.226044 (0.335997) | 5.702340 / 2.268929 (3.433412) | 3.127609 / 55.444624 (-52.317015) | 2.823060 / 6.876477 (-4.053417) | 2.898675 / 2.142072 (0.756603) | 0.659589 / 4.805227 (-4.145638) | 0.148798 / 6.500664 (-6.351866) | 0.070787 / 0.075469 (-0.004682) |\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) | 1.478317 / 1.841788 (-0.363471) | 21.995400 / 8.074308 (13.921092) | 16.770729 / 10.191392 (6.579337) | 0.226333 / 0.680424 (-0.454091) | 0.021835 / 0.534201 (-0.512366) | 0.460373 / 0.579283 (-0.118910) | 0.479494 / 0.434364 (0.045130) | 0.529470 / 0.540337 (-0.010868) | 0.718066 / 1.386936 (-0.668870) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#9a717b8eb80b0e50b25818127f79a35e0866fb14 \"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) | 0.007824 / 0.011353 (-0.003529) | 0.004601 / 0.011008 (-0.006407) | 0.100025 / 0.038508 (0.061517) | 0.096046 / 0.023109 (0.072936) | 0.376226 / 0.275898 (0.100328) | 0.410905 / 0.323480 (0.087425) | 0.006048 / 0.007986 (-0.001938) | 0.003817 / 0.004328 (-0.000511) | 0.076624 / 0.004250 (0.072374) | 0.066390 / 0.037052 (0.029338) | 0.380098 / 0.258489 (0.121609) | 0.413603 / 0.293841 (0.119762) | 0.036546 / 0.128546 (-0.092001) | 0.009881 / 0.075646 (-0.065765) | 0.344338 / 0.419271 (-0.074934) | 0.061882 / 0.043533 (0.018350) | 0.368568 / 0.255139 (0.113429) | 0.397133 / 0.283200 (0.113934) | 0.027255 / 0.141683 (-0.114428) | 1.795099 / 1.452155 (0.342945) | 1.852443 / 1.492716 (0.359727) |\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) | 0.247436 / 0.018006 (0.229430) | 0.494119 / 0.000490 (0.493629) | 0.004359 / 0.000200 (0.004159) | 0.000089 / 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) | 0.034765 / 0.037411 (-0.002647) | 0.104541 / 0.014526 (0.090015) | 0.113898 / 0.176557 (-0.062659) | 0.183634 / 0.737135 (-0.553501) | 0.116423 / 0.296338 (-0.179916) |\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) | 0.458747 / 0.215209 (0.243538) | 4.555740 / 2.077655 (2.478085) | 2.217240 / 1.504120 (0.713121) | 2.039879 / 1.541195 (0.498684) | 2.088581 / 1.468490 (0.620091) | 0.588063 / 4.584777 (-3.996714) | 4.238226 / 3.745712 (0.492514) | 4.768060 / 5.269862 (-0.501802) | 2.857117 / 4.565676 (-1.708560) | 0.068742 / 0.424275 (-0.355533) | 0.008667 / 0.007607 (0.001059) | 0.549294 / 0.226044 (0.323249) | 5.464635 / 2.268929 (3.195706) | 2.744435 / 55.444624 (-52.700189) | 2.347660 / 6.876477 (-4.528816) | 2.616816 / 2.142072 (0.474743) | 0.703701 / 4.805227 (-4.101526) | 0.159749 / 6.500664 (-6.340915) | 0.071990 / 0.075469 (-0.003479) |\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) | 1.486599 / 1.841788 (-0.355188) | 22.745438 / 8.074308 (14.671130) | 16.822332 / 10.191392 (6.630940) | 0.184730 / 0.680424 (-0.495694) | 0.021267 / 0.534201 (-0.512934) | 0.467108 / 0.579283 (-0.112176) | 0.472674 / 0.434364 (0.038311) | 0.548094 / 0.540337 (0.007756) | 0.735885 / 1.386936 (-0.651051) |\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) | 0.007746 / 0.011353 (-0.003607) | 0.004585 / 0.011008 (-0.006423) | 0.076943 / 0.038508 (0.038435) | 0.087473 / 0.023109 (0.064363) | 0.480099 / 0.275898 (0.204201) | 0.495271 / 0.323480 (0.171791) | 0.006348 / 0.007986 (-0.001638) | 0.003902 / 0.004328 (-0.000426) | 0.077586 / 0.004250 (0.073335) | 0.066467 / 0.037052 (0.029415) | 0.468741 / 0.258489 (0.210252) | 0.506778 / 0.293841 (0.212937) | 0.036877 / 0.128546 (-0.091669) | 0.010102 / 0.075646 (-0.065545) | 0.084419 / 0.419271 (-0.334852) | 0.058721 / 0.043533 (0.015188) | 0.453633 / 0.255139 (0.198494) | 0.481171 / 0.283200 (0.197971) | 0.028716 / 0.141683 (-0.112967) | 1.853048 / 1.452155 (0.400893) | 1.885847 / 1.492716 (0.393130) |\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) | 0.192136 / 0.018006 (0.174130) | 0.484481 / 0.000490 (0.483991) | 0.002951 / 0.000200 (0.002751) | 0.000098 / 0.000054 (0.000044) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.037949 / 0.037411 (0.000538) | 0.108364 / 0.014526 (0.093838) | 0.119542 / 0.176557 (-0.057014) | 0.188542 / 0.737135 (-0.548593) | 0.122011 / 0.296338 (-0.174327) |\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) | 0.483135 / 0.215209 (0.267926) | 4.849715 / 2.077655 (2.772060) | 2.497736 / 1.504120 (0.993616) | 2.314243 / 1.541195 (0.773048) | 2.412739 / 1.468490 (0.944249) | 0.564137 / 4.584777 (-4.020639) | 4.242273 / 3.745712 (0.496561) | 6.337843 / 5.269862 (1.067982) | 3.923250 / 4.565676 (-0.642426) | 0.066464 / 0.424275 (-0.357811) | 0.009217 / 0.007607 (0.001610) | 0.575667 / 0.226044 (0.349623) | 5.746187 / 2.268929 (3.477258) | 3.069655 / 55.444624 (-52.374969) | 2.674798 / 6.876477 (-4.201679) | 2.956535 / 2.142072 (0.814463) | 0.701043 / 4.805227 (-4.104185) | 0.157241 / 6.500664 (-6.343423) | 0.073175 / 0.075469 (-0.002294) |\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) | 1.609943 / 1.841788 (-0.231844) | 23.478594 / 8.074308 (15.404286) | 17.454437 / 10.191392 (7.263045) | 0.186422 / 0.680424 (-0.494002) | 0.021703 / 0.534201 (-0.512498) | 0.471704 / 0.579283 (-0.107579) | 0.480553 / 0.434364 (0.046189) | 0.552881 / 0.540337 (0.012544) | 0.722515 / 1.386936 (-0.664421) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#84645f80049cd00d9e0d4908faf3c3203fdcf21d \"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) | 0.007542 / 0.011353 (-0.003811) | 0.004692 / 0.011008 (-0.006316) | 0.099155 / 0.038508 (0.060647) | 0.089365 / 0.023109 (0.066256) | 0.370870 / 0.275898 (0.094972) | 0.422152 / 0.323480 (0.098673) | 0.006223 / 0.007986 (-0.001763) | 0.003852 / 0.004328 (-0.000476) | 0.075438 / 0.004250 (0.071188) | 0.065973 / 0.037052 (0.028921) | 0.381513 / 0.258489 (0.123024) | 0.416196 / 0.293841 (0.122355) | 0.035483 / 0.128546 (-0.093063) | 0.009884 / 0.075646 (-0.065762) | 0.341290 / 0.419271 (-0.077982) | 0.060546 / 0.043533 (0.017014) | 0.365101 / 0.255139 (0.109962) | 0.391058 / 0.283200 (0.107859) | 0.026325 / 0.141683 (-0.115358) | 1.815168 / 1.452155 (0.363013) | 1.834711 / 1.492716 (0.341994) |\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) | 0.222177 / 0.018006 (0.204171) | 0.501151 / 0.000490 (0.500662) | 0.010202 / 0.000200 (0.010002) | 0.000102 / 0.000054 (0.000048) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.034043 / 0.037411 (-0.003368) | 0.097884 / 0.014526 (0.083358) | 0.114022 / 0.176557 (-0.062534) | 0.186200 / 0.737135 (-0.550935) | 0.115555 / 0.296338 (-0.180783) |\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) | 0.485857 / 0.215209 (0.270648) | 4.959263 / 2.077655 (2.881608) | 2.501085 / 1.504120 (0.996965) | 2.234660 / 1.541195 (0.693465) | 2.238585 / 1.468490 (0.770095) | 0.645431 / 4.584777 (-3.939345) | 4.434311 / 3.745712 (0.688599) | 4.771491 / 5.269862 (-0.498371) | 2.778963 / 4.565676 (-1.786714) | 0.075615 / 0.424275 (-0.348660) | 0.009502 / 0.007607 (0.001895) | 0.546539 / 0.226044 (0.320495) | 5.464242 / 2.268929 (3.195314) | 2.894101 / 55.444624 (-52.550524) | 2.513761 / 6.876477 (-4.362715) | 2.719843 / 2.142072 (0.577770) | 0.678828 / 4.805227 (-4.126399) | 0.157839 / 6.500664 (-6.342825) | 0.071305 / 0.075469 (-0.004164) |\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) | 1.496879 / 1.841788 (-0.344909) | 22.214452 / 8.074308 (14.140144) | 17.707541 / 10.191392 (7.516149) | 0.197008 / 0.680424 (-0.483416) | 0.024883 / 0.534201 (-0.509318) | 0.493611 / 0.579283 (-0.085672) | 0.500677 / 0.434364 (0.066313) | 0.569381 / 0.540337 (0.029044) | 0.773950 / 1.386936 (-0.612986) |\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) | 0.007337 / 0.011353 (-0.004015) | 0.004572 / 0.011008 (-0.006436) | 0.091123 / 0.038508 (0.052615) | 0.079762 / 0.023109 (0.056652) | 0.450527 / 0.275898 (0.174629) | 0.525097 / 0.323480 (0.201617) | 0.005873 / 0.007986 (-0.002112) | 0.003797 / 0.004328 (-0.000532) | 0.076259 / 0.004250 (0.072009) | 0.062745 / 0.037052 (0.025692) | 0.465553 / 0.258489 (0.207064) | 0.546026 / 0.293841 (0.252186) | 0.035638 / 0.128546 (-0.092909) | 0.010086 / 0.075646 (-0.065560) | 0.109269 / 0.419271 (-0.310002) | 0.056765 / 0.043533 (0.013233) | 0.440887 / 0.255139 (0.185748) | 0.513325 / 0.283200 (0.230125) | 0.027206 / 0.141683 (-0.114476) | 1.863564 / 1.452155 (0.411409) | 1.918206 / 1.492716 (0.425490) |\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) | 0.266479 / 0.018006 (0.248473) | 0.487971 / 0.000490 (0.487481) | 0.012246 / 0.000200 (0.012046) | 0.000119 / 0.000054 (0.000065) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.035281 / 0.037411 (-0.002130) | 0.102991 / 0.014526 (0.088465) | 0.114638 / 0.176557 (-0.061919) | 0.184117 / 0.737135 (-0.553018) | 0.117943 / 0.296338 (-0.178396) |\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) | 0.497897 / 0.215209 (0.282688) | 4.973806 / 2.077655 (2.896151) | 2.596146 / 1.504120 (1.092026) | 2.419694 / 1.541195 (0.878499) | 2.525784 / 1.468490 (1.057294) | 0.568021 / 4.584777 (-4.016756) | 4.296431 / 3.745712 (0.550719) | 3.690682 / 5.269862 (-1.579179) | 2.345965 / 4.565676 (-2.219712) | 0.066859 / 0.424275 (-0.357416) | 0.009093 / 0.007607 (0.001486) | 0.582616 / 0.226044 (0.356571) | 5.826528 / 2.268929 (3.557600) | 3.253222 / 55.444624 (-52.191403) | 2.798447 / 6.876477 (-4.078030) | 3.054609 / 2.142072 (0.912537) | 0.678816 / 4.805227 (-4.126411) | 0.157966 / 6.500664 (-6.342698) | 0.073797 / 0.075469 (-0.001672) |\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) | 1.599480 / 1.841788 (-0.242308) | 23.249738 / 8.074308 (15.175430) | 16.965406 / 10.191392 (6.774014) | 0.171390 / 0.680424 (-0.509034) | 0.021810 / 0.534201 (-0.512391) | 0.483339 / 0.579283 (-0.095944) | 0.496615 / 0.434364 (0.062251) | 0.583786 / 0.540337 (0.043448) | 0.741699 / 1.386936 (-0.645237) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#7935cd2e564f5d1c66ed1acf731703724ba7a287 \"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) | 0.006054 / 0.011353 (-0.005299) | 0.003706 / 0.011008 (-0.007302) | 0.080060 / 0.038508 (0.041552) | 0.061479 / 0.023109 (0.038370) | 0.327981 / 0.275898 (0.052083) | 0.356930 / 0.323480 (0.033450) | 0.004671 / 0.007986 (-0.003315) | 0.002901 / 0.004328 (-0.001428) | 0.062425 / 0.004250 (0.058174) | 0.046310 / 0.037052 (0.009258) | 0.323657 / 0.258489 (0.065168) | 0.370130 / 0.293841 (0.076289) | 0.027151 / 0.128546 (-0.101395) | 0.007850 / 0.075646 (-0.067797) | 0.262300 / 0.419271 (-0.156971) | 0.045456 / 0.043533 (0.001923) | 0.325569 / 0.255139 (0.070430) | 0.352962 / 0.283200 (0.069762) | 0.020156 / 0.141683 (-0.121527) | 1.429404 / 1.452155 (-0.022750) | 1.615032 / 1.492716 (0.122316) |\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) | 0.187309 / 0.018006 (0.169303) | 0.428848 / 0.000490 (0.428358) | 0.003599 / 0.000200 (0.003399) | 0.000069 / 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) | 0.023260 / 0.037411 (-0.014151) | 0.072467 / 0.014526 (0.057941) | 0.082398 / 0.176557 (-0.094159) | 0.142573 / 0.737135 (-0.594562) | 0.082570 / 0.296338 (-0.213768) |\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) | 0.426503 / 0.215209 (0.211294) | 4.267875 / 2.077655 (2.190220) | 2.189762 / 1.504120 (0.685642) | 2.027992 / 1.541195 (0.486798) | 2.053211 / 1.468490 (0.584721) | 0.503850 / 4.584777 (-4.080927) | 3.086444 / 3.745712 (-0.659268) | 3.319492 / 5.269862 (-1.950370) | 2.070714 / 4.565676 (-2.494962) | 0.057591 / 0.424275 (-0.366684) | 0.006407 / 0.007607 (-0.001200) | 0.501145 / 0.226044 (0.275100) | 5.017753 / 2.268929 (2.748825) | 2.643145 / 55.444624 (-52.801479) | 2.327440 / 6.876477 (-4.549037) | 2.460250 / 2.142072 (0.318178) | 0.589397 / 4.805227 (-4.215830) | 0.124948 / 6.500664 (-6.375716) | 0.060450 / 0.075469 (-0.015020) |\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) | 1.279870 / 1.841788 (-0.561918) | 18.115908 / 8.074308 (10.041600) | 13.570032 / 10.191392 (3.378640) | 0.132981 / 0.680424 (-0.547442) | 0.016942 / 0.534201 (-0.517259) | 0.333591 / 0.579283 (-0.245692) | 0.358844 / 0.434364 (-0.075520) | 0.395748 / 0.540337 (-0.144590) | 0.546213 / 1.386936 (-0.840723) |\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) | 0.006062 / 0.011353 (-0.005291) | 0.003673 / 0.011008 (-0.007336) | 0.064726 / 0.038508 (0.026218) | 0.061854 / 0.023109 (0.038745) | 0.385343 / 0.275898 (0.109445) | 0.441284 / 0.323480 (0.117805) | 0.004830 / 0.007986 (-0.003156) | 0.002909 / 0.004328 (-0.001420) | 0.063874 / 0.004250 (0.059624) | 0.049331 / 0.037052 (0.012278) | 0.418484 / 0.258489 (0.159995) | 0.451397 / 0.293841 (0.157556) | 0.027665 / 0.128546 (-0.100881) | 0.008088 / 0.075646 (-0.067558) | 0.069625 / 0.419271 (-0.349646) | 0.043437 / 0.043533 (-0.000095) | 0.359789 / 0.255139 (0.104650) | 0.430206 / 0.283200 (0.147007) | 0.022308 / 0.141683 (-0.119375) | 1.461030 / 1.452155 (0.008875) | 1.513683 / 1.492716 (0.020966) |\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) | 0.230958 / 0.018006 (0.212952) | 0.417553 / 0.000490 (0.417063) | 0.000802 / 0.000200 (0.000602) | 0.000066 / 0.000054 (0.000011) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025421 / 0.037411 (-0.011991) | 0.077156 / 0.014526 (0.062630) | 0.087533 / 0.176557 (-0.089024) | 0.138048 / 0.737135 (-0.599087) | 0.089358 / 0.296338 (-0.206981) |\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) | 0.439172 / 0.215209 (0.223963) | 4.409509 / 2.077655 (2.331854) | 2.491270 / 1.504120 (0.987150) | 2.308446 / 1.541195 (0.767252) | 2.378440 / 1.468490 (0.909950) | 0.499834 / 4.584777 (-4.084943) | 3.083168 / 3.745712 (-0.662544) | 2.867543 / 5.269862 (-2.402318) | 1.876354 / 4.565676 (-2.689323) | 0.057092 / 0.424275 (-0.367183) | 0.006955 / 0.007607 (-0.000653) | 0.513799 / 0.226044 (0.287754) | 5.126660 / 2.268929 (2.857731) | 2.917348 / 55.444624 (-52.527277) | 2.508035 / 6.876477 (-4.368441) | 2.698089 / 2.142072 (0.556016) | 0.586828 / 4.805227 (-4.218399) | 0.124740 / 6.500664 (-6.375924) | 0.062276 / 0.075469 (-0.013193) |\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) | 1.291624 / 1.841788 (-0.550164) | 18.199968 / 8.074308 (10.125660) | 13.888139 / 10.191392 (3.696747) | 0.162955 / 0.680424 (-0.517469) | 0.017343 / 0.534201 (-0.516858) | 0.334683 / 0.579283 (-0.244600) | 0.352708 / 0.434364 (-0.081656) | 0.400629 / 0.540337 (-0.139708) | 0.539497 / 1.386936 (-0.847439) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#e7976db7fe22c6b93a869488d07b8137ea6a0db4 \"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) | 0.004498 / 0.011008 (-0.006510) | 0.100239 / 0.038508 (0.061731) | 0.083424 / 0.023109 (0.060315) | 0.366664 / 0.275898 (0.090766) | 0.406641 / 0.323480 (0.083161) | 0.004577 / 0.007986 (-0.003409) | 0.004809 / 0.004328 (0.000480) | 0.076898 / 0.004250 (0.072647) | 0.064021 / 0.037052 (0.026969) | 0.375836 / 0.258489 (0.117347) | 0.413008 / 0.293841 (0.119167) | 0.036010 / 0.128546 (-0.092537) | 0.009655 / 0.075646 (-0.065991) | 0.342595 / 0.419271 (-0.076677) | 0.061846 / 0.043533 (0.018313) | 0.376543 / 0.255139 (0.121404) | 0.395858 / 0.283200 (0.112659) | 0.026792 / 0.141683 (-0.114891) | 1.775569 / 1.452155 (0.323414) | 1.865077 / 1.492716 (0.372360) |\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) | 0.221521 / 0.018006 (0.203514) | 0.474604 / 0.000490 (0.474114) | 0.004354 / 0.000200 (0.004154) | 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) | 0.032947 / 0.037411 (-0.004464) | 0.100454 / 0.014526 (0.085928) | 0.111955 / 0.176557 (-0.064602) | 0.179752 / 0.737135 (-0.557383) | 0.114282 / 0.296338 (-0.182056) |\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) | 0.458261 / 0.215209 (0.243052) | 4.563536 / 2.077655 (2.485881) | 2.231928 / 1.504120 (0.727808) | 2.036751 / 1.541195 (0.495556) | 2.170413 / 1.468490 (0.701923) | 0.570825 / 4.584777 (-4.013952) | 4.505762 / 3.745712 (0.760050) | 5.033461 / 5.269862 (-0.236401) | 2.704989 / 4.565676 (-1.860687) | 0.067011 / 0.424275 (-0.357264) | 0.008568 / 0.007607 (0.000961) | 0.545151 / 0.226044 (0.319106) | 5.438984 / 2.268929 (3.170055) | 2.771818 / 55.444624 (-52.672806) | 2.393082 / 6.876477 (-4.483395) | 2.467173 / 2.142072 (0.325101) | 0.678849 / 4.805227 (-4.126379) | 0.160480 / 6.500664 (-6.340184) | 0.073681 / 0.075469 (-0.001788) |\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) | 1.532272 / 1.841788 (-0.309516) | 22.548741 / 8.074308 (14.474433) | 17.091044 / 10.191392 (6.899652) | 0.172100 / 0.680424 (-0.508324) | 0.022220 / 0.534201 (-0.511981) | 0.467871 / 0.579283 (-0.111412) | 0.491135 / 0.434364 (0.056771) | 0.548433 / 0.540337 (0.008096) | 0.733340 / 1.386936 (-0.653596) |\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) | 0.007593 / 0.011353 (-0.003760) | 0.004656 / 0.011008 (-0.006352) | 0.076940 / 0.038508 (0.038431) | 0.085183 / 0.023109 (0.062073) | 0.447178 / 0.275898 (0.171280) | 0.469545 / 0.323480 (0.146065) | 0.006023 / 0.007986 (-0.001962) | 0.003808 / 0.004328 (-0.000520) | 0.076767 / 0.004250 (0.072517) | 0.065713 / 0.037052 (0.028661) | 0.445573 / 0.258489 (0.187084) | 0.481689 / 0.293841 (0.187848) | 0.036893 / 0.128546 (-0.091654) | 0.009976 / 0.075646 (-0.065670) | 0.084443 / 0.419271 (-0.334829) | 0.058829 / 0.043533 (0.015297) | 0.429291 / 0.255139 (0.174152) | 0.454016 / 0.283200 (0.170816) | 0.027289 / 0.141683 (-0.114394) | 1.806786 / 1.452155 (0.354632) | 1.887680 / 1.492716 (0.394964) |\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) | 0.241012 / 0.018006 (0.223006) | 0.470629 / 0.000490 (0.470139) | 0.003213 / 0.000200 (0.003013) | 0.000107 / 0.000054 (0.000052) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.036896 / 0.037411 (-0.000515) | 0.106932 / 0.014526 (0.092406) | 0.120333 / 0.176557 (-0.056223) | 0.186271 / 0.737135 (-0.550865) | 0.121581 / 0.296338 (-0.174758) |\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) | 0.507782 / 0.215209 (0.292573) | 5.062932 / 2.077655 (2.985278) | 2.689539 / 1.504120 (1.185419) | 2.482978 / 1.541195 (0.941784) | 2.561320 / 1.468490 (1.092830) | 0.570664 / 4.584777 (-4.014113) | 4.346051 / 3.745712 (0.600339) | 6.479374 / 5.269862 (1.209513) | 4.096483 / 4.565676 (-0.469194) | 0.067564 / 0.424275 (-0.356711) | 0.009147 / 0.007607 (0.001540) | 0.596059 / 0.226044 (0.370015) | 5.963223 / 2.268929 (3.694295) | 3.201039 / 55.444624 (-52.243585) | 2.816581 / 6.876477 (-4.059896) | 3.047821 / 2.142072 (0.905748) | 0.687749 / 4.805227 (-4.117478) | 0.158174 / 6.500664 (-6.342490) | 0.073329 / 0.075469 (-0.002140) |\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) | 1.601346 / 1.841788 (-0.240441) | 23.712210 / 8.074308 (15.637902) | 16.567272 / 10.191392 (6.375880) | 0.224745 / 0.680424 (-0.455679) | 0.021662 / 0.534201 (-0.512539) | 0.471427 / 0.579283 (-0.107856) | 0.498751 / 0.434364 (0.064387) | 0.572047 / 0.540337 (0.031710) | 0.821868 / 1.386936 (-0.565068) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#34d0c9027c750adc89f3d04a6bf2e9cb95915da4 \"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) | 0.006371 / 0.011353 (-0.004981) | 0.003749 / 0.011008 (-0.007259) | 0.084155 / 0.038508 (0.045647) | 0.072450 / 0.023109 (0.049340) | 0.308002 / 0.275898 (0.032104) | 0.340471 / 0.323480 (0.016991) | 0.005054 / 0.007986 (-0.002931) | 0.003176 / 0.004328 (-0.001152) | 0.064867 / 0.004250 (0.060616) | 0.054305 / 0.037052 (0.017252) | 0.321047 / 0.258489 (0.062558) | 0.345999 / 0.293841 (0.052158) | 0.030507 / 0.128546 (-0.098039) | 0.008299 / 0.075646 (-0.067347) | 0.287682 / 0.419271 (-0.131590) | 0.052048 / 0.043533 (0.008515) | 0.308322 / 0.255139 (0.053183) | 0.333220 / 0.283200 (0.050020) | 0.022698 / 0.141683 (-0.118985) | 1.474033 / 1.452155 (0.021879) | 1.544790 / 1.492716 (0.052074) |\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) | 0.200612 / 0.018006 (0.182606) | 0.450934 / 0.000490 (0.450445) | 0.005383 / 0.000200 (0.005183) | 0.000200 / 0.000054 (0.000145) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027759 / 0.037411 (-0.009652) | 0.080935 / 0.014526 (0.066409) | 0.093041 / 0.176557 (-0.083516) | 0.148643 / 0.737135 (-0.588492) | 0.093463 / 0.296338 (-0.202876) |\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) | 0.381653 / 0.215209 (0.166444) | 3.810699 / 2.077655 (1.733044) | 1.866858 / 1.504120 (0.362738) | 1.716985 / 1.541195 (0.175790) | 1.788071 / 1.468490 (0.319581) | 0.481130 / 4.584777 (-4.103647) | 3.529798 / 3.745712 (-0.215914) | 3.982037 / 5.269862 (-1.287824) | 2.324866 / 4.565676 (-2.240811) | 0.056767 / 0.424275 (-0.367508) | 0.007306 / 0.007607 (-0.000301) | 0.459472 / 0.226044 (0.233428) | 4.602808 / 2.268929 (2.333879) | 2.332014 / 55.444624 (-53.112610) | 2.044858 / 6.876477 (-4.831619) | 2.204165 / 2.142072 (0.062093) | 0.577946 / 4.805227 (-4.227281) | 0.130900 / 6.500664 (-6.369764) | 0.059054 / 0.075469 (-0.016415) |\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) | 1.245211 / 1.841788 (-0.596576) | 19.176397 / 8.074308 (11.102089) | 13.995280 / 10.191392 (3.803888) | 0.171743 / 0.680424 (-0.508681) | 0.018038 / 0.534201 (-0.516163) | 0.392338 / 0.579283 (-0.186945) | 0.419370 / 0.434364 (-0.014994) | 0.477829 / 0.540337 (-0.062508) | 0.677409 / 1.386936 (-0.709527) |\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) | 0.006513 / 0.011353 (-0.004840) | 0.003984 / 0.011008 (-0.007024) | 0.064516 / 0.038508 (0.026008) | 0.070504 / 0.023109 (0.047395) | 0.384509 / 0.275898 (0.108611) | 0.410564 / 0.323480 (0.087084) | 0.005310 / 0.007986 (-0.002675) | 0.003268 / 0.004328 (-0.001061) | 0.064684 / 0.004250 (0.060433) | 0.055367 / 0.037052 (0.018315) | 0.399108 / 0.258489 (0.140619) | 0.422740 / 0.293841 (0.128900) | 0.031624 / 0.128546 (-0.096922) | 0.008617 / 0.075646 (-0.067030) | 0.070929 / 0.419271 (-0.348342) | 0.049146 / 0.043533 (0.005613) | 0.385492 / 0.255139 (0.130353) | 0.407434 / 0.283200 (0.124234) | 0.021972 / 0.141683 (-0.119711) | 1.496135 / 1.452155 (0.043980) | 1.533739 / 1.492716 (0.041023) |\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) | 0.226218 / 0.018006 (0.208211) | 0.443176 / 0.000490 (0.442686) | 0.000376 / 0.000200 (0.000176) | 0.000055 / 0.000054 (0.000000) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030315 / 0.037411 (-0.007097) | 0.086416 / 0.014526 (0.071890) | 0.097725 / 0.176557 (-0.078831) | 0.150407 / 0.737135 (-0.586728) | 0.099914 / 0.296338 (-0.196424) |\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) | 0.409807 / 0.215209 (0.194598) | 4.099086 / 2.077655 (2.021431) | 2.103160 / 1.504120 (0.599040) | 1.927927 / 1.541195 (0.386733) | 1.977751 / 1.468490 (0.509261) | 0.476995 / 4.584777 (-4.107781) | 3.521835 / 3.745712 (-0.223877) | 3.237695 / 5.269862 (-2.032167) | 1.995953 / 4.565676 (-2.569724) | 0.056208 / 0.424275 (-0.368068) | 0.007660 / 0.007607 (0.000053) | 0.483537 / 0.226044 (0.257492) | 4.833974 / 2.268929 (2.565046) | 2.589115 / 55.444624 (-52.855510) | 2.228076 / 6.876477 (-4.648401) | 2.395271 / 2.142072 (0.253198) | 0.577534 / 4.805227 (-4.227694) | 0.131432 / 6.500664 (-6.369232) | 0.060999 / 0.075469 (-0.014471) |\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) | 1.356043 / 1.841788 (-0.485745) | 19.470401 / 8.074308 (11.396093) | 14.091266 / 10.191392 (3.899874) | 0.166809 / 0.680424 (-0.513615) | 0.018782 / 0.534201 (-0.515419) | 0.394916 / 0.579283 (-0.184367) | 0.411378 / 0.434364 (-0.022986) | 0.466886 / 0.540337 (-0.073451) | 0.617369 / 1.386936 (-0.769567) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#601ae6c7baff33a600fd10b12940966024fd2221 \"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) | 0.007590 / 0.011353 (-0.003762) | 0.004068 / 0.011008 (-0.006941) | 0.105479 / 0.038508 (0.066971) | 0.085614 / 0.023109 (0.062505) | 0.384325 / 0.275898 (0.108427) | 0.467867 / 0.323480 (0.144387) | 0.004652 / 0.007986 (-0.003333) | 0.005445 / 0.004328 (0.001117) | 0.079604 / 0.004250 (0.075353) | 0.066031 / 0.037052 (0.028978) | 0.426184 / 0.258489 (0.167695) | 0.480712 / 0.293841 (0.186871) | 0.037837 / 0.128546 (-0.090709) | 0.009765 / 0.075646 (-0.065882) | 0.351316 / 0.419271 (-0.067955) | 0.063634 / 0.043533 (0.020101) | 0.420297 / 0.255139 (0.165158) | 0.449169 / 0.283200 (0.165969) | 0.030947 / 0.141683 (-0.110736) | 1.840184 / 1.452155 (0.388029) | 1.934074 / 1.492716 (0.441357) |\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) | 0.223483 / 0.018006 (0.205477) | 0.521086 / 0.000490 (0.520596) | 0.000379 / 0.000200 (0.000179) | 0.000065 / 0.000054 (0.000011) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.032011 / 0.037411 (-0.005400) | 0.101474 / 0.014526 (0.086948) | 0.108652 / 0.176557 (-0.067904) | 0.173340 / 0.737135 (-0.563796) | 0.114186 / 0.296338 (-0.182153) |\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) | 0.478020 / 0.215209 (0.262811) | 4.645400 / 2.077655 (2.567746) | 2.590763 / 1.504120 (1.086643) | 2.383002 / 1.541195 (0.841807) | 2.482550 / 1.468490 (1.014060) | 0.572417 / 4.584777 (-4.012360) | 4.233436 / 3.745712 (0.487724) | 4.858823 / 5.269862 (-0.411038) | 2.838913 / 4.565676 (-1.726764) | 0.070010 / 0.424275 (-0.354265) | 0.009602 / 0.007607 (0.001995) | 0.538735 / 0.226044 (0.312691) | 5.534340 / 2.268929 (3.265411) | 2.915006 / 55.444624 (-52.529619) | 2.625132 / 6.876477 (-4.251345) | 2.537838 / 2.142072 (0.395766) | 0.667870 / 4.805227 (-4.137357) | 0.146330 / 6.500664 (-6.354334) | 0.071631 / 0.075469 (-0.003838) |\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) | 1.594686 / 1.841788 (-0.247101) | 22.311113 / 8.074308 (14.236804) | 17.603983 / 10.191392 (7.412591) | 0.195995 / 0.680424 (-0.484428) | 0.022254 / 0.534201 (-0.511947) | 0.479661 / 0.579283 (-0.099622) | 0.463626 / 0.434364 (0.029262) | 0.483465 / 0.540337 (-0.056873) | 0.676141 / 1.386936 (-0.710795) |\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) | 0.006146 / 0.011353 (-0.005207) | 0.004856 / 0.011008 (-0.006152) | 0.067506 / 0.038508 (0.028998) | 0.073968 / 0.023109 (0.050859) | 0.470013 / 0.275898 (0.194115) | 0.479022 / 0.323480 (0.155542) | 0.005972 / 0.007986 (-0.002014) | 0.003846 / 0.004328 (-0.000483) | 0.075141 / 0.004250 (0.070890) | 0.058597 / 0.037052 (0.021544) | 0.481454 / 0.258489 (0.222965) | 0.515634 / 0.293841 (0.221793) | 0.034979 / 0.128546 (-0.093567) | 0.010385 / 0.075646 (-0.065261) | 0.072649 / 0.419271 (-0.346622) | 0.058183 / 0.043533 (0.014650) | 0.462138 / 0.255139 (0.206999) | 0.476093 / 0.283200 (0.192893) | 0.032918 / 0.141683 (-0.108765) | 1.820530 / 1.452155 (0.368375) | 1.626360 / 1.492716 (0.133644) |\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) | 0.208970 / 0.018006 (0.190964) | 0.492478 / 0.000490 (0.491988) | 0.005487 / 0.000200 (0.005287) | 0.000140 / 0.000054 (0.000086) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.037896 / 0.037411 (0.000484) | 0.089752 / 0.014526 (0.075227) | 0.107445 / 0.176557 (-0.069111) | 0.181260 / 0.737135 (-0.555876) | 0.105700 / 0.296338 (-0.190639) |\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) | 0.495031 / 0.215209 (0.279821) | 4.806939 / 2.077655 (2.729284) | 2.227928 / 1.504120 (0.723808) | 2.067117 / 1.541195 (0.525922) | 2.348982 / 1.468490 (0.880492) | 0.567201 / 4.584777 (-4.017576) | 4.166592 / 3.745712 (0.420880) | 3.654329 / 5.269862 (-1.615533) | 2.331092 / 4.565676 (-2.234584) | 0.062212 / 0.424275 (-0.362063) | 0.008775 / 0.007607 (0.001168) | 0.515413 / 0.226044 (0.289369) | 5.449300 / 2.268929 (3.180371) | 3.206574 / 55.444624 (-52.238050) | 2.600455 / 6.876477 (-4.276022) | 3.041162 / 2.142072 (0.899089) | 0.681899 / 4.805227 (-4.123328) | 0.155400 / 6.500664 (-6.345265) | 0.073933 / 0.075469 (-0.001537) |\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) | 1.572329 / 1.841788 (-0.269459) | 23.638519 / 8.074308 (15.564211) | 17.145663 / 10.191392 (6.954271) | 0.232690 / 0.680424 (-0.447734) | 0.028620 / 0.534201 (-0.505581) | 0.488105 / 0.579283 (-0.091178) | 0.490365 / 0.434364 (0.056001) | 0.599501 / 0.540337 (0.059164) | 0.708101 / 1.386936 (-0.678835) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#4a761315900880a25b347ad19b78bd567cfce1f0 \"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) | 0.005947 / 0.011353 (-0.005406) | 0.003577 / 0.011008 (-0.007431) | 0.081631 / 0.038508 (0.043122) | 0.058651 / 0.023109 (0.035541) | 0.342742 / 0.275898 (0.066843) | 0.384130 / 0.323480 (0.060650) | 0.004620 / 0.007986 (-0.003366) | 0.002885 / 0.004328 (-0.001444) | 0.063698 / 0.004250 (0.059448) | 0.048953 / 0.037052 (0.011901) | 0.367880 / 0.258489 (0.109391) | 0.407050 / 0.293841 (0.113209) | 0.027242 / 0.128546 (-0.101305) | 0.007914 / 0.075646 (-0.067733) | 0.262156 / 0.419271 (-0.157116) | 0.044750 / 0.043533 (0.001218) | 0.351613 / 0.255139 (0.096474) | 0.380284 / 0.283200 (0.097084) | 0.020080 / 0.141683 (-0.121603) | 1.498101 / 1.452155 (0.045946) | 1.543608 / 1.492716 (0.050892) |\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) | 0.180014 / 0.018006 (0.162008) | 0.436172 / 0.000490 (0.435682) | 0.003694 / 0.000200 (0.003494) | 0.000071 / 0.000054 (0.000017) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024389 / 0.037411 (-0.013022) | 0.072874 / 0.014526 (0.058348) | 0.083469 / 0.176557 (-0.093088) | 0.144600 / 0.737135 (-0.592536) | 0.084229 / 0.296338 (-0.212110) |\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) | 0.391636 / 0.215209 (0.176427) | 3.906941 / 2.077655 (1.829286) | 1.901944 / 1.504120 (0.397825) | 1.762702 / 1.541195 (0.221507) | 1.817970 / 1.468490 (0.349480) | 0.500345 / 4.584777 (-4.084432) | 3.011351 / 3.745712 (-0.734361) | 4.417763 / 5.269862 (-0.852098) | 2.689744 / 4.565676 (-1.875933) | 0.057765 / 0.424275 (-0.366511) | 0.006412 / 0.007607 (-0.001195) | 0.468156 / 0.226044 (0.242112) | 4.664975 / 2.268929 (2.396047) | 2.323355 / 55.444624 (-53.121270) | 1.984280 / 6.876477 (-4.892197) | 2.165215 / 2.142072 (0.023142) | 0.586950 / 4.805227 (-4.218278) | 0.124363 / 6.500664 (-6.376301) | 0.060702 / 0.075469 (-0.014767) |\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) | 1.238870 / 1.841788 (-0.602917) | 18.587360 / 8.074308 (10.513052) | 13.831674 / 10.191392 (3.640282) | 0.143542 / 0.680424 (-0.536882) | 0.016913 / 0.534201 (-0.517288) | 0.332314 / 0.579283 (-0.246969) | 0.345419 / 0.434364 (-0.088945) | 0.381257 / 0.540337 (-0.159081) | 0.537844 / 1.386936 (-0.849092) |\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) | 0.006294 / 0.011353 (-0.005059) | 0.003714 / 0.011008 (-0.007294) | 0.062684 / 0.038508 (0.024176) | 0.063520 / 0.023109 (0.040411) | 0.389591 / 0.275898 (0.113693) | 0.444278 / 0.323480 (0.120798) | 0.004825 / 0.007986 (-0.003160) | 0.003010 / 0.004328 (-0.001318) | 0.062767 / 0.004250 (0.058517) | 0.051739 / 0.037052 (0.014686) | 0.434299 / 0.258489 (0.175810) | 0.452003 / 0.293841 (0.158162) | 0.027375 / 0.128546 (-0.101171) | 0.008135 / 0.075646 (-0.067511) | 0.067401 / 0.419271 (-0.351871) | 0.042752 / 0.043533 (-0.000780) | 0.367633 / 0.255139 (0.112494) | 0.433039 / 0.283200 (0.149840) | 0.021086 / 0.141683 (-0.120597) | 1.488024 / 1.452155 (0.035870) | 1.507767 / 1.492716 (0.015050) |\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) | 0.230046 / 0.018006 (0.212040) | 0.428085 / 0.000490 (0.427595) | 0.002188 / 0.000200 (0.001988) | 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) | 0.026705 / 0.037411 (-0.010706) | 0.082466 / 0.014526 (0.067940) | 0.089378 / 0.176557 (-0.087179) | 0.147287 / 0.737135 (-0.589849) | 0.090426 / 0.296338 (-0.205913) |\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) | 0.430882 / 0.215209 (0.215672) | 4.296224 / 2.077655 (2.218569) | 2.229982 / 1.504120 (0.725862) | 2.048506 / 1.541195 (0.507311) | 2.129514 / 1.468490 (0.661024) | 0.502964 / 4.584777 (-4.081813) | 3.048125 / 3.745712 (-0.697587) | 4.208636 / 5.269862 (-1.061226) | 2.594015 / 4.565676 (-1.971661) | 0.057967 / 0.424275 (-0.366308) | 0.006875 / 0.007607 (-0.000732) | 0.513872 / 0.226044 (0.287828) | 5.126435 / 2.268929 (2.857506) | 2.691278 / 55.444624 (-52.753346) | 2.361723 / 6.876477 (-4.514754) | 2.511213 / 2.142072 (0.369141) | 0.593558 / 4.805227 (-4.211670) | 0.129332 / 6.500664 (-6.371332) | 0.064051 / 0.075469 (-0.011418) |\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) | 1.289049 / 1.841788 (-0.552739) | 18.912363 / 8.074308 (10.838055) | 14.226500 / 10.191392 (4.035108) | 0.131392 / 0.680424 (-0.549032) | 0.016750 / 0.534201 (-0.517451) | 0.330078 / 0.579283 (-0.249205) | 0.347588 / 0.434364 (-0.086776) | 0.383234 / 0.540337 (-0.157103) | 0.510967 / 1.386936 (-0.875969) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#d7892beb30bab0633b84398c5ea43d7e69fe38cc \"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) | 0.005974 / 0.011353 (-0.005379) | 0.003691 / 0.011008 (-0.007317) | 0.079410 / 0.038508 (0.040902) | 0.061769 / 0.023109 (0.038660) | 0.323310 / 0.275898 (0.047412) | 0.354325 / 0.323480 (0.030845) | 0.004794 / 0.007986 (-0.003191) | 0.002899 / 0.004328 (-0.001430) | 0.062104 / 0.004250 (0.057854) | 0.048973 / 0.037052 (0.011921) | 0.326497 / 0.258489 (0.068008) | 0.361347 / 0.293841 (0.067506) | 0.026741 / 0.128546 (-0.101805) | 0.007936 / 0.075646 (-0.067710) | 0.259168 / 0.419271 (-0.160104) | 0.044859 / 0.043533 (0.001327) | 0.319342 / 0.255139 (0.064203) | 0.343711 / 0.283200 (0.060511) | 0.022298 / 0.141683 (-0.119384) | 1.451595 / 1.452155 (-0.000560) | 1.573730 / 1.492716 (0.081014) |\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) | 0.173086 / 0.018006 (0.155080) | 0.432400 / 0.000490 (0.431910) | 0.003739 / 0.000200 (0.003539) | 0.000073 / 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) | 0.024477 / 0.037411 (-0.012934) | 0.073463 / 0.014526 (0.058937) | 0.083410 / 0.176557 (-0.093146) | 0.144760 / 0.737135 (-0.592376) | 0.084199 / 0.296338 (-0.212140) |\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) | 0.388251 / 0.215209 (0.173042) | 3.875375 / 2.077655 (1.797720) | 1.875515 / 1.504120 (0.371395) | 1.729282 / 1.541195 (0.188087) | 1.784732 / 1.468490 (0.316242) | 0.496985 / 4.584777 (-4.087792) | 3.030276 / 3.745712 (-0.715436) | 2.813192 / 5.269862 (-2.456669) | 1.868647 / 4.565676 (-2.697030) | 0.057376 / 0.424275 (-0.366899) | 0.006463 / 0.007607 (-0.001144) | 0.462153 / 0.226044 (0.236108) | 4.586583 / 2.268929 (2.317654) | 2.287730 / 55.444624 (-53.156894) | 1.972177 / 6.876477 (-4.904299) | 2.151592 / 2.142072 (0.009520) | 0.587169 / 4.805227 (-4.218058) | 0.127063 / 6.500664 (-6.373601) | 0.060297 / 0.075469 (-0.015172) |\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) | 1.267651 / 1.841788 (-0.574136) | 18.426011 / 8.074308 (10.351703) | 14.050470 / 10.191392 (3.859078) | 0.148063 / 0.680424 (-0.532361) | 0.017112 / 0.534201 (-0.517089) | 0.330051 / 0.579283 (-0.249232) | 0.358730 / 0.434364 (-0.075634) | 0.392365 / 0.540337 (-0.147972) | 0.534650 / 1.386936 (-0.852286) |\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) | 0.005936 / 0.011353 (-0.005417) | 0.003652 / 0.011008 (-0.007356) | 0.063066 / 0.038508 (0.024558) | 0.060617 / 0.023109 (0.037507) | 0.388293 / 0.275898 (0.112395) | 0.411422 / 0.323480 (0.087942) | 0.004691 / 0.007986 (-0.003295) | 0.002857 / 0.004328 (-0.001472) | 0.064198 / 0.004250 (0.059947) | 0.049124 / 0.037052 (0.012071) | 0.403601 / 0.258489 (0.145112) | 0.413619 / 0.293841 (0.119778) | 0.027279 / 0.128546 (-0.101267) | 0.008072 / 0.075646 (-0.067575) | 0.067890 / 0.419271 (-0.351381) | 0.041866 / 0.043533 (-0.001667) | 0.393438 / 0.255139 (0.138299) | 0.402865 / 0.283200 (0.119666) | 0.023381 / 0.141683 (-0.118302) | 1.496324 / 1.452155 (0.044170) | 1.538080 / 1.492716 (0.045364) |\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) | 0.212065 / 0.018006 (0.194059) | 0.410511 / 0.000490 (0.410021) | 0.001236 / 0.000200 (0.001036) | 0.000067 / 0.000054 (0.000012) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026012 / 0.037411 (-0.011399) | 0.076592 / 0.014526 (0.062066) | 0.085963 / 0.176557 (-0.090594) | 0.137803 / 0.737135 (-0.599332) | 0.087594 / 0.296338 (-0.208745) |\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) | 0.434283 / 0.215209 (0.219074) | 4.345478 / 2.077655 (2.267824) | 2.400954 / 1.504120 (0.896834) | 2.282024 / 1.541195 (0.740829) | 2.414247 / 1.468490 (0.945757) | 0.501855 / 4.584777 (-4.082922) | 3.059433 / 3.745712 (-0.686279) | 2.811288 / 5.269862 (-2.458574) | 1.856839 / 4.565676 (-2.708838) | 0.058017 / 0.424275 (-0.366258) | 0.006844 / 0.007607 (-0.000763) | 0.515376 / 0.226044 (0.289332) | 5.148775 / 2.268929 (2.879847) | 2.930807 / 55.444624 (-52.513817) | 2.520532 / 6.876477 (-4.355944) | 2.746299 / 2.142072 (0.604227) | 0.590102 / 4.805227 (-4.215125) | 0.125747 / 6.500664 (-6.374917) | 0.061873 / 0.075469 (-0.013597) |\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) | 1.306247 / 1.841788 (-0.535541) | 18.366048 / 8.074308 (10.291740) | 13.855617 / 10.191392 (3.664225) | 0.150124 / 0.680424 (-0.530300) | 0.017189 / 0.534201 (-0.517012) | 0.336285 / 0.579283 (-0.242998) | 0.344985 / 0.434364 (-0.089379) | 0.397973 / 0.540337 (-0.142364) | 0.536142 / 1.386936 (-0.850794) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#1ae24cf12054b4a512f198979b1ca7707bb99d56 \"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) | 0.006401 / 0.011353 (-0.004952) | 0.003789 / 0.011008 (-0.007219) | 0.079516 / 0.038508 (0.041008) | 0.068279 / 0.023109 (0.045170) | 0.295691 / 0.275898 (0.019793) | 0.327208 / 0.323480 (0.003728) | 0.005070 / 0.007986 (-0.002915) | 0.003044 / 0.004328 (-0.001285) | 0.061411 / 0.004250 (0.057161) | 0.053227 / 0.037052 (0.016175) | 0.297368 / 0.258489 (0.038879) | 0.334740 / 0.293841 (0.040899) | 0.029459 / 0.128546 (-0.099087) | 0.008080 / 0.075646 (-0.067566) | 0.267344 / 0.419271 (-0.151927) | 0.049877 / 0.043533 (0.006344) | 0.293853 / 0.255139 (0.038714) | 0.319819 / 0.283200 (0.036620) | 0.022593 / 0.141683 (-0.119089) | 1.459054 / 1.452155 (0.006900) | 1.471250 / 1.492716 (-0.021466) |\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) | 0.194326 / 0.018006 (0.176320) | 0.443565 / 0.000490 (0.443075) | 0.003745 / 0.000200 (0.003545) | 0.000075 / 0.000054 (0.000021) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026640 / 0.037411 (-0.010772) | 0.077630 / 0.014526 (0.063104) | 0.089364 / 0.176557 (-0.087192) | 0.147327 / 0.737135 (-0.589809) | 0.089603 / 0.296338 (-0.206735) |\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) | 0.373758 / 0.215209 (0.158549) | 3.746778 / 2.077655 (1.669123) | 1.814991 / 1.504120 (0.310871) | 1.645650 / 1.541195 (0.104455) | 1.690752 / 1.468490 (0.222262) | 0.472117 / 4.584777 (-4.112660) | 3.457346 / 3.745712 (-0.288367) | 3.138869 / 5.269862 (-2.130993) | 1.934924 / 4.565676 (-2.630753) | 0.055709 / 0.424275 (-0.368566) | 0.006680 / 0.007607 (-0.000927) | 0.446874 / 0.226044 (0.220829) | 4.458409 / 2.268929 (2.189480) | 2.253932 / 55.444624 (-53.190693) | 2.007240 / 6.876477 (-4.869237) | 2.081687 / 2.142072 (-0.060386) | 0.563379 / 4.805227 (-4.241848) | 0.128694 / 6.500664 (-6.371970) | 0.057409 / 0.075469 (-0.018060) |\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) | 1.212231 / 1.841788 (-0.629556) | 18.519121 / 8.074308 (10.444813) | 13.582243 / 10.191392 (3.390851) | 0.142488 / 0.680424 (-0.537936) | 0.017421 / 0.534201 (-0.516780) | 0.366864 / 0.579283 (-0.212419) | 0.401467 / 0.434364 (-0.032897) | 0.443659 / 0.540337 (-0.096679) | 0.618854 / 1.386936 (-0.768082) |\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) | 0.006121 / 0.011353 (-0.005232) | 0.003690 / 0.011008 (-0.007318) | 0.060340 / 0.038508 (0.021832) | 0.067215 / 0.023109 (0.044106) | 0.382846 / 0.275898 (0.106948) | 0.415774 / 0.323480 (0.092294) | 0.004868 / 0.007986 (-0.003118) | 0.003108 / 0.004328 (-0.001221) | 0.060572 / 0.004250 (0.056321) | 0.050453 / 0.037052 (0.013401) | 0.400494 / 0.258489 (0.142005) | 0.424368 / 0.293841 (0.130527) | 0.030279 / 0.128546 (-0.098267) | 0.008151 / 0.075646 (-0.067495) | 0.066707 / 0.419271 (-0.352564) | 0.046118 / 0.043533 (0.002585) | 0.386697 / 0.255139 (0.131558) | 0.410156 / 0.283200 (0.126957) | 0.020688 / 0.141683 (-0.120995) | 1.418162 / 1.452155 (-0.033993) | 1.463057 / 1.492716 (-0.029659) |\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) | 0.216081 / 0.018006 (0.198075) | 0.440541 / 0.000490 (0.440051) | 0.000371 / 0.000200 (0.000171) | 0.000054 / 0.000054 (-0.000000) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.027763 / 0.037411 (-0.009648) | 0.082316 / 0.014526 (0.067791) | 0.094086 / 0.176557 (-0.082471) | 0.144738 / 0.737135 (-0.592398) | 0.094837 / 0.296338 (-0.201501) |\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) | 0.396277 / 0.215209 (0.181068) | 3.958791 / 2.077655 (1.881136) | 2.021367 / 1.504120 (0.517247) | 1.860112 / 1.541195 (0.318917) | 1.886032 / 1.468490 (0.417541) | 0.468536 / 4.584777 (-4.116241) | 3.417950 / 3.745712 (-0.327762) | 4.849991 / 5.269862 (-0.419871) | 2.773935 / 4.565676 (-1.791742) | 0.055813 / 0.424275 (-0.368462) | 0.007053 / 0.007607 (-0.000554) | 0.470167 / 0.226044 (0.244122) | 4.702969 / 2.268929 (2.434041) | 2.474161 / 55.444624 (-52.970464) | 2.171256 / 6.876477 (-4.705220) | 2.315373 / 2.142072 (0.173301) | 0.589195 / 4.805227 (-4.216032) | 0.128237 / 6.500664 (-6.372427) | 0.058641 / 0.075469 (-0.016828) |\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) | 1.292947 / 1.841788 (-0.548841) | 18.851300 / 8.074308 (10.776992) | 14.089764 / 10.191392 (3.898372) | 0.164853 / 0.680424 (-0.515571) | 0.017281 / 0.534201 (-0.516920) | 0.359112 / 0.579283 (-0.220171) | 0.386696 / 0.434364 (-0.047668) | 0.428222 / 0.540337 (-0.112115) | 0.568659 / 1.386936 (-0.818277) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#563864ded894b468e2ba3f677ef79c5ab3fe65df \"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) | 0.006051 / 0.011353 (-0.005301) | 0.003654 / 0.011008 (-0.007355) | 0.080081 / 0.038508 (0.041572) | 0.062925 / 0.023109 (0.039815) | 0.358097 / 0.275898 (0.082199) | 0.405728 / 0.323480 (0.082248) | 0.005359 / 0.007986 (-0.002627) | 0.002820 / 0.004328 (-0.001508) | 0.063108 / 0.004250 (0.058858) | 0.049627 / 0.037052 (0.012575) | 0.397870 / 0.258489 (0.139381) | 0.437157 / 0.293841 (0.143316) | 0.027707 / 0.128546 (-0.100839) | 0.007911 / 0.075646 (-0.067735) | 0.260991 / 0.419271 (-0.158280) | 0.044771 / 0.043533 (0.001238) | 0.340230 / 0.255139 (0.085091) | 0.384925 / 0.283200 (0.101725) | 0.021369 / 0.141683 (-0.120314) | 1.431439 / 1.452155 (-0.020715) | 1.478794 / 1.492716 (-0.013922) |\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) | 0.182626 / 0.018006 (0.164620) | 0.435551 / 0.000490 (0.435061) | 0.003015 / 0.000200 (0.002815) | 0.000064 / 0.000054 (0.000009) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.024703 / 0.037411 (-0.012708) | 0.073640 / 0.014526 (0.059114) | 0.084598 / 0.176557 (-0.091959) | 0.145810 / 0.737135 (-0.591325) | 0.085125 / 0.296338 (-0.211213) |\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) | 0.394539 / 0.215209 (0.179330) | 3.945882 / 2.077655 (1.868227) | 1.947166 / 1.504120 (0.443046) | 1.763305 / 1.541195 (0.222111) | 1.816208 / 1.468490 (0.347718) | 0.498880 / 4.584777 (-4.085897) | 3.098283 / 3.745712 (-0.647429) | 2.823474 / 5.269862 (-2.446388) | 1.873993 / 4.565676 (-2.691684) | 0.058097 / 0.424275 (-0.366179) | 0.006488 / 0.007607 (-0.001119) | 0.466711 / 0.226044 (0.240667) | 4.671520 / 2.268929 (2.402592) | 2.363381 / 55.444624 (-53.081243) | 2.052092 / 6.876477 (-4.824385) | 2.209212 / 2.142072 (0.067140) | 0.594650 / 4.805227 (-4.210577) | 0.125604 / 6.500664 (-6.375060) | 0.061511 / 0.075469 (-0.013958) |\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) | 1.226564 / 1.841788 (-0.615224) | 18.583605 / 8.074308 (10.509297) | 13.993091 / 10.191392 (3.801699) | 0.146185 / 0.680424 (-0.534239) | 0.016839 / 0.534201 (-0.517362) | 0.334116 / 0.579283 (-0.245167) | 0.360780 / 0.434364 (-0.073584) | 0.386008 / 0.540337 (-0.154329) | 0.643278 / 1.386936 (-0.743658) |\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) | 0.006174 / 0.011353 (-0.005179) | 0.003658 / 0.011008 (-0.007350) | 0.063250 / 0.038508 (0.024742) | 0.063542 / 0.023109 (0.040433) | 0.366845 / 0.275898 (0.090947) | 0.409794 / 0.323480 (0.086314) | 0.005678 / 0.007986 (-0.002308) | 0.003061 / 0.004328 (-0.001268) | 0.063561 / 0.004250 (0.059311) | 0.052648 / 0.037052 (0.015596) | 0.378096 / 0.258489 (0.119607) | 0.410706 / 0.293841 (0.116865) | 0.027668 / 0.128546 (-0.100878) | 0.008045 / 0.075646 (-0.067601) | 0.068290 / 0.419271 (-0.350981) | 0.042602 / 0.043533 (-0.000930) | 0.364976 / 0.255139 (0.109837) | 0.395599 / 0.283200 (0.112400) | 0.022733 / 0.141683 (-0.118950) | 1.522473 / 1.452155 (0.070319) | 1.515891 / 1.492716 (0.023175) |\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) | 0.232554 / 0.018006 (0.214547) | 0.420702 / 0.000490 (0.420213) | 0.002161 / 0.000200 (0.001961) | 0.000064 / 0.000054 (0.000009) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026276 / 0.037411 (-0.011135) | 0.078504 / 0.014526 (0.063978) | 0.088989 / 0.176557 (-0.087567) | 0.144044 / 0.737135 (-0.593091) | 0.091074 / 0.296338 (-0.205265) |\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) | 0.420189 / 0.215209 (0.204980) | 4.189596 / 2.077655 (2.111941) | 2.316425 / 1.504120 (0.812305) | 2.186877 / 1.541195 (0.645682) | 2.259065 / 1.468490 (0.790575) | 0.502827 / 4.584777 (-4.081950) | 3.135266 / 3.745712 (-0.610446) | 2.838808 / 5.269862 (-2.431053) | 1.876519 / 4.565676 (-2.689158) | 0.057802 / 0.424275 (-0.366473) | 0.006824 / 0.007607 (-0.000784) | 0.500213 / 0.226044 (0.274168) | 4.999798 / 2.268929 (2.730869) | 2.627713 / 55.444624 (-52.816911) | 2.344263 / 6.876477 (-4.532214) | 2.415449 / 2.142072 (0.273376) | 0.593082 / 4.805227 (-4.212145) | 0.125787 / 6.500664 (-6.374877) | 0.062699 / 0.075469 (-0.012770) |\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) | 1.308219 / 1.841788 (-0.533569) | 18.703099 / 8.074308 (10.628791) | 13.976234 / 10.191392 (3.784842) | 0.144037 / 0.680424 (-0.536387) | 0.016592 / 0.534201 (-0.517609) | 0.333078 / 0.579283 (-0.246206) | 0.342317 / 0.434364 (-0.092047) | 0.396837 / 0.540337 (-0.143500) | 0.532641 / 1.386936 (-0.854295) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#14f6edd9222e577dccb962ed5338b79b73502fa5 \"CML watermark\")\n" ]

[]

[]

[ { "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" } ]

[ "I am trying to load a dataset using Hugging Face Datasets load_dataset method. I am getting the value error as show below. Can someone help with this? I am using Windows laptop and Google Colab notebook.\r\n\r\n```\r\n!pip install zstandard\r\nfrom datasets import load_dataset\r\n\r\nlds = load_dataset(\r\n \"json\",\r\n data_files=\"https://the-eye.eu/public/AI/pile_preliminary_components/FreeLaw_Opinions.jsonl.zst\",\r\n split=\"train\",\r\n streaming=True,\r\n)\r\n\r\nWARNING:datasets.builder:Using custom data configuration default-a1d9e8eaedd958cd\r\n---------------------------------------------------------------------------\r\nValueError Traceback (most recent call last)\r\n[<ipython-input-12-5b4fdcb8e6d5>](https://localhost:8080/#) in <module>\r\n 6 )\r\n 7 \r\n----> 8 next(iter(law_dataset_streamed))\r\n\r\n17 frames\r\n[/usr/local/lib/python3.8/dist-packages/fsspec/core.py](https://localhost:8080/#) in get_compression(urlpath, compression)\r\n 485 compression = infer_compression(urlpath)\r\n 486 if compression is not None and compression not in compr:\r\n--> 487 raise ValueError(\"Compression type %s not supported\" % compression)\r\n 488 return compression\r\n 489 \r\n\r\nValueError: Compression type zstd not supported\r\n```", "I just tried on google colab and this works:\r\n```python\r\n!pip install zstandard\r\n!pip install datasets\r\nfrom datasets import load_dataset\r\n\r\nlds = load_dataset(\r\n \"json\",\r\n data_files=\"https://the-eye.eu/public/AI/pile_preliminary_components/FreeLaw_Opinions.jsonl.zst\",\r\n split=\"train\",\r\n streaming=True,\r\n)\r\nnext(iter(lds))\r\n```\r\n\r\nCan you check that you have a correct installation of `zstandard` ?", "@lhoestq please note [this](https://github.com/huggingface/datasets/issues/2572#issuecomment-1363718916) is a duplicate of:\r\n- #5388", "Oh thanks I missed that one !", "> I just tried on google colab and this works:\r\n> \r\n> ```python\r\n> !pip install zstandard\r\n> !pip install datasets\r\n> from datasets import load_dataset\r\n> \r\n> lds = load_dataset(\r\n> \"json\",\r\n> data_files=\"https://the-eye.eu/public/AI/pile_preliminary_components/FreeLaw_Opinions.jsonl.zst\",\r\n> split=\"train\",\r\n> streaming=True,\r\n> )\r\n> next(iter(lds))\r\n> ```\r\n> \r\n> Can you check that you have a correct installation of `zstandard` ?\r\n\r\nI was downloading datasets first then was doing zstandard installation and that was causing the issue. This was highlighted by the Hugging Face staff and that helped. Now the issue is resolved. Thank you." ]

[ { "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 agree we should improve the features equality checks to account for this particular case. However, your code fails due to `answer_start` having the dtype `int64` instead of `int32` after loading from JSON (it's not possible to embed type precision info into a JSON file; `save_to_disk` does that for arrow files), which would lead to the concatenation error as PyArrow does not support this sort of type promotion. This can be fixed as follows:\r\n```python\r\ntemp = load_dataset(\"json\", data_files={\"train\": \"output.jsonl\"}, features=squad[\"train\"].features)\r\n``` ", "That makes sense. I totally missed the `int64` and `int32` part. Thanks for pointing it out! Will close this issue for now." ]

[]