Datasets:

---

Francesco-A

/

github-issues_huggingface-datasets

like 0

[]

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

[]

[ "Can you run the style formatting tools to pass the code quality test?\r\n\r\nYou can find all the details in CONTRIBUTING.md: https://github.com/huggingface/nlp/blob/master/CONTRIBUTING.md#how-to-contribute-to-nlp", "Nice ! ", "``make style`` done! Thanks for the approvals." ]

[]

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

[]

[ "Nice, is that the reason one cannot rerun the tests without deleting the cache? \r\n", "Yes exactly. It was not using the temporary dir for tests." ]

[]

[ "Just for your information, Tensorflow confirmed this issue [here.](https://github.com/tensorflow/tensorflow/issues/59279)", "Thanks for reporting, @HaoyuYang59.\r\n\r\nPlease note that these are different \"dataset\" objects: our docs refer to Hugging Face `datasets.Dataset` and not to TensorFlow `tf.data.Dataset`.\r\n\r\nOur `datasets.Dataset.shuffle` method does not have a `reshuffle_each_iteration` argument. Therefore, I would say the statement in our docs is True because they refer to `datasets.Dataset.shuffle`, `datasets.Dataset.skip` and `datasets.Dataset.take`.\r\n\r\nI think this issue is restricted to TensorFlow dataset, and this would be addressed by them in the issue you opened in their repo: https://github.com/tensorflow/tensorflow/issues/59279", "Also note that you are referring to an outdated documentation page: datasets 1.10.2 version\r\n\r\nCurrent datasets version is 2.8.0 and the corresponding documentation page is: https://huggingface.co/docs/datasets/stream#split-dataset", "Hi @albertvillanova thanks for your reply and your explaination here. \r\n\r\nSorry for the confusion as I'm not actually a user of your repo and I just happen to find the thread by Google (and didn't read carefully).\r\n\r\nGreat to know that and you made everything very clear now.\r\n\r\nThanks for your time and sorry for the consusion.\r\n\r\nWishing you a wonderful time. \r\n\r\nRegards" ]

[ { "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" } ]

[ "It works if we either remove the `ds = ds.cast_column(\"id\", Value(\"int32\"))` line from the code above, or if instead calling `ds.remove_columns()` we remove the columns inside each mapping as `ds.map(..., remove_columns=[...])` instead of right after the mapping.\r\n\r\nBoth of those solutions seem to fix the issue, so the root cause of it may be around that. Sorry I cannot provide you more insights, in case I get to fix it I'll submit a PR, in the meanwhile the code that I'm using as a workaround is the following:\r\n\r\n```python\r\nfrom transformers import DPRContextEncoder, DPRContextEncoderTokenizer\r\nimport torch\r\n\r\ntorch.set_grad_enabled(False)\r\nctx_encoder = DPRContextEncoder.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\nctx_tokenizer = DPRContextEncoderTokenizer.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n\r\nfrom datasets import load_dataset, Value\r\n\r\nds = load_dataset(\"csv\", data_files=[\"sample.csv\"], split=\"train\")\r\nds = ds.cast_column(\"id\", Value(\"int32\"))\r\nds = ds.map(lambda x: {\"inputs\": f\"{ctx_tokenizer.sep_token}\".join([\"title\", \"summary\"])}, remove_columns=[\"title\", \"summary\"])\r\n\r\ndef generate_embeddings(x):\r\n return {\"embeddings\": ctx_encoder(**ctx_tokenizer(x[\"inputs\"], return_tensors=\"pt\"))[0][0].numpy()}\r\n\r\nds = ds.map(generate_embeddings, remove_columns=[\"inputs\"])\r\nds.add_faiss_index(column=\"embeddings\")\r\n```", "FYI the main reason I want to use `dataset.remove_columns` rather than the function inside `dataset.map` is because according to the 🤗 Datasets documentation, it's faster.\r\n\r\n\"🤗 Datasets also has a [Dataset.remove_columns()](https://huggingface.co/docs/datasets/v2.2.1/en/package_reference/main_classes#datasets.Dataset.remove_columns) method that is functionally identical, but faster, because it doesn’t copy the data of the remaining columns.\"\r\n\r\nMore information at https://huggingface.co/docs/datasets/process#map", "Here I'm presenting all the scenarios so that you can further investigate the issue:\r\n\r\n- ✅ `cast_column` -> `map` with `remove_columns` -> `map` with `remove_columns` -> `add_faiss_index`\r\n\r\n ```python\r\n from transformers import DPRContextEncoder, DPRContextEncoderTokenizer\r\n import torch\r\n \r\n torch.set_grad_enabled(False)\r\n ctx_encoder = DPRContextEncoder.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n ctx_tokenizer = DPRContextEncoderTokenizer.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n \r\n from datasets import load_dataset, Value\r\n \r\n ds = load_dataset(\"csv\", data_files=[\"sample.csv\"], split=\"train\")\r\n ds = ds.cast_column(\"id\", Value(\"int32\"))\r\n ds = ds.map(lambda x: {\"inputs\": f\"{ctx_tokenizer.sep_token}\".join([\"title\", \"summary\"])}, remove_columns=[\"title\", \"summary\"])\r\n \r\n def generate_embeddings(x):\r\n return {\"embeddings\": ctx_encoder(**ctx_tokenizer(x[\"inputs\"], return_tensors=\"pt\"))[0][0].numpy()}\r\n \r\n ds = ds.map(generate_embeddings, remove_columns=[\"inputs\"])\r\n ds.add_faiss_index(column=\"embeddings\")\r\n ```\r\n\r\n- ❌ `cast_column` -> `map` -> `remove_columns` -> `map` -> `remove_columns` -> `add_faiss_index`\r\n\r\n ```python\r\n from transformers import DPRContextEncoder, DPRContextEncoderTokenizer\r\n import torch\r\n \r\n torch.set_grad_enabled(False)\r\n ctx_encoder = DPRContextEncoder.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n ctx_tokenizer = DPRContextEncoderTokenizer.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n \r\n from datasets import load_dataset, Value\r\n \r\n ds = load_dataset(\"csv\", data_files=[\"sample.csv\"], split=\"train\")\r\n ds = ds.cast_column(\"id\", Value(\"int32\"))\r\n ds = ds.map(lambda x: {\"inputs\": f\"{ctx_tokenizer.sep_token}\".join([\"title\", \"summary\"])})\r\n ds = ds.remove_columns([\"title\", \"summary\"])\r\n \r\n def generate_embeddings(x):\r\n return {\"embeddings\": ctx_encoder(**ctx_tokenizer(x[\"inputs\"], return_tensors=\"pt\"))[0][0].numpy()}\r\n \r\n ds = ds.map(generate_embeddings)\r\n ds = ds.remove_columns([\"inputs\"])\r\n ds.add_faiss_index(column=\"embeddings\")\r\n ```\r\n\r\n- ❌ `cast_column` -> `map` with `remove_columns` -> `map` -> `remove_columns` -> `add_faiss_index`\r\n\r\n\r\n ```python\r\n from transformers import DPRContextEncoder, DPRContextEncoderTokenizer\r\n import torch\r\n \r\n torch.set_grad_enabled(False)\r\n ctx_encoder = DPRContextEncoder.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n ctx_tokenizer = DPRContextEncoderTokenizer.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n \r\n from datasets import load_dataset, Value\r\n \r\n ds = load_dataset(\"csv\", data_files=[\"sample.csv\"], split=\"train\")\r\n ds = ds.cast_column(\"id\", Value(\"int32\"))\r\n ds = ds.map(lambda x: {\"inputs\": f\"{ctx_tokenizer.sep_token}\".join([\"title\", \"summary\"])}, remove_columns=[\"title\", \"summary\"])\r\n \r\n def generate_embeddings(x):\r\n return {\"embeddings\": ctx_encoder(**ctx_tokenizer(x[\"inputs\"], return_tensors=\"pt\"))[0][0].numpy()}\r\n \r\n ds = ds.map(generate_embeddings)\r\n ds = ds.remove_columns([\"inputs\"])\r\n ds.add_faiss_index(column=\"embeddings\")\r\n ```\r\n\r\n- ✅ `cast_column` -> `map` -> `remove_columns` -> `map` with `remove_columns` -> `add_faiss_index`\r\n\r\n\r\n ```python\r\n from transformers import DPRContextEncoder, DPRContextEncoderTokenizer\r\n import torch\r\n \r\n torch.set_grad_enabled(False)\r\n ctx_encoder = DPRContextEncoder.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n ctx_tokenizer = DPRContextEncoderTokenizer.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n \r\n from datasets import load_dataset, Value\r\n \r\n ds = load_dataset(\"csv\", data_files=[\"sample.csv\"], split=\"train\")\r\n ds = ds.cast_column(\"id\", Value(\"int32\"))\r\n ds = ds.map(lambda x: {\"inputs\": f\"{ctx_tokenizer.sep_token}\".join([\"title\", \"summary\"])})\r\n ds = ds.remove_columns([\"title\", \"summary\"])\r\n \r\n def generate_embeddings(x):\r\n return {\"embeddings\": ctx_encoder(**ctx_tokenizer(x[\"inputs\"], return_tensors=\"pt\"))[0][0].numpy()}\r\n \r\n ds = ds.map(generate_embeddings, remove_columns=[\"inputs\"])\r\n ds.add_faiss_index(column=\"embeddings\")\r\n ```\r\n\r\n- ✅ `map` -> `remove_columns` -> `map` -> `remove_columns` -> `add_faiss_index`\r\n\r\n\r\n ```python\r\n from transformers import DPRContextEncoder, DPRContextEncoderTokenizer\r\n import torch\r\n \r\n torch.set_grad_enabled(False)\r\n ctx_encoder = DPRContextEncoder.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n ctx_tokenizer = DPRContextEncoderTokenizer.from_pretrained(\"facebook/dpr-ctx_encoder-single-nq-base\")\r\n \r\n from datasets import load_dataset, Value\r\n \r\n ds = load_dataset(\"csv\", data_files=[\"sample.csv\"], split=\"train\")\r\n ds = ds.map(lambda x: {\"inputs\": f\"{ctx_tokenizer.sep_token}\".join([\"title\", \"summary\"])})\r\n ds = ds.remove_columns([\"title\", \"summary\"])\r\n \r\n def generate_embeddings(x):\r\n return {\"embeddings\": ctx_encoder(**ctx_tokenizer(x[\"inputs\"], return_tensors=\"pt\"))[0][0].numpy()}\r\n \r\n ds = ds.map(generate_embeddings)\r\n ds = ds.remove_columns([\"inputs\"])\r\n ds.add_faiss_index(column=\"embeddings\")\r\n ```", "So on, I've created #4411 so as to fix the bug with `remove_columns` under certain conditions before `add_faiss_index`, which means that the scenarios not working above are already working fine." ]

[]

[]

[]

[ "@jplu @thomwolf @patrickvonplaten @lhoestq -- How do I request reviewers? Thanks." ]

[]

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

[]

[]

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

[ "I tried in this way.\r\n\r\n```python\r\nfrom datasets import load_dataset\r\ndataset = load_dataset(path=\"wikitext\", name=\"wikitext-103-v1\", split=\"train\")\r\n```" ]

[]

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

[]

[ "Moved to https://github.com/huggingface/datasets/pull/1315" ]

[]

[ "We have to update the URL in the XTREME benchmark as well:\r\n\r\nhttps://github.com/huggingface/datasets/blob/0dfc639cec450ed8762a997789a2ed63e63cdcf2/datasets/xtreme/xtreme.py#L411-L411\r\n\r\n" ]

[]

[]

[ { "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" }, { "color": "d93f0b", "default": false, "description": "", "id": 2725241052, "name": "speech", "node_id": "MDU6TGFiZWwyNzI1MjQxMDUy", "url": "https://api.github.com/repos/huggingface/datasets/labels/speech" } ]

[]

[]

[]

[]

[ "_The documentation is not available anymore as the PR was closed or merged._", "Superseded by huggingface/doc-builder@31fe6c8bc7225810e281c2f6c6cd32f38828c504" ]

[]

[ "Hi :) \r\nIn your dummy data zip file you can just have `subset000.xz` as directories instead of compressed files.\r\nLet me know if it helps", "Thanks for your comment @lhoestq ,\r\nJust for confirmation, changing dummy data like this won't make dummy test test the functionality to extract `subsetxxx.xz` but actually kind of circumvent it. But since we will test the real data so it is ok ?", "Yes it's fine for now. We plan to add a job for slow tests.\r\nAnd at one point we'll also do another pass on the dummy data handling and consider extracting files.", "Thanks for the confirmation.\r\nAlso the suggestion works. Thank you." ]

[]

[ "@lhoestq Thank you for all the feedback. I've added the suggested changes in my last commit." ]

[ { "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" } ]

[ "#self-assign" ]

[]

[ "Hi, thanks for the feedback! Would it help to add a tip or note saying the dataset is gated and you need to accept the license before downloading it?", "The error is now more informative:\r\n```\r\nFileNotFoundError: Couldn't find a dataset script at /content/imagenet-1k/imagenet-1k.py or any data file in the same directory. Couldn't find 'imagenet-1k' on the Hugging Face Hub either: FileNotFoundError: Dataset 'imagenet-1k' doesn't exist on the Hub. If the repo is private or gated, make sure to log in with `huggingface-cli login`.\r\n```\r\n\r\n" ]

[]

[ "Hi !\r\nAbout the problems you mentioned:\r\n- Saving the infos is only done for the configurations inside the BUILDER_CONFIGS. Otherwise you would need to run the scripts on ALL language pairs, which is not what we want.\r\n- Moreover when you're on your branch, please specify the path to your local version of the dataset script, like \"./datasets/open_subtitles\". Otherwise the dataset is loaded from the master branch on github.\r\nHope that clarifies things a bit\r\n\r\nAnd of course feel free to add methods or classmethods to your builder.\r\n", "Great! Thank you :)\r\nI'll close the issue as well." ]

[]

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

[]

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

[]

[ "Thanks for reporting, @TJ-Solergibert.\r\n\r\nWe cannot access your Colab notebook: `There was an error loading this notebook. Ensure that the file is accessible and try again.`\r\nCould you please make it publicly accessible?\r\n", "I swear it's public, I've checked the settings and I've been able to open it in incognito mode.\r\n\r\nNotebook: https://colab.research.google.com/drive/1JCrS7FlGfu_kFqChMrwKZ_bpabnIMqbP?usp=sharing\r\n\r\nAnyway, this is the code to reproduce the error:\r\n\r\n```python3\r\nfrom datasets import ClassLabel\r\nfrom datasets import load_dataset\r\n\r\neuroparl_ds = load_dataset(\"tj-solergibert/Europarl-ST\")\r\n\r\nsource_lang = \"nl\"\r\nlanguages = list(europarl_ds[\"train\"][0][\"transcriptions\"].keys())\r\nClassLabels = ClassLabel(num_classes = len(languages), names = languages)\r\n\r\ndef map_label2id(example):\r\n example['dest_lang'] = ClassLabels.str2int(example['dest_lang'])\r\n return example\r\n\r\ndef unfold_transcriptions(example):\r\n for lang in languages:\r\n example[lang] = example[\"transcriptions\"][lang]\r\n return example\r\n\r\ndef unroll(batch, src_lang, dest_langs):\r\n source_t, dest_t, dest_l = [], [], []\r\n for lang in dest_langs: \r\n source_t += batch[src_lang]\r\n dest_t += batch[lang]\r\n dest_l += [lang]\r\n return_dict = {\"source_text\": source_t, \"dest_text\": dest_t, \"dest_lang\": dest_l}\r\n return return_dict\r\n\r\ndef preprocess_split(ds_split, src_lang):\r\n dest_langs = [x for x in languages if x != src_lang]\r\n\r\n ds_split = ds_split.map(unroll, fn_kwargs= {\"src_lang\": src_lang, \"dest_langs\": dest_langs}, batched = True, batch_size = 1, remove_columns= list(languages))\r\n ds_split = ds_split.filter(lambda x: x[\"source_text\"] != None and x[\"dest_text\"] != None) # Remove incomplete translations\r\n ds_split = ds_split.filter(lambda x: x[\"source_text\"] != \"None\" and x[\"dest_text\"] != \"None\")\r\n ds_split = ds_split.map(map_label2id) \r\n ds_split = ds_split.cast_column(\"dest_lang\", ClassLabels)\r\n return ds_split\r\n\r\ndef reset_cortas(example):\r\n for lang in languages:\r\n if isinstance(example[lang], str):\r\n if example[lang].isnumeric () or len(example[lang]) <= 5:\r\n example[lang] = \"None\"\r\n return example\r\n\r\ndef clean_dataset(dataset):\r\n # Remove columns\r\n dataset = dataset.remove_columns([\"original_speech\", \"original_language\", \"audio_path\", \"segment_start\", \"segment_end\"])\r\n # Unfold\r\n dataset = dataset.map(unfold_transcriptions, remove_columns = [\"transcriptions\"])\r\n dataset = dataset.map(reset_cortas)\r\n return dataset\r\n\r\nprocessed_europarl = clean_dataset(europarl_ds[\"test\"])\r\nnew_train_ds = preprocess_split(processed_europarl, 'nl')\r\n```", "Thanks, @TJ-Solergibert. I can access your notebook now. Maybe it was just a temporary issue.\r\n\r\nAt first sight, it seems something related to your data: maybe some of the examples do not have all the transcriptions for all the languages. Then, some of them are null when unrolled. And when trying to concatenate with the other rows containing strings, the cast issue is raised (the arrays to be concatenated have different types).\r\n\r\nDo you think this could be the case?", "See, in this example, \"nl\" and \"ro\" transcripts are null:\r\n```python\r\n>>> europarl_ds[\"test\"][:1]\r\n{'original_speech': ['− Señor Presidente, en primer lugar, quisiera felicitar al señor Seeber por el trabajo realizado, porque en su informe se recogen muchas de las preocupaciones manifestadas en esta'],\r\n 'original_language': ['es'],\r\n 'audio_path': ['es/audios/en.20081008.24.3-238.m4a'],\r\n 'segment_start': [0.6200000047683716],\r\n 'segment_end': [11.319999694824219],\r\n 'transcriptions': [{'de': '− Herr Präsident! Zunächst möchte ich Richard Seeber zu der von ihm geleisteten Arbeit gratulieren, denn sein Bericht greift viele der in diesem Haus zum Ausdruck gebrachten Anliegen',\r\n 'en': '− Mr President, firstly I would like to congratulate Mr Seeber on the work he has done, because his report picks up many of the concerns expressed in this',\r\n 'es': '− Señor Presidente, en primer lugar, quisiera felicitar al señor Seeber por el trabajo realizado, porque en su informe se recogen muchas de las preocupaciones manifestadas en esta',\r\n 'fr': '− Monsieur le Président, je voudrais tout d ’ abord féliciter M. Seeber pour le travail qu ’ il a effectué, parce que son rapport reprend beaucoup des inquiétudes exprimées au sein de cette',\r\n 'it': \"− Signor Presidente, mi congratulo innanzi tutto con l'onorevole Seeber per il lavoro svolto, perché la sua relazione accoglie molti dei timori espressi da quest'Aula\",\r\n 'nl': None,\r\n 'pl': '− Panie przewodniczący! Po pierwsze chciałabym pogratulować panu posłowi Seeberowi wykonanej pracy, ponieważ jego sprawozdanie podejmuje szereg podnoszonych w tej Izbie',\r\n 'pt': '− Senhor Presidente, começo por felicitar o senhor deputado Seeber pelo trabalho que desenvolveu em torno deste relatório, que retoma muitas das preocupações expressas nesta',\r\n 'ro': None}]}\r\n```\r\n```python\r\n>>> processed_europarl[0]\r\n{'de': '− Herr Präsident! Zunächst möchte ich Richard Seeber zu der von ihm geleisteten Arbeit gratulieren, denn sein Bericht greift viele der in diesem Haus zum Ausdruck gebrachten Anliegen',\r\n 'en': '− Mr President, firstly I would like to congratulate Mr Seeber on the work he has done, because his report picks up many of the concerns expressed in this',\r\n 'es': '− Señor Presidente, en primer lugar, quisiera felicitar al señor Seeber por el trabajo realizado, porque en su informe se recogen muchas de las preocupaciones manifestadas en esta',\r\n 'fr': '− Monsieur le Président, je voudrais tout d ’ abord féliciter M. Seeber pour le travail qu ’ il a effectué, parce que son rapport reprend beaucoup des inquiétudes exprimées au sein de cette',\r\n 'it': \"− Signor Presidente, mi congratulo innanzi tutto con l'onorevole Seeber per il lavoro svolto, perché la sua relazione accoglie molti dei timori espressi da quest'Aula\",\r\n 'nl': None,\r\n 'pl': '− Panie przewodniczący! Po pierwsze chciałabym pogratulować panu posłowi Seeberowi wykonanej pracy, ponieważ jego sprawozdanie podejmuje szereg podnoszonych w tej Izbie',\r\n 'pt': '− Senhor Presidente, começo por felicitar o senhor deputado Seeber pelo trabalho que desenvolveu em torno deste relatório, que retoma muitas das preocupações expressas nesta',\r\n 'ro': None}\r\n```", "You can fix this issue by forcing the cast of None to str by hand:\r\n- If you replace this line:\r\n```python\r\nsource_t += batch[src_lang]\r\n```\r\n- With this line (because the batch size is 1):\r\n```python\r\nsource_t += [str(batch[src_lang][0])]\r\n```\r\n- Or with this line (if the batch size were larger than 1):\r\n```python\r\nsource_t += [str(text) for text in batch[src_lang]]\r\n```", "Problem solved! Thanks @albertvillanova, now I have even increased the batch size and it's crazy fast :rocket: !" ]

[]

[]

[ { "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" } ]

[]

[]

[ "_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) | 0.006087 / 0.011353 (-0.005265) | 0.003567 / 0.011008 (-0.007442) | 0.079668 / 0.038508 (0.041160) | 0.063647 / 0.023109 (0.040538) | 0.323082 / 0.275898 (0.047184) | 0.348679 / 0.323480 (0.025199) | 0.004726 / 0.007986 (-0.003259) | 0.002955 / 0.004328 (-0.001373) | 0.062724 / 0.004250 (0.058473) | 0.050194 / 0.037052 (0.013142) | 0.321407 / 0.258489 (0.062918) | 0.355053 / 0.293841 (0.061212) | 0.026992 / 0.128546 (-0.101554) | 0.007994 / 0.075646 (-0.067653) | 0.260562 / 0.419271 (-0.158710) | 0.050933 / 0.043533 (0.007400) | 0.316644 / 0.255139 (0.061505) | 0.336759 / 0.283200 (0.053560) | 0.022581 / 0.141683 (-0.119101) | 1.481259 / 1.452155 (0.029104) | 1.535191 / 1.492716 (0.042475) |\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) | 0.194111 / 0.018006 (0.176104) | 0.448146 / 0.000490 (0.447656) | 0.000321 / 0.000200 (0.000121) | 0.000056 / 0.000054 (0.000001) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.023908 / 0.037411 (-0.013503) | 0.073316 / 0.014526 (0.058790) | 0.085588 / 0.176557 (-0.090968) | 0.145377 / 0.737135 (-0.591759) | 0.084788 / 0.296338 (-0.211550) |\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) | 0.439327 / 0.215209 (0.224118) | 4.384833 / 2.077655 (2.307179) | 2.322943 / 1.504120 (0.818823) | 2.147737 / 1.541195 (0.606542) | 2.226725 / 1.468490 (0.758235) | 0.502957 / 4.584777 (-4.081820) | 3.098106 / 3.745712 (-0.647606) | 4.194642 / 5.269862 (-1.075220) | 2.598820 / 4.565676 (-1.966856) | 0.057942 / 0.424275 (-0.366333) | 0.006857 / 0.007607 (-0.000750) | 0.511517 / 0.226044 (0.285472) | 5.121797 / 2.268929 (2.852868) | 2.756506 / 55.444624 (-52.688118) | 2.424602 / 6.876477 (-4.451875) | 2.608342 / 2.142072 (0.466270) | 0.589498 / 4.805227 (-4.215729) | 0.126065 / 6.500664 (-6.374600) | 0.061456 / 0.075469 (-0.014013) |\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) | 1.239928 / 1.841788 (-0.601860) | 18.423532 / 8.074308 (10.349224) | 13.935148 / 10.191392 (3.743756) | 0.129913 / 0.680424 (-0.550511) | 0.016744 / 0.534201 (-0.517457) | 0.333468 / 0.579283 (-0.245815) | 0.359615 / 0.434364 (-0.074749) | 0.383678 / 0.540337 (-0.156659) | 0.533007 / 1.386936 (-0.853929) |\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) | 0.005980 / 0.011353 (-0.005373) | 0.003640 / 0.011008 (-0.007368) | 0.062500 / 0.038508 (0.023992) | 0.059843 / 0.023109 (0.036733) | 0.360993 / 0.275898 (0.085095) | 0.401981 / 0.323480 (0.078501) | 0.005495 / 0.007986 (-0.002490) | 0.002862 / 0.004328 (-0.001467) | 0.062491 / 0.004250 (0.058240) | 0.050778 / 0.037052 (0.013726) | 0.371007 / 0.258489 (0.112518) | 0.405154 / 0.293841 (0.111313) | 0.027390 / 0.128546 (-0.101156) | 0.008042 / 0.075646 (-0.067604) | 0.067590 / 0.419271 (-0.351681) | 0.042485 / 0.043533 (-0.001048) | 0.361305 / 0.255139 (0.106166) | 0.388669 / 0.283200 (0.105469) | 0.024143 / 0.141683 (-0.117540) | 1.451508 / 1.452155 (-0.000647) | 1.490431 / 1.492716 (-0.002285) |\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) | 0.175976 / 0.018006 (0.157970) | 0.428923 / 0.000490 (0.428434) | 0.002099 / 0.000200 (0.001899) | 0.000068 / 0.000054 (0.000014) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.026346 / 0.037411 (-0.011065) | 0.078084 / 0.014526 (0.063558) | 0.087287 / 0.176557 (-0.089269) | 0.144179 / 0.737135 (-0.592957) | 0.088286 / 0.296338 (-0.208053) |\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) | 0.450436 / 0.215209 (0.235227) | 4.488801 / 2.077655 (2.411146) | 2.479303 / 1.504120 (0.975184) | 2.305396 / 1.541195 (0.764201) | 2.370370 / 1.468490 (0.901879) | 0.502355 / 4.584777 (-4.082422) | 3.094733 / 3.745712 (-0.650979) | 4.062367 / 5.269862 (-1.207495) | 2.587506 / 4.565676 (-1.978170) | 0.058245 / 0.424275 (-0.366030) | 0.006487 / 0.007607 (-0.001120) | 0.524147 / 0.226044 (0.298102) | 5.236876 / 2.268929 (2.967947) | 2.897134 / 55.444624 (-52.547490) | 2.574631 / 6.876477 (-4.301846) | 2.620307 / 2.142072 (0.478235) | 0.586963 / 4.805227 (-4.218265) | 0.125761 / 6.500664 (-6.374903) | 0.062264 / 0.075469 (-0.013205) |\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) | 1.299668 / 1.841788 (-0.542120) | 19.004441 / 8.074308 (10.930133) | 13.841867 / 10.191392 (3.650475) | 0.159674 / 0.680424 (-0.520750) | 0.016699 / 0.534201 (-0.517502) | 0.331868 / 0.579283 (-0.247415) | 0.344604 / 0.434364 (-0.089760) | 0.379391 / 0.540337 (-0.160947) | 0.514790 / 1.386936 (-0.872146) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#47a006a90e9711b33db70b0ef2d2cefaadfa2179 \"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) | 0.005792 / 0.011353 (-0.005561) | 0.003519 / 0.011008 (-0.007489) | 0.079133 / 0.038508 (0.040625) | 0.057858 / 0.023109 (0.034749) | 0.314206 / 0.275898 (0.038308) | 0.346939 / 0.323480 (0.023459) | 0.004583 / 0.007986 (-0.003403) | 0.002824 / 0.004328 (-0.001504) | 0.061652 / 0.004250 (0.057402) | 0.048520 / 0.037052 (0.011467) | 0.318018 / 0.258489 (0.059529) | 0.350350 / 0.293841 (0.056509) | 0.026284 / 0.128546 (-0.102262) | 0.007827 / 0.075646 (-0.067819) | 0.259624 / 0.419271 (-0.159647) | 0.052318 / 0.043533 (0.008786) | 0.317400 / 0.255139 (0.062261) | 0.340530 / 0.283200 (0.057331) | 0.025181 / 0.141683 (-0.116501) | 1.459208 / 1.452155 (0.007053) | 1.529158 / 1.492716 (0.036442) |\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) | 0.169692 / 0.018006 (0.151686) | 0.432638 / 0.000490 (0.432148) | 0.003675 / 0.000200 (0.003475) | 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) | 0.022956 / 0.037411 (-0.014456) | 0.071860 / 0.014526 (0.057334) | 0.082159 / 0.176557 (-0.094398) | 0.142560 / 0.737135 (-0.594576) | 0.082333 / 0.296338 (-0.214006) |\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) | 0.397923 / 0.215209 (0.182714) | 3.958757 / 2.077655 (1.881102) | 1.925837 / 1.504120 (0.421717) | 1.758114 / 1.541195 (0.216919) | 1.808845 / 1.468490 (0.340354) | 0.501116 / 4.584777 (-4.083661) | 3.007739 / 3.745712 (-0.737973) | 3.295755 / 5.269862 (-1.974106) | 2.123843 / 4.565676 (-2.441833) | 0.057174 / 0.424275 (-0.367101) | 0.006426 / 0.007607 (-0.001182) | 0.468196 / 0.226044 (0.242152) | 4.677392 / 2.268929 (2.408464) | 2.334179 / 55.444624 (-53.110446) | 1.989283 / 6.876477 (-4.887194) | 2.140091 / 2.142072 (-0.001981) | 0.590700 / 4.805227 (-4.214527) | 0.124066 / 6.500664 (-6.376598) | 0.059931 / 0.075469 (-0.015538) |\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) | 1.224547 / 1.841788 (-0.617240) | 17.866979 / 8.074308 (9.792671) | 13.142009 / 10.191392 (2.950617) | 0.147081 / 0.680424 (-0.533343) | 0.016777 / 0.534201 (-0.517424) | 0.327766 / 0.579283 (-0.251517) | 0.343988 / 0.434364 (-0.090376) | 0.383268 / 0.540337 (-0.157070) | 0.528109 / 1.386936 (-0.858827) |\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) | 0.006145 / 0.011353 (-0.005208) | 0.003634 / 0.011008 (-0.007374) | 0.062887 / 0.038508 (0.024379) | 0.062659 / 0.023109 (0.039550) | 0.362962 / 0.275898 (0.087064) | 0.405149 / 0.323480 (0.081669) | 0.004821 / 0.007986 (-0.003164) | 0.002888 / 0.004328 (-0.001441) | 0.062982 / 0.004250 (0.058732) | 0.051929 / 0.037052 (0.014877) | 0.366825 / 0.258489 (0.108336) | 0.409830 / 0.293841 (0.115989) | 0.027263 / 0.128546 (-0.101283) | 0.007972 / 0.075646 (-0.067674) | 0.067413 / 0.419271 (-0.351858) | 0.044233 / 0.043533 (0.000700) | 0.365087 / 0.255139 (0.109948) | 0.393845 / 0.283200 (0.110646) | 0.027740 / 0.141683 (-0.113943) | 1.497896 / 1.452155 (0.045741) | 1.549419 / 1.492716 (0.056703) |\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) | 0.225510 / 0.018006 (0.207503) | 0.417054 / 0.000490 (0.416564) | 0.002184 / 0.000200 (0.001984) | 0.000075 / 0.000054 (0.000020) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.025503 / 0.037411 (-0.011908) | 0.076164 / 0.014526 (0.061638) | 0.086110 / 0.176557 (-0.090446) | 0.140387 / 0.737135 (-0.596748) | 0.086956 / 0.296338 (-0.209382) |\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) | 0.469667 / 0.215209 (0.254458) | 4.689915 / 2.077655 (2.612261) | 2.685000 / 1.504120 (1.180880) | 2.516160 / 1.541195 (0.974965) | 2.531733 / 1.468490 (1.063243) | 0.501675 / 4.584777 (-4.083102) | 3.000579 / 3.745712 (-0.745133) | 2.853376 / 5.269862 (-2.416486) | 1.810677 / 4.565676 (-2.754999) | 0.057632 / 0.424275 (-0.366643) | 0.006390 / 0.007607 (-0.001217) | 0.543986 / 0.226044 (0.317941) | 5.432837 / 2.268929 (3.163908) | 3.138797 / 55.444624 (-52.305827) | 2.813141 / 6.876477 (-4.063336) | 2.803681 / 2.142072 (0.661609) | 0.588736 / 4.805227 (-4.216491) | 0.125696 / 6.500664 (-6.374968) | 0.062492 / 0.075469 (-0.012977) |\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) | 1.337163 / 1.841788 (-0.504624) | 18.611715 / 8.074308 (10.537407) | 13.953016 / 10.191392 (3.761624) | 0.154670 / 0.680424 (-0.525754) | 0.016523 / 0.534201 (-0.517678) | 0.333898 / 0.579283 (-0.245385) | 0.336520 / 0.434364 (-0.097844) | 0.389032 / 0.540337 (-0.151305) | 0.529202 / 1.386936 (-0.857734) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#01d4b3330f2cc243a3f3b0cd61ec5558466c40fd \"CML watermark\")\n" ]

[]

[ "Hi! Can you please paste the entire error stack trace, not only the last few lines?", "`----> 1 dataset = datasets.load_dataset(\"glue\", \"ax\")\r\n\r\nFile ~\\anaconda3\\envs\\huggingface\\Lib\\site-packages\\datasets\\load.py:1767, in load_dataset(path, name, data_dir, data_files, split, cache_dir, features, download_config, download_mode, verification_mode, ignore_verifications, keep_in_memory, save_infos, revision, use_auth_token, task, streaming, num_proc, storage_options, **config_kwargs)\r\n 1762 verification_mode = VerificationMode(\r\n 1763 (verification_mode or VerificationMode.BASIC_CHECKS) if not save_infos else VerificationMode.ALL_CHECKS\r\n 1764 )\r\n 1766 # Create a dataset builder\r\n-> 1767 builder_instance = load_dataset_builder(\r\n 1768 path=path,\r\n 1769 name=name,\r\n 1770 data_dir=data_dir,\r\n 1771 data_files=data_files,\r\n 1772 cache_dir=cache_dir,\r\n 1773 features=features,\r\n 1774 download_config=download_config,\r\n 1775 download_mode=download_mode,\r\n 1776 revision=revision,\r\n 1777 use_auth_token=use_auth_token,\r\n 1778 storage_options=storage_options,\r\n 1779 **config_kwargs,\r\n 1780 )\r\n 1782 # Return iterable dataset in case of streaming\r\n 1783 if streaming:\r\n\r\nFile ~\\anaconda3\\envs\\huggingface\\Lib\\site-packages\\datasets\\load.py:1498, in load_dataset_builder(path, name, data_dir, data_files, cache_dir, features, download_config, download_mode, revision, use_auth_token, storage_options, **config_kwargs)\r\n 1496 download_config = download_config.copy() if download_config else DownloadConfig()\r\n 1497 download_config.use_auth_token = use_auth_token\r\n-> 1498 dataset_module = dataset_module_factory(\r\n 1499 path,\r\n 1500 revision=revision,\r\n 1501 download_config=download_config,\r\n 1502 download_mode=download_mode,\r\n 1503 data_dir=data_dir,\r\n 1504 data_files=data_files,\r\n 1505 )\r\n 1507 # Get dataset builder class from the processing script\r\n 1508 builder_cls = import_main_class(dataset_module.module_path)\r\n\r\nFile ~\\anaconda3\\envs\\huggingface\\Lib\\site-packages\\datasets\\load.py:1211, in dataset_module_factory(path, revision, download_config, download_mode, dynamic_modules_path, data_dir, data_files, **download_kwargs)\r\n 1209 raise e1 from None\r\n 1210 if isinstance(e1, FileNotFoundError):\r\n-> 1211 raise FileNotFoundError(\r\n 1212 f\"Couldn't find a dataset script at {relative_to_absolute_path(combined_path)} or any data file in the same directory. \"\r\n 1213 f\"Couldn't find '{path}' on the Hugging Face Hub either: {type(e1).__name__}: {e1}\"\r\n 1214 ) from None\r\n 1215 raise e1 from None\r\n 1216 else:`", "Okay, this is the issue:\r\n```\r\nFileNotFoundError: [WinError 3] The system cannot find the path specified: \r\n'C:\\\\Users\\\\...\\\\.cache\\\\huggingface'\r\n``` \r\n\r\nI don't remember seeing this error before.\r\n\r\nI guess it could happen in a multi-process environment if one of the processes deletes the `datasets` cache as the other one is loading a dataset (with `load_dataset`), so make sure that's not the case. Also, you can disable the Windows max path length limit (if enabled), but this is most likely not the problem.", "Closing due to inactivity." ]

[]

[ "Looks like a flaky CI, failed download from S3." ]

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

[ "the same issue with datasets.load_dataset(\"iwslt2017\", 'iwslt2017-en-nl', split=split), ..... ", "even with setting master like the following command, still remains \r\n\r\ndatasets.load_dataset(\"iwslt2017\", 'iwslt2017-en-nl', split=\"train\", script_version=\"master\")\r\n", "Looks like the data has been moved from its original location to google drive\r\n\r\nNew url: https://drive.google.com/u/0/uc?id=12ycYSzLIG253AFN35Y6qoyf9wtkOjakp&export=download", "Fixed by #4481 " ]

[]

[]

[]

[ "Unfortunately it's not possible to download only the dev set of NQ.\r\n\r\nI think we could add a way to download only the test set by adding a custom configuration to the processing script though.", "Ok, got it. I think this could be a valuable feature - especially for large datasets like NQ, but potentially also others. \r\nFor us, it will in this case make the difference of using the library or keeping the old downloads of the raw dev datasets. \r\nHowever, I don't know if that fits into your plans with the library and can also understand if you don't want to support this.", "I don't think we could force this behavior generally since the dataset script authors are free to organize the file download as they want (sometimes the mapping between split and files can be very much nontrivial) but we can add an additional configuration for Natural Question indeed as @lhoestq indicate." ]

[]

[]

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

[ "for reference Kaggle has a good number of open source datasets stored in sqlite\r\n\r\nAlternatively a tutorial or tool on how to convert from sqlite to parquet would be cool too", "Hello, could we leverage [`pandas.read_sql`](https://pandas.pydata.org/docs/reference/api/pandas.read_sql.html) for this? \r\n\r\nThis would be basically the same as [`CSVBuilder`](https://github.com/huggingface/datasets/blob/7380140accf522a4363bb56c0b77a4190f49bed6/src/datasets/packaged_modules/csv/csv.py#L127)\r\n, but uses `pandas.read_sql(..., chunksize=1)` instead of `pandas.read_csv(..., iterator=True)`\r\n\r\nI'm happy to work on this :) \r\n\r\nself-assign" ]

[ { "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 ! cc @Rocketknight1 maybe you know better ?\r\n\r\nI'm not too familiar with `tf.data.experimental.save`. Note that `datasets` uses memory mapping, so depending on your hardware and the disk you are using you can expect performance differences with a dataset loaded in RAM", "Hi @lehrig, I suspect what's happening here is that our `to_tf_dataset()` method has some performance issues when streaming samples. This is usually not a problem, but they become apparent when streaming a vision dataset into a very small vision model, which will need a lot of sample throughput to saturate the GPU.\r\n\r\nWhen you save a `tf.data.Dataset` with `tf.data.experimental.save`, all of the samples from the dataset (which are, in this case, batches of images), are saved to disk. When you load this saved dataset, you're effectively bypassing `to_tf_dataset()` entirely, which alleviates this performance bottleneck.\r\n\r\n`to_tf_dataset()` is something we're actively working on overhauling right now - particularly for image datasets, we want to make it possible to access the underlying images with `tf.data` without going through the current layer of indirection with `Arrow`, which should massively improve simplicity and performance. \r\n\r\nHowever, if you just want this to work quickly but without needing your save/load hack, my advice would be to simply load the dataset into memory if it's small enough to fit. Since all your samples have the same dimensions, you can do this simply with:\r\n\r\n```\r\ndataset = load_from_disk(prep_data_dir)\r\ndataset = dataset.with_format(\"numpy\")\r\ndata_in_memory = dataset[:]\r\n```\r\n\r\nThen you can simply do something like:\r\n\r\n```\r\nmodel.fit(data_in_memory[\"pixel_values\"], data_in_memory[\"labels\"])\r\n```", "Thanks for the information! \r\n\r\nI have now updated the training code like so:\r\n\r\n```\r\ndataset = load_from_disk(prep_data_dir)\r\ntrain_dataset = dataset[\"train\"][:]\r\nvalidation_dataset = dataset[\"dev\"][:]\r\n\r\n...\r\n\r\nmodel.fit(\r\n train_dataset[\"pixel_values\"],\r\n train_dataset[\"label\"],\r\n epochs=epochs,\r\n validation_data=(\r\n validation_dataset[\"pixel_values\"],\r\n validation_dataset[\"label\"]\r\n ),\r\n callbacks=[earlyStopping, mcp_save, reduce_lr_loss]\r\n)\r\n```\r\n\r\n- Creating the in-memory dataset is quite quick\r\n- But: There is now a long wait (~4-5 Minutes) before the training starts (why?)\r\n- And: Training times have improved but the very first epoch leaves me wondering why it takes so long (why?)\r\n\r\n**Epoch Breakdown:**\r\n- Epoch 1/10\r\n78s 12s/step - loss: 3.1307 - accuracy: 0.0737 - val_loss: 2.2827 - val_accuracy: 0.1273 - lr: 0.0010\r\n- Epoch 2/10\r\n1s 168ms/step - loss: 2.3616 - accuracy: 0.2350 - val_loss: 2.2679 - val_accuracy: 0.2182 - lr: 0.0010\r\n- Epoch 3/10\r\n1s 189ms/step - loss: 2.0221 - accuracy: 0.3180 - val_loss: 2.2670 - val_accuracy: 0.1818 - lr: 0.0010\r\n- Epoch 4/10\r\n0s 67ms/step - loss: 1.8895 - accuracy: 0.3548 - val_loss: 2.2771 - val_accuracy: 0.1273 - lr: 0.0010\r\n- Epoch 5/10\r\n0s 67ms/step - loss: 1.7846 - accuracy: 0.3963 - val_loss: 2.2860 - val_accuracy: 0.1455 - lr: 0.0010\r\n- Epoch 6/10\r\n0s 65ms/step - loss: 1.5946 - accuracy: 0.4516 - val_loss: 2.2938 - val_accuracy: 0.1636 - lr: 0.0010\r\n- Epoch 7/10\r\n0s 63ms/step - loss: 1.4217 - accuracy: 0.5115 - val_loss: 2.2968 - val_accuracy: 0.2182 - lr: 0.0010\r\n- Epoch 8/10\r\n0s 67ms/step - loss: 1.3089 - accuracy: 0.5438 - val_loss: 2.2842 - val_accuracy: 0.2182 - lr: 0.0010\r\n- Epoch 9/10\r\n1s 184ms/step - loss: 1.2480 - accuracy: 0.5806 - val_loss: 2.2652 - val_accuracy: 0.1818 - lr: 0.0010\r\n- Epoch 10/10\r\n0s 65ms/step - loss: 1.2699 - accuracy: 0.5622 - val_loss: 2.2670 - val_accuracy: 0.2000 - lr: 0.0010\r\n\r\n", "Regarding the new long ~5 min. wait introduced by the in-memory dataset update: this might be causing it? https://datascience.stackexchange.com/questions/33364/why-model-fit-generator-in-keras-is-taking-so-much-time-even-before-picking-the\r\n\r\nFor now, my save/load hack is still more performant, even though having more boiler-plate code :/ ", "That 5 minute wait is quite surprising! I don't have a good explanation for why it's happening, but it can't be an issue with `datasets` or `tf.data` because you're just fitting directly on Numpy arrays at this point. All I can suggest is seeing if you can isolate the issue - for example, does fitting on a smaller dataset containing only 10% of the original data reduce the wait? This might indicate the delay is caused by your data being copied or converted somehow. Alternatively, you could try removing things like callbacks and seeing if you could isolate the issue there." ]

[]

[]

[ { "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" } ]

[ "You are likely trying to get the `features` from a `DatasetDict`, a dictionary containing `Datasets`. You probably first want to index into a particular split from your `DatasetDict` i.e. `dataset['train'].features`. \r\n\r\nFor example \r\n\r\n```python \r\nds = load_dataset('mnist')\r\nds.features\r\n```\r\nReturns \r\n```python\r\n---------------------------------------------------------------------------\r\n\r\nAttributeError Traceback (most recent call last)\r\n\r\n[<ipython-input-39-791c1f9df6c2>](https://localhost:8080/#) in <module>()\r\n----> 1 ds.features\r\n\r\nAttributeError: 'DatasetDict' object has no attribute 'features'\r\n```\r\n\r\nIf we look at the dataset variable, we see it is a `DatasetDict`:\r\n\r\n```python \r\nprint(ds)\r\nDatasetDict({\r\n train: Dataset({\r\n features: ['image', 'label'],\r\n num_rows: 60000\r\n })\r\n test: Dataset({\r\n features: ['image', 'label'],\r\n num_rows: 10000\r\n })\r\n})\r\n```\r\n\r\nWe can grab the features from a split by indexing into `train`:\r\n```python\r\nds['train'].features\r\n{'image': Image(decode=True, id=None),\r\n 'label': ClassLabel(num_classes=10, names=['0', '1', '2', '3', '4', '5', '6', '7', '8', '9'], id=None)}\r\n```\r\n\r\nHope that helps ", "Yes, Thanks for that clarification," ]

[]

[ "Also, mayjor issue for me is the format issue, even if I go through changing the whole code to use load_from_disk, then if I do \r\n\r\nd = datasets.load_from_disk(\"imdb\")\r\nd = d[\"train\"][:10] => the format of this is no more in datasets format\r\nthis is different from you call load_datasets(\"train[10]\")\r\n\r\ncould you tell me how I can make the two datastes the same format @lhoestq \r\n\r\n", "> `\r\nrequests.exceptions.ConnectTimeout: HTTPSConnectionPool(host='s3.amazonaws.com', port=443): Max retries exceeded with url: /datasets.huggingface.co/datasets/datasets/glue/glue.py (Caused by ConnectTimeoutError(<urllib3.connection.HTTPSConnection object at 0x7ff6d6c60a20>, 'Connection to s3.amazonaws.com timed out. (connect timeout=10)'))`\r\n\r\nDo you have an internet connection on the machine ? Is there a proxy that might block requests to aws ?\r\n\r\n> I tried to do read the data, save it to a path and then set HF_HOME, which does not work and this is still not reading from the old set path, could you assist me how to save the datasets in a path, and let dataset library read from this path to avoid connection issue. thanks\r\n\r\nHF_HOME is used to specify the directory for the cache files of this library.\r\nYou can use save_to_disk and load_from_disk without changing the HF_HOME:\r\n```python\r\nimdb = datasets.load_dataset(\"imdb\")\r\nimdb.save_to_disk(\"/idiap/temp/rkarimi/hf_datasets/imdb\")\r\nimdb = datasets.load_from_disk(\"/idiap/temp/rkarimi/hf_datasets/imdb\")\r\n```\r\n\r\n> could you tell me how I can make the two datastes the same format\r\n\r\nIndeed they returns different things:\r\n- `load_dataset` returns a `Dataset` object if the split is specified, or a `DatasetDict` if no split is given. Therefore `load_datasets(\"imdb\", split=\"train[10]\")` returns a `Dataset` object containing 10 elements.\r\n- doing `d[\"train\"][:10]` on a DatasetDict \"d\" gets the train split `d[\"train\"]` as a `Dataset` object and then gets the first 10 elements as a dictionary" ]

[]

[]

[ { "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" } ]

[ "https://github.com/huggingface/datasets/blob/6c766f9115d686182d76b1b937cb27e099c45d68/src/datasets/builder.py#L179-L186", "Thanks a lot!!!" ]

[]

[ "Could you provide more details ? What's the code you ran ?", "```python\r\ntokenizer = FunnelTokenizer.from_pretrained('funnel-transformer/small')\r\n\r\ndef tokenize(batch):\r\n return tokenizer(batch['text'], padding='max_length', truncation=True,max_length=512)\r\n\r\ndataset = load_dataset(\"bookcorpus\",split='train[:1000]').shuffle()\r\ndataset = dataset.map(tokenize, batched=True, batch_size=512)\r\n\r\n# dataset = LineByLineTextDataset(\r\n# tokenizer=tokenizer,\r\n# file_path=\"./wiki1000.txt\",\r\n# block_size=128\r\n# )\r\n\r\ndata_collator = DataCollatorForLanguageModeling(\r\n tokenizer=tokenizer, mlm=True, mlm_probability=0.15\r\n)\r\n\r\nconfig=FunnelConfig(\r\n return_dict=True\r\n)\r\n\r\nmodel= FunnelForMaskedLM(config=config)\r\n\r\ntraining_args = TrainingArguments(\r\n output_dir=\"./checkpoints\",\r\n overwrite_output_dir=True,\r\n do_train=True,\r\n num_train_epochs=1,\r\n per_device_train_batch_size=16,\r\n per_device_eval_batch_size=16,\r\n save_steps=10000,\r\n logging_dir='./ptlogs'\r\n)\r\n\r\ntrainer = Trainer(\r\n model=model,\r\n args=training_args,\r\n data_collator=data_collator,\r\n train_dataset=dataset,\r\n)\r\ntrainer.train()\r\n```", "`RuntimeError: CUDA out of memory. Tried to allocate 954.00 MiB (GPU 0; 15.90 GiB total capacity; 14.35 GiB already allocated; 753.75 MiB free; 14.39 GiB reserved in total by PyTorch)\r\nException raised from malloc at /pytorch/c10/cuda/CUDACachingAllocator.cpp:272 (most recent call first):`\r\n\r\npart of error output", "from funnel model to bert model : error still happened\r\n\r\nfrom your dataset to LineByLineTextDataset : error disapeared", "notice i just loaded 1000 rows of data", "the error happens when executing loss.backward()", "Since you're using a data collator you don't need to tokenizer the dataset using `map`. Could you try not to use `map` and only the data collator instead ? The data collator is supposed to pad to the longest sequence in each batch afaik, instead of padding to 512.\r\n\r\nAlso cc @sgugger ", "Closing this one.\r\nFeel free to re-open if you have other questions about this issue" ]

[]

[ "Yes good point !", "I think using `sorted(glob.glob())` would actually solve this problem. Can you think of other reasons why dataset loading might not be deterministic? @mariamabarham @yjernite @lhoestq @thomwolf . \r\n\r\nI can do a sweep through the dataset scripts and fix the glob.glob() if you guys are ok with it", "I'm pretty sure it would solve the problem too.\r\n\r\nThe only other dataset that is not deterministic right now is `blog_authorship_corpus` (see #215) but this is a problem related to string encodings.", "I think we should do the same also for `os.list_dir`" ]

[]

[ "@lhoestq @albertvillanova it would be super cool if we could get the Image Classification task to work with this. I'm not sure how to have the dataset find the unique label names _after_ the dataset has been loaded. Is that even possible? \r\n\r\nMy hacky community version [here](https://huggingface.co/datasets/nateraw/image-folder) does this, but it wouldn't pass the test suite here. Any thoughts?", "Hi ! Dataset builders that require some `data_files` like `csv` or `json` are handled differently that actual dataset scripts.\r\n\r\nIn particular:\r\n- they are placed directly in the `src` folder of the lib so that you can use it without internet connection (more exactly in `src/datasets/packaged_modules/<builder_name>.py`). So feel free to move the dataset python file there. You also need to register it in `src/datasets/packaked_modules.__init__.py`\r\n- they are handled a bit differently in our test suite (see the `PackagedDatasetTest` class in `test_dataset_common.py`). To be able to test the builder with your dummy data, you just need to modify `get_packaged_dataset_dummy_data_files` in `test_dataset_common.py` to return the right `data_files` for your builder. The dummy data can stay in `datasets/image_folder/dummy`\r\n\r\nLet me know if you have questions or if I can help !", "Hey @lhoestq , I actually already did both of those things. I'm trying to get the `image-classification` task to work now. \r\n\r\nFor example...When you run `ds = load_dataset('imagefolder', data_files='my_files')`, with a directory called `./my_files` that looks like this:\r\n\r\n```\r\nmy_files\r\n----| Cat\r\n--------| image1.jpg\r\n--------| ...\r\n----| Dog\r\n--------| image1.jpg\r\n--------| ...\r\n```\r\n\r\n...We should set the dataset's `labels` feature to `datasets.features.ClassLabel(names=['cat', 'dog'])` dynamically with class names we find by getting a list of directories in `my_files` (via `data_files`). Otherwise the `datasets.tasks.ImageClassification` task will break, as the `labels` feature is not a `ClassLabel`.\r\n\r\nI couldn't figure out how to access the `data_files` in the builder's `_info` function in a way that would pass in the test suite. ", "Nice ! Then maybe you can use `self.config.data_files` in `_info()` ?\r\nWhat error are you getting in the test suite ?\r\n\r\nAlso note that `data_files` was first developed to accept paths to actual files, not directories. In particular, it fetches the metadata of all the data_files to get a unique hash for the caching mechanism. So we may need to do a few changes first.", "I'm trying to make it work by getting the label names in the _info automatically.\r\nI'll let you know tomorrow how it goes :)\r\n\r\nAlso cc @mariosasko since we're going to use #3163 \r\n\r\nRight now I'm getting the label name per file by taking the first word (from regex `\\w+`) after the common prefix of all the files per split", "Data files resolution takes too much time on my side for a dataset of a few 10,000s of examples. I'll speed it up with some multihreading tomorrow, and maybe by removing the unnecessary checksum verification", "The code is a bit ugly for my taste. I'll try to simplify it tomorrow by avoiding the `os.path.commonprefix` computation and do something similar to @nateraw's ImageFolder instead, where only the second-to-last path component is considered a label (and see if I can update the class labels lazily in `_generate_examples`).\r\n\r\nAlso, as discussed offline with @lhoestq, I reverted the automatic directory globbing change in `data_files.py` and will investigate if we can use `data_dir` for that (e.g. `load_dataset(\"imagefolder\", data_dir=\"path/to/data\")` would be equal to `load_dataset(\"imagefolder\", data_files=[\"path/to/data/**/*\", \"path/to/data/*\"])`. The only problem with `data_dir` that it's equal to `dl_manager.manual_dir`, which would break scripts with `manul_download_instructions`, so maybe we can limit this behavior only to the packaged loaders? WDYT?", "An updated example of usage: [](https://colab.research.google.com/drive/1zJkAQm0Dk372EHcBq6hYgHuunsK4rR0r?usp=sharing)", "> The code is a bit ugly for my taste. I'll try to simplify it tomorrow by avoiding the os.path.commonprefix computation and do something similar to @nateraw's ImageFolder instead, where only the second-to-last path component is considered a label (and see if I can update the class labels lazily in _generate_examples).\r\n\r\nSounds good ! It's fine if we just support the same format as pytorch ImageFolder.\r\n\r\nRegarding the `data_dir` parameter, what do you think is best ?\r\n\r\n1. `dl_manager.data_dir = data_dir`\r\n2. `dl_manager.data_files = resolve(os.path.join(data_dir, \"**\"))`\r\n\r\nor something else ?\r\n\r\n> The only problem with data_dir that it's equal to dl_manager.manual_dir, which would break scripts with manul_download_instructions, so maybe we can limit this behavior only to the packaged loaders? WDYT?\r\n\r\nWe can still have `dl_manager.manual_dir = data_dir` though", "The example colab is amazing !", "@lhoestq \r\n>Regarding the `data_dir` parameter, what do you think is best ?\r\n>\r\n>1. `dl_manager.data_dir = data_dir`\r\n>2. `dl_manager.data_files = resolve(os.path.join(data_dir, \"**\"))`\r\n\r\nThe second option. Basically, I would like `data_files` to be equal to:\r\n```python\r\ndef _split_generators(self, dl_manager):\r\n data_files = self.config.data_files\r\n if data_files is None: \r\n data_files = glob.glob(\"{self.config.data_dir}/**\", recursive=True)\r\n else:\r\n raise ValueError(f\"At least one data file must be specified, but got data_files={data_files}\")\r\n```\r\nin the scripts of packaged modules. It's probably better to do the resolution in `data_files.py` tho (to handle relative file paths on the Hub, for instance)", "> The second option. Basically, I would like data_files to be equal to:\r\n> ```python\r\n> def _split_generators(self, dl_manager):\r\n> data_files = self.config.data_files\r\n> if data_files is None: \r\n> data_files = glob.glob(\"{self.config.data_dir}/**\", recursive=True)\r\n> else:\r\n> raise ValueError(f\"At least one data file must be specified, but got data_files={data_files}\")\r\n> ```\r\n> in the scripts of packaged modules. It's probably better to do the resolution in data_files.py tho (to handle relative file paths on the Hub, for instance)\r\n\r\nsounds good !", "🙌", "Hey @mariosasko are we still actually able to load an image folder?\r\n\r\nFor example...\r\n\r\n```\r\n! wget https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip\r\n! unzip kagglecatsanddogs_3367a.zip\r\n```\r\n\r\nfollowed by\r\n\r\n```python\r\nfrom datasets import load_dataset\r\n\r\n# Does not work\r\nds = load_dataset('imagefolder', data_files='/PetImages')\r\n\r\n# Also doesn't work\r\nds = load_dataset('imagefolder', data_dir='/PetImages')\r\n```\r\n\r\nAre we going forward with the assumption that the user always wants to download from URL and that they won't have a dataset locally already? This at least gets us part of the way, but is technically not an \"imagefolder\" as intended. \r\n\r\nEither way, was delighted to see the colab notebook work smoothly outside of the case I just described above. ❤️ thanks so much for the work here.", "> Hey @mariosasko are we still actually able to load an image folder?\r\n> \r\n> For example...\r\n> \r\n> ```\r\n> ! wget https://download.microsoft.com/download/3/E/1/3E1C3F21-ECDB-4869-8368-6DEBA77B919F/kagglecatsanddogs_3367a.zip\r\n> ! unzip kagglecatsanddogs_3367a.zip\r\n> ```\r\n> \r\n> followed by\r\n> \r\n> ```python\r\n> from datasets import load_dataset\r\n> \r\n> # Does not work\r\n> ds = load_dataset('imagefolder', data_files='/PetImages')\r\n\r\nI ran into this too when I was trying to out. At the moment you can still load from a local on disk directory using a glob pattern i.e. \r\n\r\n```python\r\nfrom datasets import load_dataset\r\nds = load_dataset(\"imagefolder\", data_files=\"PetImages/**/*\")\r\n```\r\n[Colab example](https://colab.research.google.com/drive/1IvAyYSAADHphzbtJMt02OXmnGwGtku3k?usp=sharing). I'm not sure if that is the intended behaviour or not. If it is, I think it would be good to document this because I also assumed the approach @nateraw used would work." ]

[]

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

[ { "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" } ]

[ "Thanks for reporting! You are more than welcome to improve `BuilderConfig`'s docstring.\r\n\r\nThis class serves an identical purpose as `tensorflow_datasets`'s `BuilderConfig`, and its docstring is [here](https://github.com/tensorflow/datasets/blob/a95e38b5bb018312c3d3720619c2a8ef83ebf57f/tensorflow_datasets/core/dataset_builder.py#L81), so feel free to re-use parts of it." ]

[ { "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! We've just started discussing the video support in `datasets` (decoding backends, video feature type, etc.), so I believe we should have something tangible by the end of this year.\r\n\r\nAlso, if you have additional video features in mind that you would like to see, feel free to let us know", "Coool thanks @mariosasko " ]

[]

[]

[]

[ "_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) | 0.011596 / 0.011353 (0.000244) | 0.005845 / 0.011008 (-0.005164) | 0.121302 / 0.038508 (0.082794) | 0.034306 / 0.023109 (0.011196) | 0.355973 / 0.275898 (0.080075) | 0.419903 / 0.323480 (0.096423) | 0.009049 / 0.007986 (0.001064) | 0.004245 / 0.004328 (-0.000084) | 0.092004 / 0.004250 (0.087753) | 0.042782 / 0.037052 (0.005730) | 0.355805 / 0.258489 (0.097316) | 0.407298 / 0.293841 (0.113457) | 0.052481 / 0.128546 (-0.076066) | 0.020880 / 0.075646 (-0.054766) | 0.379948 / 0.419271 (-0.039324) | 0.061337 / 0.043533 (0.017804) | 0.359829 / 0.255139 (0.104690) | 0.379244 / 0.283200 (0.096044) | 0.116692 / 0.141683 (-0.024990) | 1.733717 / 1.452155 (0.281562) | 1.700246 / 1.492716 (0.207530) |\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) | 0.014622 / 0.018006 (-0.003384) | 0.518777 / 0.000490 (0.518288) | 0.004086 / 0.000200 (0.003886) | 0.000136 / 0.000054 (0.000082) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.031208 / 0.037411 (-0.006204) | 0.143003 / 0.014526 (0.128477) | 0.132625 / 0.176557 (-0.043932) | 0.187681 / 0.737135 (-0.549455) | 0.136576 / 0.296338 (-0.159763) |\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) | 0.626516 / 0.215209 (0.411307) | 6.282558 / 2.077655 (4.204904) | 2.702686 / 1.504120 (1.198566) | 2.287445 / 1.541195 (0.746250) | 2.333014 / 1.468490 (0.864524) | 1.227815 / 4.584777 (-3.356962) | 5.545640 / 3.745712 (1.799928) | 4.953226 / 5.269862 (-0.316635) | 2.774549 / 4.565676 (-1.791128) | 0.145257 / 0.424275 (-0.279018) | 0.014887 / 0.007607 (0.007280) | 0.812226 / 0.226044 (0.586182) | 8.002727 / 2.268929 (5.733798) | 3.314852 / 55.444624 (-52.129773) | 2.602348 / 6.876477 (-4.274128) | 2.593511 / 2.142072 (0.451438) | 1.440498 / 4.805227 (-3.364730) | 0.254849 / 6.500664 (-6.245815) | 0.077020 / 0.075469 (0.001551) |\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) | 1.487633 / 1.841788 (-0.354155) | 17.385773 / 8.074308 (9.311465) | 21.775511 / 10.191392 (11.584118) | 0.273514 / 0.680424 (-0.406910) | 0.059644 / 0.534201 (-0.474557) | 0.578710 / 0.579283 (-0.000573) | 0.630221 / 0.434364 (0.195857) | 0.632089 / 0.540337 (0.091752) | 0.762367 / 1.386936 (-0.624569) |\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) | 0.009513 / 0.011353 (-0.001840) | 0.006009 / 0.011008 (-0.004999) | 0.087589 / 0.038508 (0.049081) | 0.037487 / 0.023109 (0.014378) | 0.397660 / 0.275898 (0.121762) | 0.474438 / 0.323480 (0.150958) | 0.007373 / 0.007986 (-0.000613) | 0.005839 / 0.004328 (0.001511) | 0.092759 / 0.004250 (0.088509) | 0.052128 / 0.037052 (0.015075) | 0.382378 / 0.258489 (0.123889) | 0.458244 / 0.293841 (0.164403) | 0.057232 / 0.128546 (-0.071314) | 0.020662 / 0.075646 (-0.054984) | 0.110314 / 0.419271 (-0.308957) | 0.063014 / 0.043533 (0.019481) | 0.386020 / 0.255139 (0.130881) | 0.476169 / 0.283200 (0.192970) | 0.118081 / 0.141683 (-0.023602) | 1.724158 / 1.452155 (0.272003) | 1.862257 / 1.492716 (0.369541) |\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) | 0.224288 / 0.018006 (0.206281) | 0.523631 / 0.000490 (0.523141) | 0.004420 / 0.000200 (0.004220) | 0.000127 / 0.000054 (0.000073) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.032359 / 0.037411 (-0.005052) | 0.140045 / 0.014526 (0.125519) | 0.138164 / 0.176557 (-0.038393) | 0.181068 / 0.737135 (-0.556067) | 0.143965 / 0.296338 (-0.152374) |\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) | 0.573809 / 0.215209 (0.358600) | 6.083247 / 2.077655 (4.005592) | 2.671258 / 1.504120 (1.167138) | 2.277062 / 1.541195 (0.735868) | 2.299544 / 1.468490 (0.831054) | 1.267351 / 4.584777 (-3.317425) | 5.494461 / 3.745712 (1.748749) | 5.083169 / 5.269862 (-0.186692) | 2.531738 / 4.565676 (-2.033938) | 0.151834 / 0.424275 (-0.272441) | 0.014123 / 0.007607 (0.006516) | 0.800222 / 0.226044 (0.574177) | 7.637624 / 2.268929 (5.368695) | 3.325574 / 55.444624 (-52.119050) | 2.563008 / 6.876477 (-4.313468) | 2.596259 / 2.142072 (0.454187) | 1.459206 / 4.805227 (-3.346021) | 0.237771 / 6.500664 (-6.262893) | 0.071854 / 0.075469 (-0.003615) |\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) | 1.605504 / 1.841788 (-0.236284) | 17.593594 / 8.074308 (9.519285) | 20.618005 / 10.191392 (10.426612) | 0.270938 / 0.680424 (-0.409486) | 0.026205 / 0.534201 (-0.507996) | 0.562223 / 0.579283 (-0.017060) | 0.617571 / 0.434364 (0.183207) | 0.616398 / 0.540337 (0.076060) | 0.715293 / 1.386936 (-0.671643) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#673dc0dd7d063b2313f7adcc9e0be53d4718f5cf \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.013213 / 0.011353 (0.001860) | 0.006253 / 0.011008 (-0.004756) | 0.125175 / 0.038508 (0.086667) | 0.037491 / 0.023109 (0.014382) | 0.401379 / 0.275898 (0.125481) | 0.395826 / 0.323480 (0.072346) | 0.009224 / 0.007986 (0.001238) | 0.005163 / 0.004328 (0.000835) | 0.096490 / 0.004250 (0.092239) | 0.042473 / 0.037052 (0.005420) | 0.383713 / 0.258489 (0.125224) | 0.429234 / 0.293841 (0.135393) | 0.063261 / 0.128546 (-0.065285) | 0.020114 / 0.075646 (-0.055532) | 0.401687 / 0.419271 (-0.017585) | 0.062831 / 0.043533 (0.019298) | 0.405211 / 0.255139 (0.150072) | 0.380810 / 0.283200 (0.097610) | 0.109166 / 0.141683 (-0.032517) | 1.869580 / 1.452155 (0.417426) | 1.949947 / 1.492716 (0.457231) |\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) | 0.207481 / 0.018006 (0.189475) | 0.504161 / 0.000490 (0.503671) | 0.008429 / 0.000200 (0.008229) | 0.000101 / 0.000054 (0.000047) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.029182 / 0.037411 (-0.008229) | 0.126284 / 0.014526 (0.111758) | 0.140381 / 0.176557 (-0.036175) | 0.175878 / 0.737135 (-0.561257) | 0.138824 / 0.296338 (-0.157514) |\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) | 0.643658 / 0.215209 (0.428449) | 6.396224 / 2.077655 (4.318569) | 2.600702 / 1.504120 (1.096582) | 2.176721 / 1.541195 (0.635526) | 2.216116 / 1.468490 (0.747626) | 1.235069 / 4.584777 (-3.349708) | 5.457228 / 3.745712 (1.711516) | 3.060455 / 5.269862 (-2.209407) | 2.028123 / 4.565676 (-2.537554) | 0.141617 / 0.424275 (-0.282658) | 0.016596 / 0.007607 (0.008989) | 0.804915 / 0.226044 (0.578870) | 7.968821 / 2.268929 (5.699893) | 3.340650 / 55.444624 (-52.103974) | 2.533620 / 6.876477 (-4.342856) | 2.457388 / 2.142072 (0.315315) | 1.486527 / 4.805227 (-3.318700) | 0.253767 / 6.500664 (-6.246897) | 0.082192 / 0.075469 (0.006723) |\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) | 1.470896 / 1.841788 (-0.370892) | 17.566637 / 8.074308 (9.492329) | 23.144148 / 10.191392 (12.952756) | 0.235510 / 0.680424 (-0.444913) | 0.046051 / 0.534201 (-0.488150) | 0.559954 / 0.579283 (-0.019329) | 0.645390 / 0.434364 (0.211026) | 0.690983 / 0.540337 (0.150646) | 0.776252 / 1.386936 (-0.610684) |\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) | 0.010564 / 0.011353 (-0.000789) | 0.006150 / 0.011008 (-0.004858) | 0.100030 / 0.038508 (0.061522) | 0.036873 / 0.023109 (0.013764) | 0.448508 / 0.275898 (0.172610) | 0.492593 / 0.323480 (0.169113) | 0.007337 / 0.007986 (-0.000648) | 0.004804 / 0.004328 (0.000475) | 0.099218 / 0.004250 (0.094967) | 0.055513 / 0.037052 (0.018461) | 0.462147 / 0.258489 (0.203658) | 0.510229 / 0.293841 (0.216388) | 0.055307 / 0.128546 (-0.073239) | 0.021989 / 0.075646 (-0.053657) | 0.118487 / 0.419271 (-0.300785) | 0.071752 / 0.043533 (0.028219) | 0.456572 / 0.255139 (0.201433) | 0.475160 / 0.283200 (0.191961) | 0.117472 / 0.141683 (-0.024211) | 1.813212 / 1.452155 (0.361058) | 1.908413 / 1.492716 (0.415696) |\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) | 0.352929 / 0.018006 (0.334923) | 0.543874 / 0.000490 (0.543384) | 0.078529 / 0.000200 (0.078329) | 0.000669 / 0.000054 (0.000614) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.033157 / 0.037411 (-0.004254) | 0.162503 / 0.014526 (0.147977) | 0.146424 / 0.176557 (-0.030132) | 0.201781 / 0.737135 (-0.535354) | 0.168110 / 0.296338 (-0.128229) |\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) | 0.644205 / 0.215209 (0.428996) | 6.327519 / 2.077655 (4.249865) | 2.728102 / 1.504120 (1.223982) | 2.306426 / 1.541195 (0.765232) | 2.373125 / 1.468490 (0.904635) | 1.350649 / 4.584777 (-3.234128) | 5.652714 / 3.745712 (1.907002) | 3.175335 / 5.269862 (-2.094526) | 2.222902 / 4.565676 (-2.342775) | 0.160609 / 0.424275 (-0.263666) | 0.015596 / 0.007607 (0.007989) | 0.790357 / 0.226044 (0.564313) | 8.289758 / 2.268929 (6.020830) | 3.479215 / 55.444624 (-51.965410) | 2.860063 / 6.876477 (-4.016413) | 2.806720 / 2.142072 (0.664648) | 1.639046 / 4.805227 (-3.166181) | 0.267017 / 6.500664 (-6.233648) | 0.083990 / 0.075469 (0.008521) |\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) | 1.632262 / 1.841788 (-0.209525) | 17.794357 / 8.074308 (9.720049) | 21.203547 / 10.191392 (11.012155) | 0.250899 / 0.680424 (-0.429525) | 0.024502 / 0.534201 (-0.509699) | 0.519960 / 0.579283 (-0.059323) | 0.615412 / 0.434364 (0.181048) | 0.641914 / 0.540337 (0.101577) | 0.772355 / 1.386936 (-0.614581) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#32cc4d10243b0feb69650f007d010971fd861dc1 \"CML watermark\")\n", "<details>\n<summary>Show benchmarks</summary>\n\nPyArrow==6.0.0\n\n<details>\n<summary>Show updated benchmarks!</summary>\n\n### Benchmark: benchmark_array_xd.json\n\n| metric | read_batch_formatted_as_numpy after write_array2d | read_batch_formatted_as_numpy after write_flattened_sequence | read_batch_formatted_as_numpy after write_nested_sequence | read_batch_unformated after write_array2d | read_batch_unformated after write_flattened_sequence | read_batch_unformated after write_nested_sequence | read_col_formatted_as_numpy after write_array2d | read_col_formatted_as_numpy after write_flattened_sequence | read_col_formatted_as_numpy after write_nested_sequence | read_col_unformated after write_array2d | read_col_unformated after write_flattened_sequence | read_col_unformated after write_nested_sequence | read_formatted_as_numpy after write_array2d | read_formatted_as_numpy after write_flattened_sequence | read_formatted_as_numpy after write_nested_sequence | read_unformated after write_array2d | read_unformated after write_flattened_sequence | read_unformated after write_nested_sequence | write_array2d | write_flattened_sequence | write_nested_sequence |\n|--------|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|\n| new / old (diff) | 0.009501 / 0.011353 (-0.001852) | 0.005262 / 0.011008 (-0.005747) | 0.100809 / 0.038508 (0.062301) | 0.036601 / 0.023109 (0.013492) | 0.299612 / 0.275898 (0.023714) | 0.366970 / 0.323480 (0.043490) | 0.007879 / 0.007986 (-0.000107) | 0.004216 / 0.004328 (-0.000113) | 0.076749 / 0.004250 (0.072498) | 0.042081 / 0.037052 (0.005029) | 0.299572 / 0.258489 (0.041083) | 0.339687 / 0.293841 (0.045846) | 0.038706 / 0.128546 (-0.089840) | 0.012295 / 0.075646 (-0.063352) | 0.336172 / 0.419271 (-0.083100) | 0.047524 / 0.043533 (0.003992) | 0.296800 / 0.255139 (0.041661) | 0.331592 / 0.283200 (0.048393) | 0.101191 / 0.141683 (-0.040491) | 1.486200 / 1.452155 (0.034046) | 1.509955 / 1.492716 (0.017239) |\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) | 0.204735 / 0.018006 (0.186728) | 0.446381 / 0.000490 (0.445891) | 0.005177 / 0.000200 (0.004977) | 0.000099 / 0.000054 (0.000045) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.028655 / 0.037411 (-0.008756) | 0.116559 / 0.014526 (0.102033) | 0.122551 / 0.176557 (-0.054006) | 0.189764 / 0.737135 (-0.547372) | 0.126446 / 0.296338 (-0.169892) |\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) | 0.400104 / 0.215209 (0.184895) | 4.001524 / 2.077655 (1.923869) | 1.779267 / 1.504120 (0.275147) | 1.580168 / 1.541195 (0.038974) | 1.684100 / 1.468490 (0.215610) | 0.703354 / 4.584777 (-3.881423) | 3.828131 / 3.745712 (0.082419) | 2.098500 / 5.269862 (-3.171362) | 1.331161 / 4.565676 (-3.234516) | 0.085417 / 0.424275 (-0.338858) | 0.012380 / 0.007607 (0.004772) | 0.504189 / 0.226044 (0.278144) | 5.094672 / 2.268929 (2.825743) | 2.264352 / 55.444624 (-53.180272) | 1.909573 / 6.876477 (-4.966904) | 2.005425 / 2.142072 (-0.136648) | 0.840893 / 4.805227 (-3.964335) | 0.164689 / 6.500664 (-6.335975) | 0.062754 / 0.075469 (-0.012715) |\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) | 1.250001 / 1.841788 (-0.591786) | 14.993313 / 8.074308 (6.919005) | 14.880601 / 10.191392 (4.689209) | 0.175141 / 0.680424 (-0.505283) | 0.028952 / 0.534201 (-0.505249) | 0.447073 / 0.579283 (-0.132210) | 0.445993 / 0.434364 (0.011629) | 0.525527 / 0.540337 (-0.014811) | 0.613156 / 1.386936 (-0.773780) |\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) | 0.007796 / 0.011353 (-0.003557) | 0.005399 / 0.011008 (-0.005609) | 0.078240 / 0.038508 (0.039732) | 0.035303 / 0.023109 (0.012193) | 0.364603 / 0.275898 (0.088705) | 0.400794 / 0.323480 (0.077314) | 0.006152 / 0.007986 (-0.001834) | 0.004324 / 0.004328 (-0.000004) | 0.074949 / 0.004250 (0.070698) | 0.051939 / 0.037052 (0.014887) | 0.377079 / 0.258489 (0.118590) | 0.413630 / 0.293841 (0.119789) | 0.037567 / 0.128546 (-0.090979) | 0.012793 / 0.075646 (-0.062854) | 0.089013 / 0.419271 (-0.330258) | 0.050748 / 0.043533 (0.007215) | 0.370100 / 0.255139 (0.114961) | 0.384838 / 0.283200 (0.101638) | 0.105840 / 0.141683 (-0.035843) | 1.476490 / 1.452155 (0.024335) | 1.544688 / 1.492716 (0.051972) |\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) | 0.220987 / 0.018006 (0.202981) | 0.443801 / 0.000490 (0.443311) | 0.005747 / 0.000200 (0.005547) | 0.000106 / 0.000054 (0.000051) |\n\n### Benchmark: benchmark_indices_mapping.json\n\n| metric | select | shard | shuffle | sort | train_test_split |\n|--------|---|---|---|---|---|\n| new / old (diff) | 0.030187 / 0.037411 (-0.007225) | 0.118230 / 0.014526 (0.103704) | 0.126810 / 0.176557 (-0.049746) | 0.200482 / 0.737135 (-0.536654) | 0.130831 / 0.296338 (-0.165507) |\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) | 0.423231 / 0.215209 (0.208022) | 4.196576 / 2.077655 (2.118921) | 1.992919 / 1.504120 (0.488799) | 1.809172 / 1.541195 (0.267977) | 1.932706 / 1.468490 (0.464216) | 0.727319 / 4.584777 (-3.857458) | 3.833295 / 3.745712 (0.087583) | 3.527005 / 5.269862 (-1.742857) | 1.937348 / 4.565676 (-2.628329) | 0.088713 / 0.424275 (-0.335562) | 0.012711 / 0.007607 (0.005104) | 0.531385 / 0.226044 (0.305341) | 5.308051 / 2.268929 (3.039123) | 2.493494 / 55.444624 (-52.951131) | 2.168359 / 6.876477 (-4.708118) | 2.258160 / 2.142072 (0.116088) | 0.865629 / 4.805227 (-3.939598) | 0.171281 / 6.500664 (-6.329383) | 0.065746 / 0.075469 (-0.009723) |\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) | 1.290378 / 1.841788 (-0.551409) | 15.900804 / 8.074308 (7.826496) | 14.809614 / 10.191392 (4.618222) | 0.177287 / 0.680424 (-0.503137) | 0.017875 / 0.534201 (-0.516326) | 0.429646 / 0.579283 (-0.149637) | 0.451646 / 0.434364 (0.017282) | 0.545669 / 0.540337 (0.005332) | 0.633215 / 1.386936 (-0.753721) |\n\n</details>\n</details>\n\n![](https://cml.dev/watermark.png#2c67b5f4bc9cea088e977a135644d38da8c144ff \"CML watermark\")\n" ]

[]

[ "@lhoestq, should I raise a PR for this? Just a minor change while reading labels text file", "Sure feel free to open a PR thanks !" ]

[]

[ "I am wondering why only one test of \"keep_in_memory=True\" fails, when there are many other tests that test the same and it happens only in pyarrow_1..." ]

[]

[]

[]

[ "nvm, I'd made an assumption that the library gets installed with transformers. " ]

[]

[]

[ { "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" }, { "color": "d93f0b", "default": false, "description": "", "id": 2725241052, "name": "speech", "node_id": "MDU6TGFiZWwyNzI1MjQxMDUy", "url": "https://api.github.com/repos/huggingface/datasets/labels/speech" } ]

[]

[ { "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" } ]

[ "As discussed, the metadata JSON files must be regenerated because the keys were nor properly generated and they will not be read by the builder:\r\n> Indeed there is some problem/bug while reading the datasets_info.json file: there is a mismatch with the config.name keys in the file...\r\nIn the meanwhile, in order to be able to use the datasets_info.json file content, you can create the builder without passing the name :\r\n```\r\nIn [25]: builder = datasets.load_dataset_builder(\"journalists_questions\")\r\nIn [26]: builder.info.splits\r\nOut[26]: {'train': SplitInfo(name='train', num_bytes=342296, num_examples=10077, dataset_name='journalists_questions')}\r\n```\r\n\r\nAfter regenerating the metadata JSON file for this dataset, I get the right key:\r\n```\r\n{\"plain_text\": {\"description\": \"The journalists_questions corpus (\r\n```", "Thanks!" ]

[ { "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": "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" } ]

[ "Hello @mrm8488, are these datasets official datasets published in an NLP/CL/ML venue?", "What about this one: https://github.com/ccasimiro88/TranslateAlignRetrieve?" ]

[]

[ "Looks good to me. Do we also support TSV in this helper (explain if it should be text or CSV) and in the dummy-data creator?", "snli is basically based on tsv files (but named as .txt) and it is in the list of datasets of reference.\r\nThe dummy data creator supports tsv", "merging this one.\r\nIf you think of other datasets of reference to add we can still add them later" ]

[]

[ "The docs for this PR live [here](https://moon-ci-docs.huggingface.co/docs/datasets/pr_5579). All of your documentation changes will be reflected on that endpoint.", "I'm not sure we need this part as we provide a link to the notebook that shows how to train an object detection model, and this notebook instantiates a `DataLoader` before training the model. I'd like to hear what @stevhliu thinks.\r\n\r\nPS: Your `collate_fn` calls `torch.stack` on the `bbox` tensors, which don't have the same shape, so this will fail.", "I agree with @mariosasko; we also have a [Use with PyTorch](https://huggingface.co/docs/datasets/use_with_pytorch) guide that shows how you can create a `DataLoader`. " ]

[]

[ "`download_mode=datasets.GenerateMode.FORCE_REDOWNLOAD` should work.\r\nThis makes me think we we should rename this to DownloadMode.FORCE_REDOWNLOAD. Currently that's confusing", "Can we just use `features=...` in `load_dataset` for this @lhoestq?", "Indeed you should use `features` in this case. \r\n```python\r\nfeatures = Features({'text': Value('string'), 'label': Value('float32')})\r\ndataset = load_dataset('csv', data_files=['sst_test.csv'], features=features)\r\n```\r\nNote that because of an issue with the caching when you change the features (see #750 ) you still need to specify the `FORCE_REDOWNLOAD ` flag. I'm working on a fix for this one", "https://github.com/huggingface/datasets/issues/769#issuecomment-717837832\r\n> This makes me think we we should rename this to DownloadMode.FORCE_REDOWNLOAD. Currently that's confusing\r\n\r\n@lhoestq do you still think we should rename it?\r\n", "It's no big deal, but since it can be confusing to users I think it's worth renaming it, and deprecate `GenerateMode` until `datasets` 2.0 at least. IMO it's confusing to have `download_mode=GenerateMode.something`" ]

[ { "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" } ]

[ "To have some context, please see:\r\n- #4145\r\n\r\nPlease, also note that we have recently made some fixes to the script, which are in our GitHub master branch but not yet released:\r\n- #4422\r\n- #4425 \r\n- #4436", "Thanks Albert! I'll try pulling `datasets` from the git repo instead of PyPI, and/or just wait for the next release.\r\n", "I'm closing this issue then. Please, feel free to reopen it again if the problem persists." ]

[ { "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" } ]

[ "Thanks for the report !\r\n\r\nCurrently we don't have a way to let the user easily disable this behavior.\r\nHowever I agree that we should support stateful processing functions, ideally by removing `does_function_return_dict`.\r\n\r\nWe needed this function in order to know whether the `map` functions needs to write data or not. if `does_function_return_dict` returns False then we don't write anything.\r\n\r\nInstead of checking the output of the processing function outside of the for loop that iterates through the dataset to process it, we can check the output of the first processed example and at that point decide if we need to write data or not.\r\n\r\nTherefore it's definitely possible to fix this unwanted behavior, any contribution going into this direction is welcome :)", "Thanks @mariosasko for the PR!" ]

[]