id
stringlengths
2
115
lastModified
stringlengths
24
24
tags
list
author
stringlengths
2
42
description
stringlengths
0
68.7k
citation
stringlengths
0
10.7k
cardData
null
likes
int64
0
3.55k
downloads
int64
0
10.1M
card
stringlengths
0
1.01M
nRuaif/Pure-dove-sharegpt
2023-10-01T14:01:10.000Z
[ "region:us" ]
nRuaif
null
null
null
0
4
Entry not found
nikchar/retrieval_verification_bm25_bert
2023-10-01T10:50:28.000Z
[ "region:us" ]
nikchar
null
null
null
0
4
--- dataset_info: features: - name: claim dtype: string - name: evidence_wiki_url dtype: string - name: text dtype: string - name: retrieved_evidence_title sequence: string - name: retrieved_evidence_text sequence: string - name: labels dtype: int64 - name: Retrieval_Success dtype: bool - name: Predicted_Labels dtype: int64 - name: Predicted_Labels_Each_doc sequence: int64 splits: - name: train num_bytes: 66031496 num_examples: 11073 download_size: 30811942 dataset_size: 66031496 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "retrieval_verification_bm25_bert" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
prattay/abinbev
2023-10-01T09:46:36.000Z
[ "license:apache-2.0", "region:us" ]
prattay
null
null
null
0
4
--- license: apache-2.0 ---
learn3r/SDG_cs
2023-10-01T11:45:46.000Z
[ "region:us" ]
learn3r
null
null
null
0
4
--- dataset_info: features: - name: jargon dtype: string - name: definition dtype: string splits: - name: train num_bytes: 44588 num_examples: 200 download_size: 29080 dataset_size: 44588 --- # Dataset Card for "SDG_cs" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
wwwlir/langcain_docs_l1
2023-10-01T19:43:24.000Z
[ "region:us" ]
wwwlir
null
null
null
0
4
Entry not found
rmanluo/RoG-webqsp
2023-10-01T23:40:22.000Z
[ "region:us" ]
rmanluo
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: validation path: data/validation-* - split: test path: data/test-* dataset_info: features: - name: id dtype: string - name: question dtype: string - name: answer sequence: string - name: q_entity sequence: string - name: a_entity sequence: string - name: graph sequence: sequence: string - name: choices sequence: 'null' splits: - name: train num_bytes: 993540472 num_examples: 2826 - name: validation num_bytes: 84009553 num_examples: 246 - name: test num_bytes: 580788090 num_examples: 1628 download_size: 0 dataset_size: 1658338115 --- # Dataset Card for "RoG-webqsp" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
rmanluo/RoG-cwq
2023-10-01T23:47:36.000Z
[ "region:us" ]
rmanluo
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: validation path: data/validation-* - split: test path: data/test-* dataset_info: features: - name: id dtype: string - name: question dtype: string - name: answer sequence: string - name: q_entity sequence: string - name: a_entity sequence: string - name: graph sequence: sequence: string - name: choices sequence: 'null' splits: - name: train num_bytes: 8890766478 num_examples: 27639 - name: validation num_bytes: 1170336525 num_examples: 3519 - name: test num_bytes: 1208452620 num_examples: 3531 download_size: 1993772283 dataset_size: 11269555623 --- # Dataset Card for "RoG-cwq" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Sathvik-24/HinGlishLaama2
2023-10-02T07:00:41.000Z
[ "region:us" ]
Sathvik-24
null
null
null
0
4
Entry not found
JuanKO/T5_summarization_RLAIF
2023-10-02T14:57:00.000Z
[ "license:apache-2.0", "region:us" ]
JuanKO
null
null
null
0
4
--- license: apache-2.0 dataset_info: features: - name: prompt dtype: string - name: summary_1 dtype: string - name: summary_2 dtype: string splits: - name: train num_bytes: 1697095 num_examples: 1000 download_size: 906302 dataset_size: 1697095 ---
pphuc25/uit_data_sample
2023-10-02T16:54:52.000Z
[ "region:us" ]
pphuc25
null
null
null
0
4
--- dataset_info: features: - name: id dtype: string - name: context dtype: string - name: claim dtype: string - name: verdict dtype: string - name: evidence dtype: string - name: domain dtype: string splits: - name: train num_bytes: 4167523 num_examples: 1000 download_size: 1991987 dataset_size: 4167523 --- # Dataset Card for "uit_data_sample" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
adamo1139/PS_AD_Office365_02
2023-10-03T00:13:00.000Z
[ "license:apache-2.0", "region:us" ]
adamo1139
null
null
null
0
4
--- license: apache-2.0 --- Second version of the synthetic dataset created by putting a part of a textbook in the context of 7B model and then asking the model to create a few questions and answers related to the dataset. It contains information about PowerShell basics, Office 365 basics and Active Directory/GPO basics.
ZhongshengWang/PARARULE-Plus-Alpaca
2023-10-03T06:24:35.000Z
[ "license:mit", "region:us" ]
ZhongshengWang
null
null
null
0
4
--- license: mit ---
hanifabdlh/quac-lamini-instruction-indo-0k-10k
2023-10-03T05:46:21.000Z
[ "region:us" ]
hanifabdlh
null
null
null
0
4
--- dataset_info: features: - name: context dtype: string - name: instruction dtype: string - name: response dtype: string - name: instruction_source dtype: string splits: - name: train num_bytes: 4177364 num_examples: 10000 download_size: 2408739 dataset_size: 4177364 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "quac-lamini-instruction-indo-0k-10k" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
ouvic215/Soldering-Data-pix2pix-1001
2023-10-03T08:01:47.000Z
[ "region:us" ]
ouvic215
null
null
null
0
4
--- dataset_info: features: - name: mask_image dtype: image - name: text dtype: string - name: image dtype: image splits: - name: train num_bytes: 961523307.5 num_examples: 12054 download_size: 960371764 dataset_size: 961523307.5 --- # Dataset Card for "Soldering-Data-pix2pix-1001" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
nguyenthanhdo/viettel_v3.2
2023-10-03T08:52:34.000Z
[ "region:us" ]
nguyenthanhdo
null
null
null
0
4
--- dataset_info: features: - name: instruction dtype: string - name: output dtype: string - name: translated dtype: bool - name: output_len dtype: int64 - name: source dtype: string - name: input dtype: string splits: - name: train num_bytes: 327564182.0 num_examples: 100000 download_size: 157982995 dataset_size: 327564182.0 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "viettel_v3.2" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
ismailiismail/paraphrasing_french_5000
2023-10-03T19:47:58.000Z
[ "region:us" ]
ismailiismail
null
null
null
0
4
--- dataset_info: features: - name: phrase dtype: string - name: paraphrase dtype: string splits: - name: train num_bytes: 1240685 num_examples: 4972 download_size: 499325 dataset_size: 1240685 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "paraphrasing_french_5000" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Minglii/v_4096
2023-10-04T01:49:42.000Z
[ "region:us" ]
Minglii
null
null
null
0
4
--- dataset_info: features: - name: data struct: - name: conversations list: - name: from dtype: string - name: markdown struct: - name: answer dtype: string - name: index dtype: int64 - name: type dtype: string - name: text dtype: string - name: value dtype: string - name: id dtype: string splits: - name: train num_bytes: 685122486 num_examples: 80129 download_size: 278043744 dataset_size: 685122486 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "v_4096" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Minglii/W_QthenA_4096
2023-10-04T01:51:58.000Z
[ "region:us" ]
Minglii
null
null
null
0
4
--- dataset_info: features: - name: data struct: - name: conversations list: - name: from dtype: string - name: value dtype: string - name: id dtype: string splits: - name: train num_bytes: 1576315785 num_examples: 143000 download_size: 537850801 dataset_size: 1576315785 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "W_QthenA_4096" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
DamarJati/Face-Mask-Detection
2023-10-04T06:34:17.000Z
[ "task_categories:image-classification", "language:en", "art", "face mask", "mask", "region:us" ]
DamarJati
null
null
null
0
4
--- language: - en pipeline_tag: image-classification tags: - art - face mask - mask task_categories: - image-classification --- Original datasets https://www.kaggle.com/datasets/ashishjangra27/face-mask-12k-images-dataset
lakelz/mydataset-bpg
2023-10-07T06:54:47.000Z
[ "region:us" ]
lakelz
null
null
null
0
4
This dataset is a subset of the Open Assistant dataset, which you can find here: https://huggingface.co/datasets/OpenAssistant/oasst1/tree/main This subset of the data only contains the highest-rated paths in the conversation tree, with a total of 9,846 samples. This dataset was used to train Guanaco with QLoRA. For further information, please see the original dataset. License: Apache 2.0
Alexandre-Numind/TrainIE_grouped
2023-10-04T08:53:58.000Z
[ "region:us" ]
Alexandre-Numind
null
null
null
0
4
Entry not found
Alexandre-Numind/ValIE_grouped
2023-10-04T08:54:38.000Z
[ "region:us" ]
Alexandre-Numind
null
null
null
0
4
Entry not found
TheAIchemist13/hindi_asr_dataset
2023-10-10T15:11:52.000Z
[ "region:us" ]
TheAIchemist13
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* dataset_info: features: - name: audio dtype: audio: sampling_rate: 16000 - name: transcriptions dtype: string splits: - name: train num_bytes: 12211841.0 num_examples: 80 - name: test num_bytes: 12211841.0 num_examples: 80 download_size: 24346804 dataset_size: 24423682.0 --- # Dataset Card for "hindi_asr_dataset" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
AayushShah/SQL_PlainText_Combined
2023-10-04T10:49:38.000Z
[ "region:us" ]
AayushShah
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* dataset_info: features: - name: text dtype: string - name: target dtype: string splits: - name: train num_bytes: 349116676.7610253 num_examples: 306706 - name: test num_bytes: 38791374.23897472 num_examples: 34079 download_size: 98654951 dataset_size: 387908051.0 --- # Dataset Card for "SQL_PlainText_Combined" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
BlazeLlama/euclid_elements_eng
2023-10-04T18:52:14.000Z
[ "license:apache-2.0", "region:us" ]
BlazeLlama
null
null
null
0
4
--- license: apache-2.0 ---
finiteautomata/yahoo_dataset
2023-10-04T18:34:26.000Z
[ "region:us" ]
finiteautomata
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* dataset_info: features: - name: id dtype: int32 - name: topic dtype: class_label: names: '0': Society & Culture '1': Science & Mathematics '2': Health '3': Education & Reference '4': Computers & Internet '5': Sports '6': Business & Finance '7': Entertainment & Music '8': Family & Relationships '9': Politics & Government - name: question_title dtype: string - name: question_content dtype: string - name: best_answer dtype: string - name: question_title_embeddings sequence: float32 - name: question_content_embeddings sequence: float32 - name: best_answer_embeddings sequence: float32 splits: - name: train num_bytes: 1032387680 num_examples: 200000 - name: test num_bytes: 309853862 num_examples: 60000 download_size: 500190426 dataset_size: 1342241542 --- # Dataset Card for "yahoo_dataset" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
gorkaartola/ZS-train_S1-SDGdescriptions-AURORA1_S2-SDGdescriptions-SDGtitle_Negative_Sample_Filter-AURORA1
2023-10-04T20:07:43.000Z
[ "region:us" ]
gorkaartola
null
null
null
0
4
Entry not found
angellist/cupcakeLPAParsingTest
2023-10-05T06:03:51.000Z
[ "region:us" ]
angellist
null
null
null
0
4
--- dataset_info: features: - name: input dtype: string - name: output dtype: string - name: text dtype: string splits: - name: train num_bytes: 1483910 num_examples: 1138 download_size: 448368 dataset_size: 1483910 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "cupcakeLPAParsingTest" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
sugeun/231005
2023-10-05T04:20:31.000Z
[ "region:us" ]
sugeun
null
null
null
0
4
Entry not found
ManeAI31416/NASA_fine-tuning
2023-10-05T05:36:01.000Z
[ "license:llama2", "region:us" ]
ManeAI31416
null
null
null
0
4
--- license: llama2 ---
hanifabdlh/quac-lamini-instruction-indo-30k-40k
2023-10-05T06:19:20.000Z
[ "region:us" ]
hanifabdlh
null
null
null
0
4
--- dataset_info: features: - name: context dtype: string - name: instruction dtype: string - name: response dtype: string - name: instruction_source dtype: string splits: - name: train num_bytes: 4126187 num_examples: 10000 download_size: 2378575 dataset_size: 4126187 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "quac-lamini-instruction-indo-30k-40k" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
jfrei/GPTNERMED
2023-10-08T22:05:18.000Z
[ "task_categories:token-classification", "task_ids:named-entity-recognition", "annotations_creators:machine-generated", "language_creators:machine-generated", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "language:de", "bio", "biomedical", "medical", "c...
jfrei
GPTNERMED is a novel open synthesized dataset and neural named-entity-recognition (NER) model for German texts in medical natural language processing (NLP).
@article{FREI2023104478, title = {Annotated dataset creation through large language models for non-english medical NLP}, journal = {Journal of Biomedical Informatics}, volume = {145}, pages = {104478}, year = {2023}, issn = {1532-0464}, doi = {https://doi.org/10.1016/j.jbi.2023.104478}, url = {https://www.sciencedirect.com/science/article/pii/S1532046423001995}, author = {Johann Frei and Frank Kramer}, keywords = {Natural language processing, Information extraction, Named entity recognition, Data augmentation, Knowledge distillation, Medication detection}, abstract = {Obtaining text datasets with semantic annotations is an effortful process, yet crucial for supervised training in natural language processing (NLP). In general, developing and applying new NLP pipelines in domain-specific contexts for tasks often requires custom-designed datasets to address NLP tasks in a supervised machine learning fashion. When operating in non-English languages for medical data processing, this exposes several minor and major, interconnected problems such as the lack of task-matching datasets as well as task-specific pre-trained models. In our work, we suggest to leverage pre-trained large language models for training data acquisition in order to retrieve sufficiently large datasets for training smaller and more efficient models for use-case-specific tasks. To demonstrate the effectiveness of your approach, we create a custom dataset that we use to train a medical NER model for German texts, GPTNERMED, yet our method remains language-independent in principle. Our obtained dataset as well as our pre-trained models are publicly available at https://github.com/frankkramer-lab/GPTNERMED.} }
null
0
4
--- annotations_creators: - machine-generated language: - de language_creators: - machine-generated license: [] multilinguality: - monolingual pretty_name: GPTNERMED size_categories: - 1K<n<10K source_datasets: - original tags: - bio - biomedical - medical - clinical task_categories: - token-classification task_ids: - named-entity-recognition --- # GPTNERMED Dataset for German medical NER entities ## Dataset Description - **Repository:** https://github.com/frankkramer-lab/GPTNERMED - **Paper:** https://doi.org/10.1016/j.jbi.2023.104478 - **ArXiv-Preprint:** https://arxiv.org/abs/2208.14493 ## Dataset Summary This dataset contains the synthetic German sentences with annotated entities (`Medikation`, `Dosis`, `Diagnose`) from the GPTNERMED project. The sentences as well as the annotations are **not** manually validated by medical professionals and therefore this dataset is **no** gold standard dataset. The dataset consists of 9,845 sentences (121,027 tokens by SpaCy Tokenizer, 245,107 tokens by the GPT tokenizer) with the following labels: | Label | Count | #Tokens (SpaCy) | | --- | --- | -- | | Medikation | 9868 | 10138 | | Dosis | 7547 | 15845 | | Diagnose | 5996 | 7656 | ## Dataset Structure The train/test/dev-split (80%, 10%, 10%) of the data loader is as follows:\ `<-- train: 0.8 --><-- test: 0.1 --><-- dev: 0.1 -->`\ The splits are selected arbitrarily as the dataloader requires a split configuration. All sample sentences are however homogeneous in origin and splits could also be performed otherwise. Every sample is a sentence with its text (property `sentence`) and its corresponding NER labels (property `ner_labels` / List of labels).\ Every NER label entry has a char-wise start and stop index (property `start`, `stop`) and a label class (property `ner_class`). ### Citation Information If you like our work, cite our paper and give us a star on GitHub.\ (See the links above)
Sathvik-24/chachadata
2023-10-05T13:03:04.000Z
[ "region:us" ]
Sathvik-24
null
null
null
0
4
Entry not found
magnus42/test_train_hey
2023-10-05T17:03:53.000Z
[ "region:us" ]
magnus42
null
null
null
0
4
Entry not found
ninja/arabic-english-translation
2023-10-05T17:07:41.000Z
[ "region:us" ]
ninja
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* dataset_info: features: - name: arabic dtype: string - name: english dtype: string splits: - name: train num_bytes: 228876.54205607477 num_examples: 674 - name: test num_bytes: 25468.457943925234 num_examples: 75 download_size: 159571 dataset_size: 254345.0 --- # Dataset Card for "arabic-english-translation" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
MasterBruce1/test1
2023-10-05T18:56:59.000Z
[ "region:us" ]
MasterBruce1
null
null
null
0
4
Entry not found
joey234/sst2_affix_pos
2023-10-05T23:45:33.000Z
[ "region:us" ]
joey234
null
null
null
0
4
--- dataset_info: features: - name: idx dtype: int32 - name: sentence dtype: string - name: label dtype: class_label: names: '0': negative '1': positive - name: words_with_affixes sequence: string splits: - name: validation num_bytes: 9357 num_examples: 58 download_size: 9664 dataset_size: 9357 configs: - config_name: default data_files: - split: validation path: data/validation-* --- # Dataset Card for "sst2_affix_pos" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
ayan1988/diffusion.2.textual_inversion
2023-10-06T06:51:48.000Z
[ "region:us" ]
ayan1988
null
null
null
0
4
--- dataset_info: features: - name: image dtype: image splits: - name: train num_bytes: 1740639.0 num_examples: 6 download_size: 0 dataset_size: 1740639.0 --- # Dataset Card for "diffusion.2.textual_inversion" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Falah/ali_prompts
2023-10-06T07:36:22.000Z
[ "region:us" ]
Falah
null
null
null
0
4
--- dataset_info: features: - name: prompts dtype: string splits: - name: train num_bytes: 367063 num_examples: 1000 download_size: 19378 dataset_size: 367063 --- # Dataset Card for "ali_prompts" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
ikiransuryavanshi/llama_training3
2023-10-06T07:39:00.000Z
[ "region:us" ]
ikiransuryavanshi
null
null
null
0
4
Entry not found
lafnac/sl-dataset
2023-10-06T10:09:07.000Z
[ "task_categories:text-classification", "size_categories:1M<n<10M", "language:ar", "license:afl-3.0", "region:us" ]
lafnac
null
null
null
0
4
--- license: afl-3.0 task_categories: - text-classification language: - ar size_categories: - 1M<n<10M --- # Dataset Card for Dataset Name ## Dataset Description - **Homepage:** - **Repository:** - **Paper:** - **Leaderboard:** - **Point of Contact:** ### Dataset Summary This dataset card aims to be a base template for new datasets. It has been generated using [this raw template](https://github.com/huggingface/huggingface_hub/blob/main/src/huggingface_hub/templates/datasetcard_template.md?plain=1). ### Supported Tasks and Leaderboards [More Information Needed] ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information [More Information Needed] ### Contributions [More Information Needed]
dieineb/sartaj
2023-10-06T11:58:25.000Z
[ "region:us" ]
dieineb
null
null
null
0
4
Entry not found
HamdanXI/difference_analysis_data_structure
2023-10-06T12:21:19.000Z
[ "region:us" ]
HamdanXI
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* dataset_info: features: - name: en_toxic_comment dtype: string - name: en_neutral_comment dtype: string - name: edit_ops sequence: sequence: string splits: - name: train num_bytes: 4067285 num_examples: 19744 download_size: 1996316 dataset_size: 4067285 --- # Dataset Card for "difference_analysis_data_structure" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
muhammadravi251001/indonesian-nli-and-qa
2023-10-06T14:09:01.000Z
[ "license:mit", "region:us" ]
muhammadravi251001
null
null
null
0
4
--- license: mit ---
ContextualAI/nq_open
2023-10-07T00:34:08.000Z
[ "region:us" ]
ContextualAI
null
null
null
0
4
--- dataset_info: features: - name: query dtype: string - name: gold_generation sequence: string splits: - name: train num_bytes: 5990520 num_examples: 79168 - name: dev num_bytes: 660716 num_examples: 8757 - name: test num_bytes: 313829 num_examples: 3610 download_size: 4681299 dataset_size: 6965065 --- # Dataset Card for "nq_open" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
vsarathy/nl-robotics-semantic-parsing-info_structure-2k-no-context-TEST
2023-10-07T12:31:44.000Z
[ "region:us" ]
vsarathy
null
null
null
0
4
Entry not found
syaoran312/VHAC_QA_full
2023-10-07T19:51:18.000Z
[ "region:us" ]
syaoran312
null
null
null
0
4
Entry not found
rongrong77/ADL_HW1
2023-10-08T06:52:58.000Z
[ "region:us" ]
rongrong77
null
null
null
0
4
Entry not found
tyzhu/synpre_union_1M
2023-10-08T09:18:54.000Z
[ "region:us" ]
tyzhu
null
null
null
0
4
--- dataset_info: features: - name: inputs dtype: string - name: targets dtype: string splits: - name: train num_bytes: 1167868421 num_examples: 1000000 - name: validation num_bytes: 11660114 num_examples: 10000 download_size: 788391948 dataset_size: 1179528535 --- # Dataset Card for "synpre_union_1M" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
MikuHH/stagop
2023-10-08T13:14:25.000Z
[ "region:us" ]
MikuHH
null
null
null
0
4
Entry not found
Dmkond/tune-forms
2023-10-08T15:44:24.000Z
[ "region:us" ]
Dmkond
null
null
null
0
4
--- dataset_info: features: - name: text dtype: string splits: - name: train num_bytes: 842248 num_examples: 200 download_size: 221015 dataset_size: 842248 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "tune-forms" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Ayansk11/llama2_merged_file
2023-10-08T17:15:08.000Z
[ "region:us" ]
Ayansk11
null
null
null
0
4
Entry not found
gayanin/legal-es-masked
2023-10-08T21:56:11.000Z
[ "region:us" ]
gayanin
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* - split: validation path: data/validation-* dataset_info: features: - name: original_sent dtype: string - name: masked_sent dtype: string splits: - name: train num_bytes: 14319276284 num_examples: 48833571 - name: test num_bytes: 2144523252 num_examples: 6104196 - name: validation num_bytes: 2169841655 num_examples: 6104197 download_size: 8287754892 dataset_size: 18633641191 --- # Dataset Card for "legal-es-masked" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
minh21/COVID-QA-Chunk-64-question-answering-biencoder-data-90_10
2023-10-09T04:29:31.000Z
[ "region:us" ]
minh21
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: validation path: data/validation-* dataset_info: features: - name: question dtype: string - name: answer dtype: string - name: context_chunks sequence: string - name: document_id dtype: int64 - name: id dtype: int64 splits: - name: train num_bytes: 78943266 num_examples: 1631 - name: validation num_bytes: 8529659 num_examples: 185 download_size: 14143196 dataset_size: 87472925 --- # Dataset Card for "COVID-QA-Chunk-64-question-answering-biencoder-data-90_10" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
kelzla/ds_test1
2023-10-09T05:35:13.000Z
[ "region:us" ]
kelzla
null
null
null
0
4
Entry not found
krthk/kapardhi_dataset
2023-10-09T06:45:45.000Z
[ "region:us" ]
krthk
null
null
null
0
4
Entry not found
Back-up/validation_data_T5
2023-10-09T06:58:57.000Z
[ "region:us" ]
Back-up
null
null
null
0
4
--- dataset_info: features: - name: input_ids sequence: int32 - name: attention_mask sequence: int8 - name: labels sequence: int64 splits: - name: train num_bytes: 338661368 num_examples: 31984 download_size: 43689455 dataset_size: 338661368 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "validation_data_T5" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
giovanni92/MailFuncData
2023-10-09T08:00:32.000Z
[ "license:mit", "region:us" ]
giovanni92
null
null
null
0
4
--- license: mit ---
Colin23189/kaggle-exam-llm
2023-10-09T08:34:21.000Z
[ "region:us" ]
Colin23189
null
null
null
0
4
Entry not found
kolkata97/pellm0-zancanaro-split
2023-10-09T12:43:38.000Z
[ "region:us" ]
kolkata97
null
null
null
0
4
Entry not found
tonywu71/PokemonCards_fixed
2023-10-09T13:11:31.000Z
[ "license:mit", "region:us" ]
tonywu71
null
null
null
0
4
--- license: mit dataset_info: features: - name: id dtype: string - name: image_url dtype: string - name: caption dtype: string - name: name dtype: string - name: hp dtype: int64 - name: set_name dtype: string splits: - name: train num_bytes: 9474973.87624629 num_examples: 13088 download_size: 3028812 dataset_size: 9474973.87624629 configs: - config_name: default data_files: - split: train path: data/train-* ---
dmrau/cqadupstack-android-qrels
2023-10-09T12:39:31.000Z
[ "region:us" ]
dmrau
null
null
null
0
4
--- configs: - config_name: default data_files: - split: test path: data/test-* dataset_info: features: - name: query-id dtype: string - name: corpus-id dtype: string - name: score dtype: int64 splits: - name: test num_bytes: 43411 num_examples: 1696 download_size: 0 dataset_size: 43411 --- # Dataset Card for "cqadupstack-android-qrels" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
dmrau/cqadupstack-gaming
2023-10-09T12:39:43.000Z
[ "region:us" ]
dmrau
null
null
null
0
4
--- configs: - config_name: default data_files: - split: queries path: data/queries-* - split: corpus path: data/corpus-* dataset_info: features: - name: _id dtype: string - name: text dtype: string - name: title dtype: string splits: - name: queries num_bytes: 105494 num_examples: 1595 - name: corpus num_bytes: 20666596 num_examples: 45301 download_size: 12946080 dataset_size: 20772090 --- # Dataset Card for "cqadupstack-gaming" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
dmrau/cqadupstack-webmasters
2023-10-09T12:41:03.000Z
[ "region:us" ]
dmrau
null
null
null
0
4
--- configs: - config_name: default data_files: - split: queries path: data/queries-* - split: corpus path: data/corpus-* dataset_info: features: - name: _id dtype: string - name: text dtype: string - name: title dtype: string splits: - name: queries num_bytes: 34792 num_examples: 506 - name: corpus num_bytes: 11659413 num_examples: 17405 download_size: 6885106 dataset_size: 11694205 --- # Dataset Card for "cqadupstack-webmasters" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
dmrau/cqadupstack-english
2023-10-09T12:41:18.000Z
[ "region:us" ]
dmrau
null
null
null
0
4
--- configs: - config_name: default data_files: - split: queries path: data/queries-* - split: corpus path: data/corpus-* dataset_info: features: - name: _id dtype: string - name: text dtype: string - name: title dtype: string splits: - name: queries num_bytes: 103588 num_examples: 1570 - name: corpus num_bytes: 18199570 num_examples: 40221 download_size: 11382247 dataset_size: 18303158 --- # Dataset Card for "cqadupstack-english" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
dmrau/cqadupstack-unix
2023-10-09T12:42:00.000Z
[ "region:us" ]
dmrau
null
null
null
0
4
--- configs: - config_name: default data_files: - split: queries path: data/queries-* - split: corpus path: data/corpus-* dataset_info: features: - name: _id dtype: string - name: text dtype: string - name: title dtype: string splits: - name: queries num_bytes: 72357 num_examples: 1072 - name: corpus num_bytes: 46102756 num_examples: 47382 download_size: 24571026 dataset_size: 46175113 --- # Dataset Card for "cqadupstack-unix" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
dmrau/cqadupstack-wordpress
2023-10-09T12:42:09.000Z
[ "region:us" ]
dmrau
null
null
null
0
4
--- configs: - config_name: default data_files: - split: queries path: data/queries-* - split: corpus path: data/corpus-* dataset_info: features: - name: _id dtype: string - name: text dtype: string - name: title dtype: string splits: - name: queries num_bytes: 35736 num_examples: 541 - name: corpus num_bytes: 53026140 num_examples: 48605 download_size: 26551471 dataset_size: 53061876 --- # Dataset Card for "cqadupstack-wordpress" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
berardi6/LBcmopcenscaspnewwsx4
2023-10-09T12:44:50.000Z
[ "region:us" ]
berardi6
null
null
null
0
4
--- dataset_info: features: - name: text dtype: string splits: - name: train num_bytes: 565921 num_examples: 1788 download_size: 180294 dataset_size: 565921 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "LBcmopcenscaspnewwsx4" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Amarjitkr/medical
2023-10-09T17:46:33.000Z
[ "license:apache-2.0", "region:us" ]
Amarjitkr
null
null
null
0
4
--- license: apache-2.0 ---
amphora/lmsys-filtered
2023-10-09T17:57:19.000Z
[ "region:us" ]
amphora
null
null
null
0
4
--- dataset_info: features: - name: conversation_id dtype: string - name: model dtype: string - name: conversation dtype: string - name: turn dtype: int64 - name: language dtype: string - name: openai_moderation dtype: string - name: redacted dtype: bool - name: __index_level_0__ dtype: int64 splits: - name: train num_bytes: 317822351 num_examples: 62968 download_size: 122101594 dataset_size: 317822351 --- # Dataset Card for "lmsys-filtered" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
rcherukuri14/science-qa-instructions
2023-10-09T21:46:14.000Z
[ "region:us" ]
rcherukuri14
null
null
null
0
4
Entry not found
promptora11/train
2023-10-10T04:28:14.000Z
[ "region:us" ]
promptora11
null
null
null
0
4
--- configs: - config_name: default data_files: - split: train path: data/train-* - split: test path: data/test-* dataset_info: features: - name: instruction dtype: string - name: input dtype: string - name: output dtype: string - name: text dtype: string splits: - name: train num_bytes: 90417.6 num_examples: 108 - name: test num_bytes: 10046.4 num_examples: 12 download_size: 16905 dataset_size: 100464.0 --- # Dataset Card for "train" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Falah/invention_prompts
2023-10-10T05:47:15.000Z
[ "region:us" ]
Falah
null
null
null
0
4
--- dataset_info: features: - name: prompts dtype: string splits: - name: train num_bytes: 96461 num_examples: 1000 download_size: 2138 dataset_size: 96461 --- # Dataset Card for "invention_prompts" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Luciya/llama-2-nuv-intent-noE-pp
2023-10-10T05:58:08.000Z
[ "region:us" ]
Luciya
null
null
null
0
4
--- dataset_info: features: - name: text dtype: string splits: - name: train num_bytes: 791845 num_examples: 1585 download_size: 111893 dataset_size: 791845 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "llama-2-nuv-intent-noE-pp" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Jagadeesh-ti/sql-v4
2023-10-10T06:00:29.000Z
[ "region:us" ]
Jagadeesh-ti
null
null
null
0
4
Entry not found
Luciya/llama-2-nuv-intent-noE-pp-oos
2023-10-10T06:50:06.000Z
[ "region:us" ]
Luciya
null
null
null
0
4
--- dataset_info: features: - name: text dtype: string splits: - name: train num_bytes: 921669 num_examples: 1834 download_size: 134964 dataset_size: 921669 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "llama-2-nuv-intent-noE-pp-oos" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
Luciya/llama-2-nuv-intent-noE-oos
2023-10-10T06:50:18.000Z
[ "region:us" ]
Luciya
null
null
null
0
4
--- dataset_info: features: - name: text dtype: string splits: - name: train num_bytes: 828135 num_examples: 1834 download_size: 127293 dataset_size: 828135 configs: - config_name: default data_files: - split: train path: data/train-* --- # Dataset Card for "llama-2-nuv-intent-noE-oos" [More Information needed](https://github.com/huggingface/datasets/blob/main/CONTRIBUTING.md#how-to-contribute-to-the-dataset-cards)
ilyas3141/ilias_test20
2023-10-10T08:41:24.000Z
[ "region:us" ]
ilyas3141
null
null
null
0
4
Entry not found
Oscaraandersson/testrag
2023-10-10T09:13:42.000Z
[ "region:us" ]
Oscaraandersson
null
null
null
0
4
Entry not found
autshumato
2023-06-01T14:59:51.000Z
[ "task_categories:translation", "annotations_creators:expert-generated", "language_creators:expert-generated", "multilinguality:multilingual", "size_categories:100K<n<1M", "size_categories:10K<n<100K", "source_datasets:original", "language:en", "language:tn", "language:ts", "language:zu", "lice...
null
Multilingual information access is stipulated in the South African constitution. In practise, this is hampered by a lack of resources and capacity to perform the large volumes of translation work required to realise multilingual information access. One of the aims of the Autshumato project is to develop machine translation systems for three South African languages pairs.
@article{groenewald2010processing, title={Processing parallel text corpora for three South African language pairs in the Autshumato project}, author={Groenewald, Hendrik J and du Plooy, Liza}, journal={AfLaT 2010}, pages={27}, year={2010} }
null
2
3
--- annotations_creators: - expert-generated language_creators: - expert-generated language: - en - tn - ts - zu license: - cc-by-2.5 multilinguality: - multilingual size_categories: - 100K<n<1M - 10K<n<100K source_datasets: - original task_categories: - translation task_ids: [] paperswithcode_id: null pretty_name: autshumato dataset_info: - config_name: autshumato-en-tn features: - name: translation dtype: translation: languages: - en - tn splits: - name: train num_bytes: 28826392 num_examples: 159000 download_size: 9458762 dataset_size: 28826392 - config_name: autshumato-en-zu features: - name: translation dtype: translation: languages: - en - zu splits: - name: train num_bytes: 7188970 num_examples: 35489 download_size: 2068891 dataset_size: 7188970 - config_name: autshumato-en-ts features: - name: translation dtype: translation: languages: - en - ts splits: - name: train num_bytes: 50803849 num_examples: 450000 download_size: 15145915 dataset_size: 50803849 - config_name: autshumato-en-ts-manual features: - name: translation dtype: translation: languages: - en - ts splits: - name: train num_bytes: 10408757 num_examples: 92396 download_size: 2876924 dataset_size: 10408757 - config_name: autshumato-tn features: - name: text dtype: string splits: - name: train num_bytes: 5132267 num_examples: 38206 download_size: 1599029 dataset_size: 5132267 - config_name: autshumato-ts features: - name: text dtype: string splits: - name: train num_bytes: 3399674 num_examples: 58398 download_size: 974488 dataset_size: 3399674 config_names: - autshumato-en-tn - autshumato-en-ts - autshumato-en-ts-manual - autshumato-en-zu - autshumato-tn - autshumato-ts --- # Dataset Card for autshumato ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [https://repo.sadilar.org/handle/20.500.12185/7/discover]() - **Repository:** []() - **Paper:** []() - **Leaderboard:** []() - **Point of Contact:** []() ### Dataset Summary Multilingual information access is stipulated in the South African constitution. In practise, this is hampered by a lack of resources and capacity to perform the large volumes of translation work required to realise multilingual information access. One of the aims of the Autshumato project is to develop machine translation systems for three South African languages pairs. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data [More Information Needed] #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations [More Information Needed] #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information [More Information Needed] ### Dataset Curators [More Information Needed] ### Licensing Information ### Citation Information ``` @article{groenewald2010processing, title={Processing parallel text corpora for three South African language pairs in the Autshumato project}, author={Groenewald, Hendrik J and du Plooy, Liza}, journal={AfLaT 2010}, pages={27}, year={2010} } ``` ### Contributions Thanks to [@Narsil](https://github.com/Narsil) for adding this dataset.
bswac
2022-11-03T16:15:55.000Z
[ "task_categories:text-generation", "task_categories:fill-mask", "task_ids:language-modeling", "task_ids:masked-language-modeling", "annotations_creators:no-annotation", "language_creators:found", "multilinguality:monolingual", "size_categories:100M<n<1B", "source_datasets:original", "language:bs",...
null
The Bosnian web corpus bsWaC was built by crawling the .ba top-level domain in 2014. The corpus was near-deduplicated on paragraph level, normalised via diacritic restoration, morphosyntactically annotated and lemmatised. The corpus is shuffled by paragraphs. Each paragraph contains metadata on the URL, domain and language identification (Bosnian vs. Croatian vs. Serbian). Version 1.0 of this corpus is described in http://www.aclweb.org/anthology/W14-0405. Version 1.1 contains newer and better linguistic annotations.
@misc{11356/1062, title = {Bosnian web corpus {bsWaC} 1.1}, author = {Ljube{\v s}i{\'c}, Nikola and Klubi{\v c}ka, Filip}, url = {http://hdl.handle.net/11356/1062}, note = {Slovenian language resource repository {CLARIN}.{SI}}, copyright = {Creative Commons - Attribution-{ShareAlike} 4.0 International ({CC} {BY}-{SA} 4.0)}, year = {2016} }
null
0
3
--- annotations_creators: - no-annotation language_creators: - found language: - bs license: - cc-by-sa-3.0 multilinguality: - monolingual size_categories: - 100M<n<1B source_datasets: - original task_categories: - text-generation - fill-mask task_ids: - language-modeling - masked-language-modeling paperswithcode_id: null pretty_name: BsWac dataset_info: features: - name: sentence dtype: string config_name: bswac splits: - name: train num_bytes: 9156258478 num_examples: 354581267 download_size: 1988514951 dataset_size: 9156258478 --- # Dataset Card for BsWac ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** http://nlp.ffzg.hr/resources/corpora/bswac/ - **Repository:** https://www.clarin.si/repository/xmlui/handle/11356/1062 - **Paper:** http://nlp.ffzg.hr/data/publications/nljubesi/ljubesic14-bs.pdf - **Leaderboard:** - **Point of Contact:** [Nikola Ljubešič](mailto:nikola.ljubesic@ffzg.hr) ### Dataset Summary The Bosnian web corpus bsWaC was built by crawling the .ba top-level domain in 2014. The corpus was near-deduplicated on paragraph level, normalised via diacritic restoration, morphosyntactically annotated and lemmatised. The corpus is shuffled by paragraphs. Each paragraph contains metadata on the URL, domain and language identification (Bosnian vs. Croatian vs. Serbian). ### Supported Tasks and Leaderboards [More Information Needed] ### Languages Dataset is monolingual in Bosnian language. ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information Dataset is under the [CC-BY-SA 3.0](http://creativecommons.org/licenses/by-sa/3.0/) license. ### Citation Information ``` @misc{11356/1062, title = {Bosnian web corpus {bsWaC} 1.1}, author = {Ljube{\v s}i{\'c}, Nikola and Klubi{\v c}ka, Filip}, url = {http://hdl.handle.net/11356/1062}, note = {Slovenian language resource repository {CLARIN}.{SI}}, copyright = {Creative Commons - Attribution-{ShareAlike} 4.0 International ({CC} {BY}-{SA} 4.0)}, year = {2016} } ``` ### Contributions Thanks to [@IvanZidov](https://github.com/IvanZidov) for adding this dataset.
capes
2022-11-03T16:15:53.000Z
[ "task_categories:translation", "annotations_creators:found", "language_creators:found", "multilinguality:multilingual", "size_categories:1M<n<10M", "source_datasets:original", "language:en", "language:pt", "license:unknown", "dissertation-abstracts-translation", "theses-translation", "region:u...
null
A parallel corpus of theses and dissertations abstracts in English and Portuguese were collected from the CAPES website (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) - Brazil. The corpus is sentence aligned for all language pairs. Approximately 240,000 documents were collected and aligned using the Hunalign algorithm.
@inproceedings{soares2018parallel, title={A Parallel Corpus of Theses and Dissertations Abstracts}, author={Soares, Felipe and Yamashita, Gabrielli Harumi and Anzanello, Michel Jose}, booktitle={International Conference on Computational Processing of the Portuguese Language}, pages={345--352}, year={2018}, organization={Springer} }
null
2
3
--- annotations_creators: - found language_creators: - found language: - en - pt license: - unknown multilinguality: - multilingual size_categories: - 1M<n<10M source_datasets: - original task_categories: - translation task_ids: [] paperswithcode_id: capes pretty_name: CAPES tags: - dissertation-abstracts-translation - theses-translation dataset_info: features: - name: translation dtype: translation: languages: - en - pt config_name: en-pt splits: - name: train num_bytes: 472484364 num_examples: 1157610 download_size: 162229298 dataset_size: 472484364 --- # Dataset Card for CAPES ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:**[Parallel corpus of theses and dissertation abstracts in Portuguese and English from CAPES](https://sites.google.com/view/felipe-soares/datasets#h.p_kxOR6EhHm2a6) - **Repository:** - **Paper:** - **Leaderboard:** - **Point of Contact:** ### Dataset Summary A parallel corpus of theses and dissertations abstracts in English and Portuguese were collected from the CAPES website (Coordenação de Aperfeiçoamento de Pessoal de Nível Superior) - Brazil. The corpus is sentence aligned for all language pairs. Approximately 240,000 documents were collected and aligned using the Hunalign algorithm. ### Supported Tasks and Leaderboards The underlying task is machine translation. ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information ``` @inproceedings{soares2018parallel, title={A Parallel Corpus of Theses and Dissertations Abstracts}, author={Soares, Felipe and Yamashita, Gabrielli Harumi and Anzanello, Michel Jose}, booktitle={International Conference on Computational Processing of the Portuguese Language}, pages={345--352}, year={2018}, organization={Springer} } ``` ### Contributions Thanks to [@patil-suraj](https://github.com/patil-suraj) for adding this dataset.
chr_en
2023-06-01T14:59:50.000Z
[ "task_categories:fill-mask", "task_categories:text-generation", "task_categories:translation", "task_ids:language-modeling", "task_ids:masked-language-modeling", "annotations_creators:expert-generated", "annotations_creators:found", "annotations_creators:no-annotation", "language_creators:found", ...
null
ChrEn is a Cherokee-English parallel dataset to facilitate machine translation research between Cherokee and English. ChrEn is extremely low-resource contains 14k sentence pairs in total, split in ways that facilitate both in-domain and out-of-domain evaluation. ChrEn also contains 5k Cherokee monolingual data to enable semi-supervised learning.
@inproceedings{zhang2020chren, title={ChrEn: Cherokee-English Machine Translation for Endangered Language Revitalization}, author={Zhang, Shiyue and Frey, Benjamin and Bansal, Mohit}, booktitle={EMNLP2020}, year={2020} }
null
3
3
--- annotations_creators: - expert-generated - found - no-annotation language_creators: - found language: - chr - en license: - other multilinguality: - monolingual - multilingual - translation size_categories: - 100K<n<1M - 10K<n<100K - 1K<n<10K source_datasets: - original task_categories: - fill-mask - text-generation - translation task_ids: - language-modeling - masked-language-modeling paperswithcode_id: chren dataset_info: - config_name: monolingual_raw features: - name: text_sentence dtype: string - name: text_title dtype: string - name: speaker dtype: string - name: date dtype: int32 - name: type dtype: string - name: dialect dtype: string splits: - name: full num_bytes: 1210828 num_examples: 5210 download_size: 28899321 dataset_size: 1210828 - config_name: parallel_raw features: - name: line_number dtype: string - name: sentence_pair dtype: translation: languages: - en - chr - name: text_title dtype: string - name: speaker dtype: string - name: date dtype: int32 - name: type dtype: string - name: dialect dtype: string splits: - name: full num_bytes: 5012923 num_examples: 14151 download_size: 28899321 dataset_size: 5012923 - config_name: monolingual features: - name: sentence dtype: string splits: - name: chr num_bytes: 882848 num_examples: 5210 - name: en5000 num_bytes: 615295 num_examples: 5000 - name: en10000 num_bytes: 1211645 num_examples: 10000 - name: en20000 num_bytes: 2432378 num_examples: 20000 - name: en50000 num_bytes: 6065780 num_examples: 49999 - name: en100000 num_bytes: 12130564 num_examples: 100000 download_size: 28899321 dataset_size: 23338510 - config_name: parallel features: - name: sentence_pair dtype: translation: languages: - en - chr splits: - name: train num_bytes: 3089658 num_examples: 11639 - name: dev num_bytes: 260409 num_examples: 1000 - name: out_dev num_bytes: 78134 num_examples: 256 - name: test num_bytes: 264603 num_examples: 1000 - name: out_test num_bytes: 80967 num_examples: 256 download_size: 28899321 dataset_size: 3773771 config_names: - monolingual - monolingual_raw - parallel - parallel_raw --- # Dataset Card for ChrEn ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Repository:** [Github repository for ChrEn](https://github.com/ZhangShiyue/ChrEn) - **Paper:** [ChrEn: Cherokee-English Machine Translation for Endangered Language Revitalization](https://arxiv.org/abs/2010.04791) - **Point of Contact:** [benfrey@email.unc.edu](benfrey@email.unc.edu) ### Dataset Summary ChrEn is a Cherokee-English parallel dataset to facilitate machine translation research between Cherokee and English. ChrEn is extremely low-resource contains 14k sentence pairs in total, split in ways that facilitate both in-domain and out-of-domain evaluation. ChrEn also contains 5k Cherokee monolingual data to enable semi-supervised learning. ### Supported Tasks and Leaderboards The dataset is intended to use for `machine-translation` between Enlish (`en`) and Cherokee (`chr`). ### Languages The dataset contains Enlish (`en`) and Cherokee (`chr`) text. The data encompasses both existing dialects of Cherokee: the Overhill dialect, mostly spoken in Oklahoma (OK), and the Middle dialect, mostly used in North Carolina (NC). ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization Many of the source texts were translations of English materials, which means that the Cherokee structures may not be 100% natural in terms of what a speaker might spontaneously produce. Each text was translated by people who speak Cherokee as the first language, which means there is a high probability of grammaticality. These data were originally available in PDF version. We apply the Optical Character Recognition (OCR) via Tesseract OCR engine to extract the Cherokee and English text. #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? The sentences were manually aligned by Dr. Benjamin Frey a proficient second-language speaker of Cherokee, who also fixed the errors introduced by OCR. This process is time-consuming and took several months. ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators The dataset was gathered and annotated by Shiyue Zhang, Benjamin Frey, and Mohit Bansal at UNC Chapel Hill. ### Licensing Information The copyright of the data belongs to original book/article authors or translators (hence, used for research purpose; and please contact Dr. Benjamin Frey for other copyright questions). ### Citation Information ``` @inproceedings{zhang2020chren, title={ChrEn: Cherokee-English Machine Translation for Endangered Language Revitalization}, author={Zhang, Shiyue and Frey, Benjamin and Bansal, Mohit}, booktitle={EMNLP2020}, year={2020} } ``` ### Contributions Thanks to [@yjernite](https://github.com/yjernite), [@lhoestq](https://github.com/lhoestq) for adding this dataset.
coached_conv_pref
2023-01-25T14:28:17.000Z
[ "task_categories:other", "task_categories:text-generation", "task_categories:fill-mask", "task_categories:token-classification", "task_ids:dialogue-modeling", "task_ids:parsing", "annotations_creators:expert-generated", "language_creators:found", "multilinguality:monolingual", "size_categories:n<1...
null
A dataset consisting of 502 English dialogs with 12,000 annotated utterances between a user and an assistant discussing movie preferences in natural language. It was collected using a Wizard-of-Oz methodology between two paid crowd-workers, where one worker plays the role of an 'assistant', while the other plays the role of a 'user'. The 'assistant' elicits the 'user’s' preferences about movies following a Coached Conversational Preference Elicitation (CCPE) method. The assistant asks questions designed to minimize the bias in the terminology the 'user' employs to convey his or her preferences as much as possible, and to obtain these preferences in natural language. Each dialog is annotated with entity mentions, preferences expressed about entities, descriptions of entities provided, and other statements of entities.
@inproceedings{48414, title = {Coached Conversational Preference Elicitation: A Case Study in Understanding Movie Preferences}, author = {Filip Radlinski and Krisztian Balog and Bill Byrne and Karthik Krishnamoorthi}, year = {2019}, booktitle = {Proceedings of the Annual SIGdial Meeting on Discourse and Dialogue} }
null
2
3
--- annotations_creators: - expert-generated language_creators: - found language: - en license: - cc-by-sa-4.0 multilinguality: - monolingual size_categories: - n<1K source_datasets: - original task_categories: - other - text-generation - fill-mask - token-classification task_ids: - dialogue-modeling - parsing paperswithcode_id: coached-conversational-preference-elicitation pretty_name: Coached Conversational Preference Elicitation tags: - Conversational Recommendation dataset_info: features: - name: conversationId dtype: string - name: utterances sequence: - name: index dtype: int32 - name: speaker dtype: class_label: names: '0': USER '1': ASSISTANT - name: text dtype: string - name: segments sequence: - name: startIndex dtype: int32 - name: endIndex dtype: int32 - name: text dtype: string - name: annotations sequence: - name: annotationType dtype: class_label: names: '0': ENTITY_NAME '1': ENTITY_PREFERENCE '2': ENTITY_DESCRIPTION '3': ENTITY_OTHER - name: entityType dtype: class_label: names: '0': MOVIE_GENRE_OR_CATEGORY '1': MOVIE_OR_SERIES '2': PERSON '3': SOMETHING_ELSE config_name: coached_conv_pref splits: - name: train num_bytes: 2295579 num_examples: 502 download_size: 5191959 dataset_size: 2295579 --- # Dataset Card for Coached Conversational Preference Elicitation ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [Coached Conversational Preference Elicitation Homepage](https://research.google/tools/datasets/coached-conversational-preference-elicitation/) - **Repository:** [Coached Conversational Preference Elicitation Repository](https://github.com/google-research-datasets/ccpe) - **Paper:** [Aclweb](https://www.aclweb.org/anthology/W19-5941/) ### Dataset Summary A dataset consisting of 502 English dialogs with 12,000 annotated utterances between a user and an assistant discussing movie preferences in natural language. It was collected using a Wizard-of-Oz methodology between two paid crowd-workers, where one worker plays the role of an 'assistant', while the other plays the role of a 'user'. The 'assistant' elicits the 'user’s' preferences about movies following a Coached Conversational Preference Elicitation (CCPE) method. The assistant asks questions designed to minimize the bias in the terminology the 'user' employs to convey his or her preferences as much as possible, and to obtain these preferences in natural language. Each dialog is annotated with entity mentions, preferences expressed about entities, descriptions of entities provided, and other statements of entities. ### Supported Tasks and Leaderboards * `other-other-Conversational Recommendation`: The dataset can be used to train a model for Conversational recommendation, which consists in Coached Conversation Preference Elicitation. ### Languages The text in the dataset is in English. The associated BCP-47 code is `en`. ## Dataset Structure ### Data Instances A typical data point comprises of a series of utterances between the 'assistant' and the 'user'. Each such utterance is annotated into categories mentioned in data fields. An example from the Coached Conversational Preference Elicitation dataset looks as follows: ``` {'conversationId': 'CCPE-6faee', 'utterances': {'index': [0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15], 'segments': [{'annotations': [{'annotationType': [], 'entityType': []}], 'endIndex': [0], 'startIndex': [0], 'text': ['']}, {'annotations': [{'annotationType': [0], 'entityType': [0]}, {'annotationType': [1], 'entityType': [0]}], 'endIndex': [20, 27], 'startIndex': [14, 0], 'text': ['comedy', 'I really like comedy movies']}, {'annotations': [{'annotationType': [0], 'entityType': [0]}], 'endIndex': [24], 'startIndex': [16], 'text': ['comedies']}, {'annotations': [{'annotationType': [1], 'entityType': [0]}], 'endIndex': [15], 'startIndex': [0], 'text': ['I love to laugh']}, {'annotations': [{'annotationType': [], 'entityType': []}], 'endIndex': [0], 'startIndex': [0], 'text': ['']}, {'annotations': [{'annotationType': [0], 'entityType': [1]}, {'annotationType': [1], 'entityType': [1]}], 'endIndex': [21, 21], 'startIndex': [8, 0], 'text': ['Step Brothers', 'I liked Step Brothers']}, {'annotations': [{'annotationType': [], 'entityType': []}], 'endIndex': [0], 'startIndex': [0], 'text': ['']}, {'annotations': [{'annotationType': [1], 'entityType': [1]}], 'endIndex': [32], 'startIndex': [0], 'text': ['Had some amazing one-liners that']}, {'annotations': [{'annotationType': [], 'entityType': []}], 'endIndex': [0], 'startIndex': [0], 'text': ['']}, {'annotations': [{'annotationType': [0], 'entityType': [1]}, {'annotationType': [1], 'entityType': [1]}], 'endIndex': [15, 15], 'startIndex': [13, 0], 'text': ['RV', "I don't like RV"]}, {'annotations': [{'annotationType': [], 'entityType': []}], 'endIndex': [0], 'startIndex': [0], 'text': ['']}, {'annotations': [{'annotationType': [1], 'entityType': [1]}, {'annotationType': [1], 'entityType': [1]}], 'endIndex': [48, 66], 'startIndex': [18, 50], 'text': ['It was just so slow and boring', "I didn't like it"]}, {'annotations': [{'annotationType': [0], 'entityType': [1]}], 'endIndex': [63], 'startIndex': [33], 'text': ['Jurassic World: Fallen Kingdom']}, {'annotations': [{'annotationType': [0], 'entityType': [1]}, {'annotationType': [3], 'entityType': [1]}], 'endIndex': [52, 52], 'startIndex': [22, 0], 'text': ['Jurassic World: Fallen Kingdom', 'I have seen the movie Jurassic World: Fallen Kingdom']}, {'annotations': [{'annotationType': [], 'entityType': []}], 'endIndex': [0], 'startIndex': [0], 'text': ['']}, {'annotations': [{'annotationType': [1], 'entityType': [1]}, {'annotationType': [1], 'entityType': [1]}, {'annotationType': [1], 'entityType': [1]}], 'endIndex': [24, 125, 161], 'startIndex': [0, 95, 135], 'text': ['I really like the actors', 'I just really like the scenery', 'the dinosaurs were awesome']}], 'speaker': [1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0, 1, 0], 'text': ['What kinds of movies do you like?', 'I really like comedy movies.', 'Why do you like comedies?', "I love to laugh and comedy movies, that's their whole purpose. Make you laugh.", 'Alright, how about a movie you liked?', 'I liked Step Brothers.', 'Why did you like that movie?', 'Had some amazing one-liners that still get used today even though the movie was made awhile ago.', 'Well, is there a movie you did not like?', "I don't like RV.", 'Why not?', "And I just didn't It was just so slow and boring. I didn't like it.", 'Ok, then have you seen the movie Jurassic World: Fallen Kingdom', 'I have seen the movie Jurassic World: Fallen Kingdom.', 'What is it about these kinds of movies that you like or dislike?', 'I really like the actors. I feel like they were doing their best to make the movie better. And I just really like the scenery, and the the dinosaurs were awesome.']}} ``` ### Data Fields Each conversation has the following fields: * `conversationId`: A unique random ID for the conversation. The ID has no meaning. * `utterances`: An array of utterances by the workers. Each utterance has the following fields: * `index`: A 0-based index indicating the order of the utterances in the conversation. * `speaker`: Either USER or ASSISTANT, indicating which role generated this utterance. * `text`: The raw text as written by the ASSISTANT, or transcribed from the spoken recording of USER. * `segments`: An array of semantic annotations of spans in the text. Each semantic annotation segment has the following fields: * `startIndex`: The position of the start of the annotation in the utterance text. * `endIndex`: The position of the end of the annotation in the utterance text. * `text`: The raw text that has been annotated. * `annotations`: An array of annotation details for this segment. Each annotation has two fields: * `annotationType`: The class of annotation (see ontology below). * `entityType`: The class of the entity to which the text refers (see ontology below). **EXPLANATION OF ONTOLOGY** In the corpus, preferences and the entities that these preferences refer to are annotated with an annotation type as well as an entity type. Annotation types fall into four categories: * `ENTITY_NAME` (0): These mark the names of relevant entities mentioned. * `ENTITY_PREFERENCE` (1): These are defined as statements indicating that the dialog participant does or does not like the relevant entity in general, or that they do or do not like some aspect of the entity. This may also be thought of the participant having some sentiment about what is being discussed. * `ENTITY_DESCRIPTION` (2): Neutral descriptions that describe an entity but do not convey an explicit liking or disliking. * `ENTITY_OTHER` (3): Other relevant statements about an entity that convey relevant information of how the participant relates to the entity but do not provide a sentiment. Most often, these relate to whether a participant has seen a particular movie, or knows a lot about a given entity. Entity types are marked as belonging to one of four categories: * `MOVIE_GENRE_OR_CATEGORY` (0): For genres or general descriptions that capture a particular type or style of movie. * `MOVIE_OR_SERIES` (1): For the full or partial name of a movie or series of movies. * `PERSON` (2): For the full or partial name of an actual person. * `SOMETHING_ELSE ` (3): For other important proper nouns, such as the names of characters or locations. ### Data Splits There is a single split of the dataset named 'train' which contains the whole datset. | | Train | | ------------------- | ----- | | Input Conversations | 502 | ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [Creative Commons Attribution 4.0 License](https://creativecommons.org/licenses/by/4.0/) ### Citation Information ``` @inproceedings{radlinski-etal-2019-ccpe, title = {Coached Conversational Preference Elicitation: A Case Study in Understanding Movie Preferences}, author = {Filip Radlinski and Krisztian Balog and Bill Byrne and Karthik Krishnamoorthi}, booktitle = {Proceedings of the Annual Meeting of the Special Interest Group on Discourse and Dialogue ({SIGDIAL})}, year = 2019 } ``` ### Contributions Thanks to [@vineeths96](https://github.com/vineeths96) for adding this dataset.
dyk
2023-01-25T14:29:39.000Z
[ "task_categories:question-answering", "task_ids:open-domain-qa", "annotations_creators:expert-generated", "language_creators:other", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "language:pl", "license:bsd-3-clause", "region:us" ]
null
The Did You Know (pol. Czy wiesz?) dataset consists of human-annotated question-answer pairs. The task is to predict if the answer is correct. We chose the negatives which have the largest token overlap with a question.
@inproceedings{marcinczuk2013open, title={Open dataset for development of Polish Question Answering systems}, author={Marcinczuk, Michal and Ptak, Marcin and Radziszewski, Adam and Piasecki, Maciej}, booktitle={Proceedings of the 6th Language & Technology Conference: Human Language Technologies as a Challenge for Computer Science and Linguistics, Wydawnictwo Poznanskie, Fundacja Uniwersytetu im. Adama Mickiewicza}, year={2013} }
null
0
3
--- annotations_creators: - expert-generated language_creators: - other language: - pl license: - bsd-3-clause multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - question-answering task_ids: - open-domain-qa pretty_name: dyk dataset_info: features: - name: q_id dtype: string - name: question dtype: string - name: answer dtype: string - name: target dtype: class_label: names: '0': '0' '1': '1' splits: - name: train num_bytes: 1388690 num_examples: 4154 - name: test num_bytes: 353643 num_examples: 1029 download_size: 685462 dataset_size: 1742333 --- # Dataset Card for [Dataset Name] ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** http://nlp.pwr.wroc.pl/en/tools-and-resources/resources/czy-wiesz-question-answering-dataset - **Repository:** - **Paper:** - **Leaderboard:** - **Point of Contact:** ### Dataset Summary The Did You Know (pol. Czy wiesz?) dataset consists of human-annotated question-answer pairs. The task is to predict if the answer is correct. We chose the negatives which have the largest token overlap with a question. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages Polish ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields - q_id: question id - question: question sentence - answer: answer sentence - target: 1 if the answer is correct, 0 otherwise. Note that the test split doesn't have target values so -1 is used instead ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information CC BY-SA 3.0 ### Citation Information [More Information Needed] ### Contributions Thanks to [@abecadel](https://github.com/abecadel) for adding this dataset.
eduge
2023-01-25T14:29:42.000Z
[ "task_categories:text-classification", "task_ids:multi-class-classification", "annotations_creators:expert-generated", "language_creators:expert-generated", "multilinguality:monolingual", "size_categories:10K<n<100K", "source_datasets:original", "language:mn", "license:unknown", "region:us" ]
null
Eduge news classification dataset is provided by Bolorsoft LLC. It is used for training the Eduge.mn production news classifier 75K news articles in 9 categories: урлаг соёл, эдийн засаг, эрүүл мэнд, хууль, улс төр, спорт, технологи, боловсрол and байгал орчин
null
null
3
3
--- annotations_creators: - expert-generated language_creators: - expert-generated language: - mn license: - unknown multilinguality: - monolingual size_categories: - 10K<n<100K source_datasets: - original task_categories: - text-classification task_ids: - multi-class-classification pretty_name: Eduge dataset_info: features: - name: news dtype: string - name: label dtype: class_label: names: '0': урлаг соёл '1': эдийн засаг '2': эрүүл мэнд '3': хууль '4': улс төр '5': спорт '6': технологи '7': боловсрол '8': байгал орчин splits: - name: train num_bytes: 255275842 num_examples: 60528 - name: test num_bytes: 64451731 num_examples: 15133 download_size: 320395067 dataset_size: 319727573 --- # Dataset Card for Eduge ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** http://eduge.mn/ - **Repository:** https://github.com/tugstugi/mongolian-nlp - **Paper:** [Needs More Information] - **Leaderboard:** [Needs More Information] - **Point of Contact:** [Needs More Information] ### Dataset Summary Eduge news classification dataset provided by Bolorsoft LLC. Used to train the Eduge.mn production news classifier 75K news articles in 9 categories: урлаг соёл, эдийн засаг, эрүүл мэнд, хууль, улс төр, спорт, технологи, боловсрол and байгал орчин ### Supported Tasks and Leaderboards - `text-classification`: We can transform the above into a 9-class classification task. ### Languages The text in the dataset is in Mongolian ## Dataset Structure ### Data Instances For the `default` configuration: ``` { 'label': 0, # 'урлаг соёл' 'news': 'Шударга өрсөлдөөн, хэрэглэгчийн төлөө газар 2013 оны дөрөвдүгээр сараас эхлэн Монгол киноны ашиг орлогын мэдээллийг олон нийтэд хүргэж байгаа. Ингэснээр Монголын кино үйлдвэрлэгчид улсад ашиг орлогоо шударгаар төлөх, мөн  чанартай уран бүтээлийн тоо өсөх боломж бүрдэж байгаа юм.', } ``` ### Data Fields - `news`: a complete news article on a specific topic as a string - `label`: the single class of the topic, among these values: "урлаг соёл" (0), "эдийн засаг" (1), "эрүүл мэнд" (2), "хууль" (3), "улс төр" (4), "спорт" (5), "технологи" (6), "боловсрол" (7), "байгал орчин" (8). ### Data Splits The set of complete articles is split into a training and test set. ## Dataset Creation ### Curation Rationale [Needs More Information] ### Source Data #### Initial Data Collection and Normalization [Needs More Information] #### Who are the source language producers? Eduge.mn which is a combination from shuud.mn, ikon.mn, olloo.mn, news.gogo.mn, montsame.mn, zaluu.com, sonin.mn, medee.mn, bloombergtv.mn. ### Annotations #### Annotation process [Needs More Information] #### Who are the annotators? [Needs More Information] ### Personal and Sensitive Information [Needs More Information] ## Considerations for Using the Data ### Social Impact of Dataset [Needs More Information] ### Discussion of Biases [Needs More Information] ### Other Known Limitations [Needs More Information] ## Additional Information ### Dataset Curators [Needs More Information] ### Licensing Information [Needs More Information] ### Citation Information No citation available for this dataset. ### Contributions Thanks to [@enod](https://github.com/enod) for adding this dataset.
eitb_parcc
2022-11-03T16:15:31.000Z
[ "task_categories:translation", "annotations_creators:found", "language_creators:found", "multilinguality:multilingual", "size_categories:100K<n<1M", "source_datasets:original", "language:es", "language:eu", "license:unknown", "region:us" ]
null
EiTB-ParCC: Parallel Corpus of Comparable News. A Basque-Spanish parallel corpus provided by Vicomtech (https://www.vicomtech.org), extracted from comparable news produced by the Basque public broadcasting group Euskal Irrati Telebista.
@InProceedings{TIEDEMANN12.463, author = {J{\"o}rg Tiedemann}, title = {Parallel Data, Tools and Interfaces in OPUS}, booktitle = {Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC'12)}, year = {2012}, month = {may}, date = {23-25}, address = {Istanbul, Turkey}, editor = {Nicoletta Calzolari (Conference Chair) and Khalid Choukri and Thierry Declerck and Mehmet Ugur Dogan and Bente Maegaard and Joseph Mariani and Jan Odijk and Stelios Piperidis}, publisher = {European Language Resources Association (ELRA)}, isbn = {978-2-9517408-7-7}, language = {english} }
null
1
3
--- annotations_creators: - found language_creators: - found language: - es - eu license: - unknown multilinguality: - multilingual size_categories: - 100K<n<1M source_datasets: - original task_categories: - translation task_ids: [] paperswithcode_id: eitb-parcc pretty_name: EiTB-ParCC dataset_info: features: - name: translation dtype: translation: languages: - es - eu config_name: es-eu splits: - name: train num_bytes: 139039398 num_examples: 637183 download_size: 57244346 dataset_size: 139039398 --- # Dataset Card for [Dataset Name] ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:**[EiTB-ParCC: Parallel Corpus of Comparable New](http://opus.nlpl.eu/EiTB-ParCC.php) - **Repository:** - **Paper:** - **Leaderboard:** - **Point of Contact:** ### Dataset Summary EiTB-ParCC: Parallel Corpus of Comparable News. A Basque-Spanish parallel corpus provided by \ Vicomtech (https://www.vicomtech.org), extracted from comparable news produced by the \ Basque public broadcasting group Euskal Irrati Telebista. ### Supported Tasks and Leaderboards The underlying task is machine translation. ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information ``` @InProceedings{TIEDEMANN12.463, author = {J{\"o}rg Tiedemann}, title = {Parallel Data, Tools and Interfaces in OPUS}, booktitle = {Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC'12)}, year = {2012}, month = {may}, date = {23-25}, address = {Istanbul, Turkey}, editor = {Nicoletta Calzolari (Conference Chair) and Khalid Choukri and Thierry Declerck and Mehmet Ugur Dogan and Bente Maegaard and Joseph Mariani and Jan Odijk and Stelios Piperidis}, publisher = {European Language Resources Association (ELRA)}, isbn = {978-2-9517408-7-7}, language = {english} } ``` ### Contributions Thanks to [@patil-suraj](https://github.com/patil-suraj) for adding this dataset.
eth_py150_open
2022-11-18T20:01:17.000Z
[ "task_categories:other", "annotations_creators:no-annotation", "language_creators:machine-generated", "multilinguality:monolingual", "size_categories:100K<n<1M", "source_datasets:original", "language:en", "license:apache-2.0", "contextual-embeddings", "region:us" ]
null
A redistributable subset of the ETH Py150 corpus, introduced in the ICML 2020 paper 'Learning and Evaluating Contextual Embedding of Source Code'
@inproceedings{kanade2020learning, title={Learning and Evaluating Contextual Embedding of Source Code}, author={Kanade, Aditya and Maniatis, Petros and Balakrishnan, Gogul and Shi, Kensen}, booktitle={International Conference on Machine Learning}, pages={5110--5121}, year={2020}, organization={PMLR} }
null
0
3
--- annotations_creators: - no-annotation language_creators: - machine-generated language: - en license: - apache-2.0 multilinguality: - monolingual size_categories: - 100K<n<1M source_datasets: - original task_categories: - other task_ids: [] paperswithcode_id: eth-py150-open pretty_name: ethpy150open tags: - contextual-embeddings dataset_info: features: - name: filepath dtype: string - name: license dtype: string config_name: eth_py150_open splits: - name: train num_bytes: 5414978 num_examples: 74749 - name: test num_bytes: 3006199 num_examples: 41457 - name: validation num_bytes: 598524 num_examples: 8302 download_size: 13875671 dataset_size: 9019701 --- # Dataset Card for ethpy150open ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** https://www.sri.inf.ethz.ch/py150 - **Repository:** https://github.com/google-research-datasets/eth_py150_open - **Paper:** https://proceedings.icml.cc/static/paper_files/icml/2020/5401-Paper.pdf - **Leaderboard:** None - **Point of Contact:** Aditya Kanade <kanade@iisc.ac.in>, Petros Maniatis <maniatis@google.com> ### Dataset Summary A redistributable subset of the [ETH Py150 corpus](https://www.sri.inf.ethz.ch/py150), introduced in the ICML 2020 paper ['Learning and Evaluating Contextual Embedding of Source Code'](https://proceedings.icml.cc/static/paper_files/icml/2020/5401-Paper.pdf) ### Supported Tasks and Leaderboards [More Information Needed] ### Languages English ## Dataset Structure List of dicts of { "filepath": The relative URL containing the path to the file on GitHub "license": The license used for that specific file or repository } ### Data Instances { "filepath": "0rpc/zerorpc-python/setup.py", "license": "mit" }, { "filepath": "0rpc/zerorpc-python/zerorpc/heartbeat.py", "license": "mit" }, ### Data Fields - `filepath`: The relative URL containing the path to the file on GitHub - `license`: The license used for that specific file or repository ### Data Splits | | Train | Valid | Test | | ----- | ------- | ----- | ----- | | Dataset Split | 74749 | 8302 | 41457 | ## Dataset Creation The original dataset is at https://www.sri.inf.ethz.ch/py150 ### Curation Rationale To generate a more redistributable version of the dataset ### Source Data #### Initial Data Collection and Normalization All the urls are filepaths relative to GitHub and the master branch was used as available at the time #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information Apache License 2.0 ### Citation Information @inproceedings{kanade2020learning, title={Learning and Evaluating Contextual Embedding of Source Code}, author={Kanade, Aditya and Maniatis, Petros and Balakrishnan, Gogul and Shi, Kensen}, booktitle={International Conference on Machine Learning}, pages={5110--5121}, year={2020}, organization={PMLR} } ### Contributions Thanks to [@Bharat123rox](https://github.com/Bharat123rox) for adding this dataset.
finer
2023-01-25T14:30:30.000Z
[ "task_categories:token-classification", "task_ids:named-entity-recognition", "annotations_creators:expert-generated", "language_creators:other", "multilinguality:monolingual", "size_categories:10K<n<100K", "source_datasets:original", "language:fi", "license:mit", "arxiv:1908.04212", "region:us" ...
null
The directory data contains a corpus of Finnish technology related news articles with a manually prepared named entity annotation (digitoday.2014.csv). The text material was extracted from the archives of Digitoday, a Finnish online technology news source (www.digitoday.fi). The corpus consists of 953 articles (193,742 word tokens) with six named entity classes (organization, location, person, product, event, and date). The corpus is available for research purposes and can be readily used for development of NER systems for Finnish.
@article{ruokolainen2019finnish, title={A finnish news corpus for named entity recognition}, author={Ruokolainen, Teemu and Kauppinen, Pekka and Silfverberg, Miikka and Lind{\'e}n, Krister}, journal={Language Resources and Evaluation}, pages={1--26}, year={2019}, publisher={Springer} }
null
1
3
--- annotations_creators: - expert-generated language_creators: - other language: - fi license: - mit multilinguality: - monolingual size_categories: - 10K<n<100K source_datasets: - original task_categories: - token-classification task_ids: - named-entity-recognition paperswithcode_id: finer pretty_name: Finnish News Corpus for Named Entity Recognition dataset_info: features: - name: id dtype: string - name: tokens sequence: string - name: ner_tags sequence: class_label: names: '0': O '1': B-DATE '2': B-EVENT '3': B-LOC '4': B-ORG '5': B-PER '6': B-PRO '7': I-DATE '8': I-EVENT '9': I-LOC '10': I-ORG '11': I-PER '12': I-PRO - name: nested_ner_tags sequence: class_label: names: '0': O '1': B-DATE '2': B-EVENT '3': B-LOC '4': B-ORG '5': B-PER '6': B-PRO '7': I-DATE '8': I-EVENT '9': I-LOC '10': I-ORG '11': I-PER '12': I-PRO config_name: finer splits: - name: train num_bytes: 5159550 num_examples: 13497 - name: validation num_bytes: 387494 num_examples: 986 - name: test num_bytes: 1327354 num_examples: 3512 - name: test_wikipedia num_bytes: 1404397 num_examples: 3360 download_size: 3733127 dataset_size: 8278795 --- # Dataset Card for [Dataset Name] ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [Github](https://github.com/mpsilfve/finer-data) - **Repository:** [Github](https://github.com/mpsilfve/finer-data) - **Paper:** [Arxiv](https://arxiv.org/abs/1908.04212) - **Leaderboard:** - **Point of Contact:** ### Dataset Summary [More Information Needed] ### Supported Tasks and Leaderboards [More Information Needed] ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields Each row consists of the following fields: * `id`: The sentence id * `tokens`: An ordered list of tokens from the full text * `ner_tags`: Named entity recognition tags for each token * `nested_ner_tags`: Nested named entity recognition tags for each token Note that by design, the length of `tokens`, `ner_tags`, and `nested_ner_tags` will always be identical. `ner_tags` and `nested_ner_tags` correspond to the list below: ``` [ "O", "B-DATE", "B-EVENT", "B-LOC", "B-ORG", "B-PER", "B-PRO", "I-DATE", "I-EVENT", "I-LOC", "I-ORG", "I-PER", "I-PRO" ] ``` IOB2 labeling scheme is used. ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information [More Information Needed] ### Contributions Thanks to [@stefan-it](https://github.com/stefan-it) for adding this dataset.
fquad
2023-04-05T10:06:27.000Z
[ "task_categories:question-answering", "task_categories:text-retrieval", "task_ids:extractive-qa", "task_ids:closed-domain-qa", "annotations_creators:crowdsourced", "language_creators:crowdsourced", "language_creators:found", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datase...
null
FQuAD: French Question Answering Dataset We introduce FQuAD, a native French Question Answering Dataset. FQuAD contains 25,000+ question and answer pairs. Finetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%.
@ARTICLE{2020arXiv200206071 author = {Martin, d'Hoffschmidt and Maxime, Vidal and Wacim, Belblidia and Tom, Brendlé}, title = "{FQuAD: French Question Answering Dataset}", journal = {arXiv e-prints}, keywords = {Computer Science - Computation and Language}, year = "2020", month = "Feb", eid = {arXiv:2002.06071}, pages = {arXiv:2002.06071}, archivePrefix = {arXiv}, eprint = {2002.06071}, primaryClass = {cs.CL} }
null
7
3
--- annotations_creators: - crowdsourced language_creators: - crowdsourced - found language: - fr license: - cc-by-nc-sa-3.0 multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - question-answering - text-retrieval task_ids: - extractive-qa - closed-domain-qa paperswithcode_id: fquad pretty_name: 'FQuAD: French Question Answering Dataset' dataset_info: features: - name: context dtype: string - name: questions sequence: string - name: answers sequence: - name: texts dtype: string - name: answers_starts dtype: int32 splits: - name: train num_bytes: 5898752 num_examples: 4921 - name: validation num_bytes: 1031456 num_examples: 768 download_size: 0 dataset_size: 6930208 --- # Dataset Card for FQuAD ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [https://fquad.illuin.tech/](https://fquad.illuin.tech/) - **Paper:** [FQuAD: French Question Answering Dataset](https://arxiv.org/abs/2002.06071) - **Point of Contact:** [https://www.illuin.tech/contact/](https://www.illuin.tech/contact/) - **Size of downloaded dataset files:** 3.29 MB - **Size of the generated dataset:** 6.94 MB - **Total amount of disk used:** 10.23 MB ### Dataset Summary FQuAD: French Question Answering Dataset We introduce FQuAD, a native French Question Answering Dataset. FQuAD contains 25,000+ question and answer pairs. Finetuning CamemBERT on FQuAD yields a F1 score of 88% and an exact match of 77.9%. Developped to provide a SQuAD equivalent in the French language. Questions are original and based on high quality Wikipedia articles. Please, note this dataset is licensed for non-commercial purposes and users must agree to the following terms and conditions: 1. Use FQuAD only for internal research purposes. 2. Not make any copy except a safety one. 3. Not redistribute it (or part of it) in any way, even for free. 4. Not sell it or use it for any commercial purpose. Contact us for a possible commercial licence. 5. Mention the corpus origin and Illuin Technology in all publications about experiments using FQuAD. 6. Redistribute to Illuin Technology any improved or enriched version you could make of that corpus. Request manually download of the data from: https://fquad.illuin.tech/ ### Supported Tasks and Leaderboards - `closed-domain-qa`, `text-retrieval`: This dataset is intended to be used for `closed-domain-qa`, but can also be used for information retrieval tasks. ### Languages This dataset is exclusively in French, with context data from Wikipedia and questions from French university students (`fr`). ## Dataset Structure ### Data Instances #### default - **Size of downloaded dataset files:** 3.29 MB - **Size of the generated dataset:** 6.94 MB - **Total amount of disk used:** 10.23 MB An example of 'validation' looks as follows. ``` This example was too long and was cropped: { "answers": { "answers_starts": [161, 46, 204], "texts": ["La Vierge aux rochers", "documents contemporains", "objets de spéculations"] }, "context": "\"Les deux tableaux sont certes décrits par des documents contemporains à leur création mais ceux-ci ne le font qu'indirectement ...", "questions": ["Que concerne principalement les documents ?", "Par quoi sont décrit les deux tableaux ?", "Quels types d'objets sont les deux tableaux aux yeux des chercheurs ?"] } ``` ### Data Fields The data fields are the same among all splits. #### default - `context`: a `string` feature. - `questions`: a `list` of `string` features. - `answers`: a dictionary feature containing: - `texts`: a `string` feature. - `answers_starts`: a `int32` feature. ### Data Splits The FQuAD dataset has 3 splits: _train_, _validation_, and _test_. The _test_ split is however not released publicly at the moment. The splits contain disjoint sets of articles. The following table contains stats about each split. Dataset Split | Number of Articles in Split | Number of paragraphs in split | Number of questions in split --------------|------------------------------|--------------------------|------------------------- Train | 117 | 4921 | 20731 Validation | 768 | 51.0% | 3188 Test | 10 | 532 | 2189 ## Dataset Creation ### Curation Rationale The FQuAD dataset was created by Illuin technology. It was developped to provide a SQuAD equivalent in the French language. Questions are original and based on high quality Wikipedia articles. ### Source Data The text used for the contexts are from the curated list of French High-Quality Wikipedia [articles](https://fr.wikipedia.org/wiki/Cat%C3%A9gorie:Article_de_qualit%C3%A9). ### Annotations Annotations (spans and questions) are written by students of the CentraleSupélec school of engineering. Wikipedia articles were scraped and Illuin used an internally-developped tool to help annotators ask questions and indicate the answer spans. Annotators were given paragraph sized contexts and asked to generate 4/5 non-trivial questions about information in the context. ### Personal and Sensitive Information No personal or sensitive information is included in this dataset. This has been manually verified by the dataset curators. ## Considerations for Using the Data Users should consider this dataset is sampled from Wikipedia data which might not be representative of all QA use cases. ### Social Impact of Dataset The social biases of this dataset have not yet been investigated. ### Discussion of Biases The social biases of this dataset have not yet been investigated, though articles have been selected by their quality and objectivity. ### Other Known Limitations The limitations of the FQuAD dataset have not yet been investigated. ## Additional Information ### Dataset Curators Illuin Technology: [https://fquad.illuin.tech/](https://fquad.illuin.tech/) ### Licensing Information The FQuAD dataset is licensed under the [CC BY-NC-SA 3.0](https://creativecommons.org/licenses/by-nc-sa/3.0/fr/) license. It allows personal and academic research uses of the dataset, but not commercial uses. So concretely, the dataset cannot be used to train a model that is then put into production within a business or a company. For this type of commercial use, we invite FQuAD users to contact [the authors](https://www.illuin.tech/contact/) to discuss possible partnerships. ### Citation Information ``` @ARTICLE{2020arXiv200206071 author = {Martin, d'Hoffschmidt and Maxime, Vidal and Wacim, Belblidia and Tom, Brendlé}, title = "{FQuAD: French Question Answering Dataset}", journal = {arXiv e-prints}, keywords = {Computer Science - Computation and Language}, year = "2020", month = "Feb", eid = {arXiv:2002.06071}, pages = {arXiv:2002.06071}, archivePrefix = {arXiv}, eprint = {2002.06071}, primaryClass = {cs.CL} } ``` ### Contributions Thanks to [@thomwolf](https://github.com/thomwolf), [@mariamabarham](https://github.com/mariamabarham), [@patrickvonplaten](https://github.com/patrickvonplaten), [@lewtun](https://github.com/lewtun), [@albertvillanova](https://github.com/albertvillanova) for adding this dataset. Thanks to [@ManuelFay](https://github.com/manuelfay) for providing information on the dataset creation process.
giga_fren
2022-11-03T16:15:21.000Z
[ "task_categories:translation", "annotations_creators:found", "language_creators:found", "multilinguality:multilingual", "size_categories:10M<n<100M", "source_datasets:original", "language:en", "language:fr", "license:unknown", "region:us" ]
null
Giga-word corpus for French-English from WMT2010 collected by Chris Callison-Burch 2 languages, total number of files: 452 total number of tokens: 1.43G total number of sentence fragments: 47.55M
@InProceedings{TIEDEMANN12.463, author = {J{\"o}rg Tiedemann}, title = {Parallel Data, Tools and Interfaces in OPUS}, booktitle = {Proceedings of the Eight International Conference on Language Resources and Evaluation (LREC'12)}, year = {2012}, month = {may}, date = {23-25}, address = {Istanbul, Turkey}, editor = {Nicoletta Calzolari (Conference Chair) and Khalid Choukri and Thierry Declerck and Mehmet Ugur Dogan and Bente Maegaard and Joseph Mariani and Jan Odijk and Stelios Piperidis}, publisher = {European Language Resources Association (ELRA)}, isbn = {978-2-9517408-7-7}, language = {english} }
null
0
3
--- annotations_creators: - found language_creators: - found language: - en - fr license: - unknown multilinguality: - multilingual size_categories: - 10M<n<100M source_datasets: - original task_categories: - translation task_ids: [] paperswithcode_id: null pretty_name: GigaFren dataset_info: features: - name: id dtype: string - name: translation dtype: translation: languages: - en - fr config_name: en-fr splits: - name: train num_bytes: 8690296821 num_examples: 22519904 download_size: 2701536198 dataset_size: 8690296821 --- # Dataset Card for GigaFren ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** http://opus.nlpl.eu/giga-fren.php - **Repository:** None - **Paper:** http://www.lrec-conf.org/proceedings/lrec2012/pdf/463_Paper.pdf - **Leaderboard:** [More Information Needed] - **Point of Contact:** [More Information Needed] ### Dataset Summary [More Information Needed] ### Supported Tasks and Leaderboards [More Information Needed] ### Languages [More Information Needed] ## Dataset Structure ### Data Instances Here are some examples of questions and facts: ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data [More Information Needed] #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations [More Information Needed] #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information [More Information Needed] ### Contributions Thanks to [@abhishekkrthakur](https://github.com/abhishekkrthakur) for adding this dataset.
hausa_voa_ner
2023-01-25T14:31:51.000Z
[ "task_categories:token-classification", "task_ids:named-entity-recognition", "annotations_creators:expert-generated", "language_creators:expert-generated", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "language:ha", "license:cc-by-4.0", "region:us" ]
null
The Hausa VOA NER dataset is a labeled dataset for named entity recognition in Hausa. The texts were obtained from Hausa Voice of America News articles https://www.voahausa.com/ . We concentrate on four types of named entities: persons [PER], locations [LOC], organizations [ORG], and dates & time [DATE]. The Hausa VOA NER data files contain 2 columns separated by a tab ('\t'). Each word has been put on a separate line and there is an empty line after each sentences i.e the CoNLL format. The first item on each line is a word, the second is the named entity tag. The named entity tags have the format I-TYPE which means that the word is inside a phrase of type TYPE. For every multi-word expression like 'New York', the first word gets a tag B-TYPE and the subsequent words have tags I-TYPE, a word with tag O is not part of a phrase. The dataset is in the BIO tagging scheme. For more details, see https://www.aclweb.org/anthology/2020.emnlp-main.204/
@inproceedings{hedderich-etal-2020-transfer, title = "Transfer Learning and Distant Supervision for Multilingual Transformer Models: A Study on {A}frican Languages", author = "Hedderich, Michael A. and Adelani, David and Zhu, Dawei and Alabi, Jesujoba and Markus, Udia and Klakow, Dietrich", booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)", month = nov, year = "2020", address = "Online", publisher = "Association for Computational Linguistics", url = "https://www.aclweb.org/anthology/2020.emnlp-main.204", doi = "10.18653/v1/2020.emnlp-main.204", pages = "2580--2591", }
null
2
3
--- annotations_creators: - expert-generated language_creators: - expert-generated language: - ha license: - cc-by-4.0 multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - token-classification task_ids: - named-entity-recognition pretty_name: Hausa VOA NER Corpus dataset_info: features: - name: id dtype: string - name: tokens sequence: string - name: ner_tags sequence: class_label: names: '0': O '1': B-PER '2': I-PER '3': B-ORG '4': I-ORG '5': B-LOC '6': I-LOC '7': B-DATE '8': I-DATE config_name: hausa_voa_ner splits: - name: train num_bytes: 483634 num_examples: 1015 - name: validation num_bytes: 69673 num_examples: 146 - name: test num_bytes: 139227 num_examples: 292 download_size: 324962 dataset_size: 692534 --- # Dataset Card for Hausa VOA NER Corpus ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** https://www.aclweb.org/anthology/2020.emnlp-main.204/ - **Repository:** [Hausa VOA NER](https://github.com/uds-lsv/transfer-distant-transformer-african/tree/master/data/hausa_ner) - **Paper:** https://www.aclweb.org/anthology/2020.emnlp-main.204/ - **Leaderboard:** - **Point of Contact:** [David Adelani](mailto:didelani@lsv.uni-saarland.de) ### Dataset Summary The Hausa VOA NER is a named entity recognition (NER) dataset for Hausa language based on the [VOA Hausa news](https://www.voahausa.com/) corpus. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages The language supported is Hausa. ## Dataset Structure ### Data Instances A data point consists of sentences seperated by empty line and tab-seperated tokens and tags. {'id': '0', 'ner_tags': [B-PER, 0, 0, B-LOC, 0], 'tokens': ['Trump', 'ya', 'ce', 'Rasha', 'ma'] } ### Data Fields - `id`: id of the sample - `tokens`: the tokens of the example text - `ner_tags`: the NER tags of each token The NER tags correspond to this list: ``` "O", "B-PER", "I-PER", "B-ORG", "I-ORG", "B-LOC", "I-LOC", "B-DATE", "I-DATE", ``` The NER tags have the same format as in the CoNLL shared task: a B denotes the first item of a phrase and an I any non-initial word. There are four types of phrases: person names (PER), organizations (ORG), locations (LOC) and dates & times (DATE). (O) is used for tokens not considered part of any named entity. ### Data Splits Training (1,014 sentences), validation (145 sentences) and test split (291 sentences) ## Dataset Creation ### Curation Rationale The data was created to help introduce resources to new language - Hausa. [More Information Needed] ### Source Data #### Initial Data Collection and Normalization The dataset is based on the news domain and was crawled from [VOA Hausa news](https://www.voahausa.com/). [More Information Needed] #### Who are the source language producers? The dataset was collected from VOA Hausa news. Most of the texts used in creating the Hausa VOA NER are news stories from Nigeria, Niger Republic, United States, and other parts of the world. [More Information Needed] ### Annotations Named entity recognition annotation #### Annotation process [More Information Needed] #### Who are the annotators? The data was annotated by Jesujoba Alabi and David Adelani for the paper: [Transfer Learning and Distant Supervision for Multilingual Transformer Models: A Study on African Languages](https://www.aclweb.org/anthology/2020.emnlp-main.204/). [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators The annotated data sets were developed by students of Saarland University, Saarbrücken, Germany . ### Licensing Information The data is under the [Creative Commons Attribution 4.0 ](https://creativecommons.org/licenses/by/4.0/) ### Citation Information ``` @inproceedings{hedderich-etal-2020-transfer, title = "Transfer Learning and Distant Supervision for Multilingual Transformer Models: A Study on {A}frican Languages", author = "Hedderich, Michael A. and Adelani, David and Zhu, Dawei and Alabi, Jesujoba and Markus, Udia and Klakow, Dietrich", booktitle = "Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)", month = nov, year = "2020", address = "Online", publisher = "Association for Computational Linguistics", url = "https://www.aclweb.org/anthology/2020.emnlp-main.204", doi = "10.18653/v1/2020.emnlp-main.204", pages = "2580--2591", } ``` ### Contributions Thanks to [@dadelani](https://github.com/dadelani) for adding this dataset.
hebrew_this_world
2022-11-03T16:08:08.000Z
[ "task_categories:text-generation", "task_categories:fill-mask", "task_ids:language-modeling", "task_ids:masked-language-modeling", "annotations_creators:expert-generated", "language_creators:found", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "language:he...
null
HebrewThisWorld is a data set consists of 2028 issues of the newspaper 'This World' edited by Uri Avnery and were published between 1950 and 1989. Released under the AGPLv3 license.
null
null
1
3
--- annotations_creators: - expert-generated language_creators: - found language: - he license: - agpl-3.0 multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - text-generation - fill-mask task_ids: - language-modeling - masked-language-modeling paperswithcode_id: null pretty_name: HebrewSentiment dataset_info: features: - name: issue_num dtype: int64 - name: page_count dtype: int64 - name: date dtype: string - name: date_he dtype: string - name: year dtype: string - name: href dtype: string - name: pdf dtype: string - name: coverpage dtype: string - name: backpage dtype: string - name: content dtype: string - name: url dtype: string splits: - name: train num_bytes: 678389435 num_examples: 2028 download_size: 678322912 dataset_size: 678389435 --- # Dataset Card for HebrewSentiment ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** https://thisworld.online/ - **Repository:** https://github.com/thisworld1/thisworld.online - **Paper:** - **Leaderboard:** - **Point of Contact:** ### Dataset Summary HebrewThisWorld is a data set consists of 2028 issues of the newspaper 'This World' edited by Uri Avnery and were published between 1950 and 1989. Released under the AGPLv3 license. Data Annotation: ### Supported Tasks and Leaderboards Language modeling ### Languages Hebrew ## Dataset Structure csv file with "," delimeter ### Data Instances Sample: ```json { "issue_num": 637, "page_count": 16, "date": "1950-01-01", "date_he": "1 בינואר 1950", "year": "1950", "href": "https://thisworld.online/1950/637", "pdf": "https://olam.eu-central-1.linodeobjects.com/pdfs/B-I0637-D010150.pdf", "coverpage": "https://olam.eu-central-1.linodeobjects.com/pages/637/t-1.png", "backpage": "https://olam.eu-central-1.linodeobjects.com/pages/637/t-16.png", "content": "\nלפיד\nהנוער ־ בירושלים צילומים :\n\nב. רותנברג\n\nוזהו הלפיד\n...", "url": "https://thisworld.online/api/1950/637" } ``` ### Data Fields - `issue_num`: ID/Number of the issue - `page_count`: Page count of the current issue - `date`: Published date - `date_he`: Published date in Hebrew - `year`: Year of the issue - `href`: URL to the issue to scan/print etc. - `pdf`: URL to the issue to scan in pdf - `coverpage`: URL to coverpage - `backpage`: URL to backpage - `content`: text content of the issue - `url`: URL ### Data Splits | | train | |--------|------:| | corpus | 2028 | ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data [thisworld.online](https://thisworld.online/) #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? Researchers ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information GNU AGPLv3+ This is free software, and you are welcome to redistribute it under certain conditions. This program is free software: you can redistribute it and/or modify it under the terms of the GNU Affero General Public License as published by the Free Software Foundation, either version 3 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU Affero General Public License for more details. You should have received a copy of the GNU Affero General Public License along with this program. If not, see <http://www.gnu.org/licenses/>. ### Citation Information https://thisworld.online/ ### Contributions Thanks to [@lhoestq](https://github.com/lhoestq), [@imvladikon](https://github.com/imvladikon) for adding this dataset.
hrenwac_para
2022-11-03T16:07:49.000Z
[ "task_categories:translation", "annotations_creators:no-annotation", "language_creators:found", "multilinguality:translation", "size_categories:10K<n<100K", "source_datasets:original", "language:en", "language:hr", "license:cc-by-sa-3.0", "region:us" ]
null
The hrenWaC corpus version 2.0 consists of parallel Croatian-English texts crawled from the .hr top-level domain for Croatia. The corpus was built with Spidextor (https://github.com/abumatran/spidextor), a tool that glues together the output of SpiderLing used for crawling and Bitextor used for bitext extraction. The accuracy of the extracted bitext on the segment level is around 80% and on the word level around 84%.
@misc{11356/1058, title = {Croatian-English parallel corpus {hrenWaC} 2.0}, author = {Ljube{\v s}i{\'c}, Nikola and Espl{\'a}-Gomis, Miquel and Ortiz Rojas, Sergio and Klubi{\v c}ka, Filip and Toral, Antonio}, url = {http://hdl.handle.net/11356/1058}, note = {Slovenian language resource repository {CLARIN}.{SI}}, copyright = {{CLARIN}.{SI} User Licence for Internet Corpora}, year = {2016} }
null
0
3
--- annotations_creators: - no-annotation language_creators: - found language: - en - hr license: - cc-by-sa-3.0 multilinguality: - translation size_categories: - 10K<n<100K source_datasets: - original task_categories: - translation task_ids: [] paperswithcode_id: null pretty_name: HrenwacPara dataset_info: features: - name: translation dtype: translation: languages: - en - hr config_name: hrenWaC splits: - name: train num_bytes: 29602110 num_examples: 99001 download_size: 11640281 dataset_size: 29602110 --- # Dataset Card for hrenwac_para ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** http://nlp.ffzg.hr/resources/corpora/hrenwac/ - **Repository:** http://nlp.ffzg.hr/data/corpora/hrenwac/hrenwac.en-hr.txt.gz - **Paper:** http://workshop2013.iwslt.org/downloads/IWSLT-2013-Cettolo.pdf - **Leaderboard:** - **Point of Contact:** [Nikola Ljubešič](mailto:nikola.ljubesic@ffzg.hr) ### Dataset Summary The hrenWaC corpus version 2.0 consists of parallel Croatian-English texts crawled from the .hr top-level domain for Croatia. The corpus was built with Spidextor (https://github.com/abumatran/spidextor), a tool that glues together the output of SpiderLing used for crawling and Bitextor used for bitext extraction. The accuracy of the extracted bitext on the segment level is around 80% and on the word level around 84%. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages Dataset is bilingual with Croatian and English languages. ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information Dataset is under the [CC-BY-SA 3.0](http://creativecommons.org/licenses/by-sa/3.0/) license. ### Citation Information ``` @misc{11356/1058, title = {Croatian-English parallel corpus {hrenWaC} 2.0}, author = {Ljube{\v s}i{\'c}, Nikola and Espl{\`a}-Gomis, Miquel and Ortiz Rojas, Sergio and Klubi{\v c}ka, Filip and Toral, Antonio}, url = {http://hdl.handle.net/11356/1058}, note = {Slovenian language resource repository {CLARIN}.{SI}}, copyright = {{CLARIN}.{SI} User Licence for Internet Corpora}, year = {2016} } ``` ### Contributions Thanks to [@IvanZidov](https://github.com/IvanZidov) for adding this dataset.
hrwac
2022-11-03T16:15:15.000Z
[ "task_categories:text-generation", "task_categories:fill-mask", "task_ids:language-modeling", "task_ids:masked-language-modeling", "annotations_creators:no-annotation", "language_creators:found", "multilinguality:monolingual", "size_categories:1B<n<10B", "source_datasets:original", "language:hr", ...
null
The Croatian web corpus hrWaC was built by crawling the .hr top-level domain in 2011 and again in 2014. The corpus was near-deduplicated on paragraph level, normalised via diacritic restoration, morphosyntactically annotated and lemmatised. The corpus is shuffled by paragraphs. Each paragraph contains metadata on the URL, domain and language identification (Croatian vs. Serbian). Version 2.0 of this corpus is described in http://www.aclweb.org/anthology/W14-0405. Version 2.1 contains newer and better linguistic annotations.
@misc{11356/1064, title = {Croatian web corpus {hrWaC} 2.1}, author = {Ljube{\v s}i{\'c}, Nikola and Klubi{\v c}ka, Filip}, url = {http://hdl.handle.net/11356/1064}, note = {Slovenian language resource repository {CLARIN}.{SI}}, copyright = {Creative Commons - Attribution-{ShareAlike} 4.0 International ({CC} {BY}-{SA} 4.0)}, year = {2016} }
null
0
3
--- annotations_creators: - no-annotation language_creators: - found language: - hr license: - cc-by-sa-3.0 multilinguality: - monolingual size_categories: - 1B<n<10B source_datasets: - original task_categories: - text-generation - fill-mask task_ids: - language-modeling - masked-language-modeling paperswithcode_id: null pretty_name: HrWac dataset_info: features: - name: sentence dtype: string config_name: hrwac splits: - name: train num_bytes: 43994569015 num_examples: 1736944727 download_size: 9217221471 dataset_size: 43994569015 --- # Dataset Card for HrWac ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** http://nlp.ffzg.hr/resources/corpora/hrwac/ - **Repository:** https://www.clarin.si/repository/xmlui/handle/11356/1064 - **Paper:** http://nlp.ffzg.hr/data/publications/nljubesi/ljubesic11-hrwac.pdf - **Leaderboard:** - **Point of Contact:** [Nikola Ljubešič](mailto:nikola.ljubesic@ffzg.hr) ### Dataset Summary The Croatian web corpus hrWaC was built by crawling the .hr top-level domain in 2011 and again in 2014. The corpus was near-deduplicated on paragraph level, normalised via diacritic restoration, morphosyntactically annotated and lemmatised. The corpus is shuffled by paragraphs. Each paragraph contains metadata on the URL, domain and language identification (Croatian vs. Serbian). ### Supported Tasks and Leaderboards [More Information Needed] ### Languages Dataset is monolingual in Croatian language. ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields - sentence: sentences as strings ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information Dataset is under the [CC-BY-SA 3.0](http://creativecommons.org/licenses/by-sa/3.0/) license. ### Citation Information ``` @misc{11356/1064, title = {Croatian web corpus {hrWaC} 2.1}, author = {Ljube{\v s}i{\'c}, Nikola and Klubi{\v c}ka, Filip}, url = {http://hdl.handle.net/11356/1064}, note = {Slovenian language resource repository {CLARIN}.{SI}}, copyright = {Creative Commons - Attribution-{ShareAlike} 4.0 International ({CC} {BY}-{SA} 4.0)}, year = {2016} } ``` ### Contributions Thanks to [@IvanZidov](https://github.com/IvanZidov) for adding this dataset.
id_puisi
2022-11-03T16:08:09.000Z
[ "task_categories:text2text-generation", "task_categories:text-generation", "task_categories:fill-mask", "annotations_creators:no-annotation", "language_creators:found", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "language:id", "license:mit", "poem-gene...
null
Puisi (poem) is an Indonesian poetic form. The dataset contains 7223 Indonesian puisi with its title and author.
null
null
2
3
--- annotations_creators: - no-annotation language_creators: - found language: - id license: - mit multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - text2text-generation - text-generation - fill-mask task_ids: [] paperswithcode_id: null pretty_name: Indonesian Puisi tags: - poem-generation dataset_info: features: - name: title dtype: string - name: author dtype: string - name: puisi dtype: string - name: puisi_with_header dtype: string splits: - name: train num_bytes: 10613475 num_examples: 7223 download_size: 10558108 dataset_size: 10613475 --- # Dataset Card for id_puisi ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [puisi-pantun-generator](https://github.com/ilhamfp/puisi-pantun-generator) - **Repository:** [puisi-pantun-generator](https://github.com/ilhamfp/puisi-pantun-generator) - **Paper:** [N/A] - **Leaderboard:** [N/A] - **Point of Contact:** [Ilham Firdausi Putra](ilhamfputra31@gmail.com) ### Dataset Summary Puisi (poem) is an Indonesian poetic form. The dataset contains 7223 Indonesian puisi with its title and author. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages Indonesian ## Dataset Structure ### Data Instances ``` { 'puisi_with_header': 'TEPERANGKAP Oleh Mangku Langit Jingga Mungkin kau membiarkan aku Membiarkan perasaan ini larut Memberi ruang jiwaku hampa Agar tetap terbiasa nikmati Perangkap yang kau buat Perisai yang kau banggakan Takkan jadi tameng bagimu Aku mengerti betapa hebatnya Perangkap mu hei sang dewi Ku akan terus merasa terbiasa Dengan pesona indahmu Ku masih akan nikmati hadirmu Berjalanlah pada hati yang sama Satu hati denganku Walau ku terperangkap Namunku nikmati dan jalani', 'title': 'TEPERANGKAP', 'author': 'Oleh Mangku Langit Jingga', 'puisi': 'Mungkin kau membiarkan aku Membiarkan perasaan ini larut Memberi ruang jiwaku hampa Agar tetap terbiasa nikmati Perangkap yang kau buat Perisai yang kau banggakan Takkan jadi tameng bagimu Aku mengerti betapa hebatnya Perangkap mu hei sang dewi Ku akan terus merasa terbiasa Dengan pesona indahmu Ku masih akan nikmati hadirmu Berjalanlah pada hati yang sama Satu hati denganku Walau ku terperangkap Namunku nikmati dan jalani', } ``` ### Data Fields - `puisi_with_header`: the raw text from scraping - `title`: the title extracted from the raw text using regex - `author`: the author extracted from the raw text using regex - `puisi`: the poem with title and author extracted out using regex ### Data Splits The dataset contains only a train set. ## Dataset Creation ### Curation Rationale The dataset was initially collected as an experiment to generate an Indonesian poem using GPT-2. ### Source Data #### Initial Data Collection and Normalization The dataset was scraped using BeautifulSoup from lokerpuisi.web.id (the data no longer exist on the original blog). The title and author column was produced using regex match from puisi_with_header column. #### Who are the source language producers? The poems were generated by humans. The users of the original blog voluntarily submit their original poems to get published on the blog. ### Annotations #### Annotation process [N/A] #### Who are the annotators? [N/A] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations The regex match used to extract the title & author from the raw text is not perfect. Some title & text is still failed to get extracted. ## Additional Information ### Dataset Curators Ilham Firdausi Putra ### Licensing Information MIT License ### Citation Information [N/A] ### Contributions Thanks to [@ilhamfp](https://github.com/ilhamfp) for adding this dataset.
imppres
2023-01-25T14:32:53.000Z
[ "task_categories:text-classification", "task_ids:natural-language-inference", "annotations_creators:machine-generated", "language_creators:machine-generated", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "language:en", "license:cc-by-nc-4.0", "region:us" ]
null
Over >25k semiautomatically generated sentence pairs illustrating well-studied pragmatic inference types. IMPPRES is an NLI dataset following the format of SNLI (Bowman et al., 2015), MultiNLI (Williams et al., 2018) and XNLI (Conneau et al., 2018), which was created to evaluate how well trained NLI models recognize several classes of presuppositions and scalar implicatures.
@inproceedings{jeretic-etal-2020-natural, title = "Are Natural Language Inference Models {IMPPRESsive}? {L}earning {IMPlicature} and {PRESupposition}", author = "Jereti\v{c}, Paloma and Warstadt, Alex and Bhooshan, Suvrat and Williams, Adina", booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics", month = jul, year = "2020", address = "Online", publisher = "Association for Computational Linguistics", url = "https://www.aclweb.org/anthology/2020.acl-main.768", doi = "10.18653/v1/2020.acl-main.768", pages = "8690--8705", abstract = "Natural language inference (NLI) is an increasingly important task for natural language understanding, which requires one to infer whether a sentence entails another. However, the ability of NLI models to make pragmatic inferences remains understudied. We create an IMPlicature and PRESupposition diagnostic dataset (IMPPRES), consisting of 32K semi-automatically generated sentence pairs illustrating well-studied pragmatic inference types. We use IMPPRES to evaluate whether BERT, InferSent, and BOW NLI models trained on MultiNLI (Williams et al., 2018) learn to make pragmatic inferences. Although MultiNLI appears to contain very few pairs illustrating these inference types, we find that BERT learns to draw pragmatic inferences. It reliably treats scalar implicatures triggered by {``}some{''} as entailments. For some presupposition triggers like {``}only{''}, BERT reliably recognizes the presupposition as an entailment, even when the trigger is embedded under an entailment canceling operator like negation. BOW and InferSent show weaker evidence of pragmatic reasoning. We conclude that NLI training encourages models to learn some, but not all, pragmatic inferences.", }
null
0
3
--- annotations_creators: - machine-generated language_creators: - machine-generated language: - en license: - cc-by-nc-4.0 multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - text-classification task_ids: - natural-language-inference paperswithcode_id: imppres pretty_name: IMPPRES dataset_info: - config_name: presupposition_all_n_presupposition features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: all_n_presupposition num_bytes: 458492 num_examples: 1900 download_size: 335088 dataset_size: 458492 - config_name: presupposition_both_presupposition features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: both_presupposition num_bytes: 432792 num_examples: 1900 download_size: 335088 dataset_size: 432792 - config_name: presupposition_change_of_state features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: change_of_state num_bytes: 308627 num_examples: 1900 download_size: 335088 dataset_size: 308627 - config_name: presupposition_cleft_existence features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: cleft_existence num_bytes: 363238 num_examples: 1900 download_size: 335088 dataset_size: 363238 - config_name: presupposition_cleft_uniqueness features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: cleft_uniqueness num_bytes: 388779 num_examples: 1900 download_size: 335088 dataset_size: 388779 - config_name: presupposition_only_presupposition features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: only_presupposition num_bytes: 349018 num_examples: 1900 download_size: 335088 dataset_size: 349018 - config_name: presupposition_possessed_definites_existence features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: possessed_definites_existence num_bytes: 362334 num_examples: 1900 download_size: 335088 dataset_size: 362334 - config_name: presupposition_possessed_definites_uniqueness features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: possessed_definites_uniqueness num_bytes: 459403 num_examples: 1900 download_size: 335088 dataset_size: 459403 - config_name: presupposition_question_presupposition features: - name: premise dtype: string - name: hypothesis dtype: string - name: trigger dtype: string - name: trigger1 dtype: string - name: trigger2 dtype: string - name: presupposition dtype: string - name: gold_label dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: UID dtype: string - name: pairID dtype: string - name: paradigmID dtype: int16 splits: - name: question_presupposition num_bytes: 397227 num_examples: 1900 download_size: 335088 dataset_size: 397227 - config_name: implicature_connectives features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: connectives num_bytes: 221868 num_examples: 1200 download_size: 335088 dataset_size: 221868 - config_name: implicature_gradable_adjective features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: gradable_adjective num_bytes: 153672 num_examples: 1200 download_size: 335088 dataset_size: 153672 - config_name: implicature_gradable_verb features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: gradable_verb num_bytes: 180702 num_examples: 1200 download_size: 335088 dataset_size: 180702 - config_name: implicature_modals features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: modals num_bytes: 178560 num_examples: 1200 download_size: 335088 dataset_size: 178560 - config_name: implicature_numerals_10_100 features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: numerals_10_100 num_bytes: 208620 num_examples: 1200 download_size: 335088 dataset_size: 208620 - config_name: implicature_numerals_2_3 features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: numerals_2_3 num_bytes: 188784 num_examples: 1200 download_size: 335088 dataset_size: 188784 - config_name: implicature_quantifiers features: - name: premise dtype: string - name: hypothesis dtype: string - name: gold_label_log dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: gold_label_prag dtype: class_label: names: '0': entailment '1': neutral '2': contradiction - name: spec_relation dtype: string - name: item_type dtype: string - name: trigger dtype: string - name: lexemes dtype: string splits: - name: quantifiers num_bytes: 176814 num_examples: 1200 download_size: 335088 dataset_size: 176814 --- # Dataset Card for IMPPRES ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [Github](https://github.com/facebookresearch/Imppres) - **Repository:** [Github](https://github.com/facebookresearch/Imppres) - **Paper:** [Aclweb](https://www.aclweb.org/anthology/2020.acl-main.768) - **Leaderboard:** - **Point of Contact:** ### Dataset Summary Over >25k semiautomatically generated sentence pairs illustrating well-studied pragmatic inference types. IMPPRES is an NLI dataset following the format of SNLI (Bowman et al., 2015), MultiNLI (Williams et al., 2018) and XNLI (Conneau et al., 2018), which was created to evaluate how well trained NLI models recognize several classes of presuppositions and scalar implicatures. ### Supported Tasks and Leaderboards Natural Language Inference. ### Languages English. ## Dataset Structure ### Data Instances The data consists of 2 configurations: implicature and presupposition. Each configuration consists of several different sub-datasets: **Pressupposition** - all_n_presupposition - change_of_state - cleft_uniqueness - possessed_definites_existence - question_presupposition - both_presupposition - cleft_existence - only_presupposition - possessed_definites_uniqueness **Implicature** - connectives - gradable_adjective - gradable_verb - modals - numerals_10_100 - numerals_2_3 - quantifiers Each sentence type in IMPPRES is generated according to a template that specifies the linear order of the constituents in the sentence. The constituents are sampled from a vocabulary of over 3000 lexical items annotated with grammatical features needed to ensure wellformedness. We semiautomatically generate IMPPRES using a codebase developed by Warstadt et al. (2019a) and significantly expanded for the BLiMP dataset (Warstadt et al., 2019b). Here is an instance of the raw presupposition data from any sub-dataset: ```buildoutcfg { "sentence1": "All ten guys that proved to boast might have been divorcing.", "sentence2": "There are exactly ten guys that proved to boast.", "trigger": "modal", "presupposition": "positive", "gold_label": "entailment", "UID": "all_n_presupposition", "pairID": "9e", "paradigmID": 0 } ``` and the raw implicature data from any sub-dataset: ```buildoutcfg { "sentence1": "That teenager couldn't yell.", "sentence2": "That teenager could yell.", "gold_label_log": "contradiction", "gold_label_prag": "contradiction", "spec_relation": "negation", "item_type": "control", "trigger": "modal", "lexemes": "can - have to" } ``` ### Data Fields **Presupposition** There is a slight mapping from the raw data fields in the presupposition sub-datasets and the fields appearing in the HuggingFace Datasets. When dealing with the HF Dataset, the following mapping of fields happens: ```buildoutcfg "premise" -> "sentence1" "hypothesis"-> "sentence2" "trigger" -> "trigger" or "Not_In_Example" "trigger1" -> "trigger1" or "Not_In_Example" "trigger2" -> "trigger2" or "Not_In_Example" "presupposition" -> "presupposition" or "Not_In_Example" "gold_label" -> "gold_label" "UID" -> "UID" "pairID" -> "pairID" "paradigmID" -> "paradigmID" ``` For the most part, the majority of the raw fields remain unchanged. However, when it comes to the various `trigger` fields, a new mapping was introduced. There are some examples in the dataset that only have the `trigger` field while other examples have the `trigger1` and `trigger2` field without the `trigger` or `presupposition` field. Nominally, most examples look like the example in the Data Instances section above. Occassionally, however, some examples will look like: ```buildoutcfg { 'sentence1': 'Did that committee know when Lissa walked through the cafe?', 'sentence2': 'That committee knew when Lissa walked through the cafe.', 'trigger1': 'interrogative', 'trigger2': 'unembedded', 'gold_label': 'neutral', 'control_item': True, 'UID': 'question_presupposition', 'pairID': '1821n', 'paradigmID': 95 } ``` In this example, `trigger1` and `trigger2` appear and `presupposition` and `trigger` are removed. This maintains the length of the dictionary. To account for these examples, we have thus introduced the mapping above such that all examples accessed through the HF Datasets interface will have the same size as well as the same fields. In the event that an example does not have a value for one of the fields, the field is maintained in the dictionary but given a value of `Not_In_Example`. To illustrate this point, the example given in the Data Instances section above would look like the following in the HF Datasets: ```buildoutcfg { "premise": "All ten guys that proved to boast might have been divorcing.", "hypothesis": "There are exactly ten guys that proved to boast.", "trigger": "modal", "trigger1": "Not_In_Example", "trigger2": "Not_In_Example" "presupposition": "positive", "gold_label": "entailment", "UID": "all_n_presupposition", "pairID": "9e", "paradigmID": 0 } ``` Below is description of the fields: ```buildoutcfg "premise": The premise. "hypothesis": The hypothesis. "trigger": A detailed discussion of trigger types appears in the paper. "trigger1": A detailed discussion of trigger types appears in the paper. "trigger2": A detailed discussion of trigger types appears in the paper. "presupposition": positive or negative. "gold_label": Corresponds to entailment, contradiction, or neutral. "UID": Unique id. "pairID": Sentence pair ID. "paradigmID": ? ``` It is not immediately clear what the difference is between `trigger`, `trigger1`, and `trigger2` is or what the `paradigmID` refers to. **Implicature** The `implicature` fields only have the mapping below: ```buildoutcfg "premise" -> "sentence1" "hypothesis"-> "sentence2" ``` Here is a description of the fields: ```buildoutcfg "premise": The premise. "hypothesis": The hypothesis. "gold_label_log": Gold label for a logical reading of the sentence pair. "gold_label_prag": Gold label for a pragmatic reading of the sentence pair. "spec_relation": ? "item_type": ? "trigger": A detailed discussion of trigger types appears in the paper. "lexemes": ? ``` ### Data Splits As the dataset was created to test already trained models, the only split that exists is for testing. ## Dataset Creation ### Curation Rationale IMPPRES was created to evaluate how well trained NLI models recognize several classes of presuppositions and scalar implicatures. ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? The annotations were generated semi-automatically. ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information IMPPRES is available under a Creative Commons Attribution-NonCommercial 4.0 International Public License ("The License"). You may not use these files except in compliance with the License. Please see the LICENSE file for more information before you use the dataset. ### Citation Information ```buildoutcfg @inproceedings{jeretic-etal-2020-natural, title = "Are Natural Language Inference Models {IMPPRESsive}? {L}earning {IMPlicature} and {PRESupposition}", author = "Jereti\v{c}, Paloma and Warstadt, Alex and Bhooshan, Suvrat and Williams, Adina", booktitle = "Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics", month = jul, year = "2020", address = "Online", publisher = "Association for Computational Linguistics", url = "https://www.aclweb.org/anthology/2020.acl-main.768", doi = "10.18653/v1/2020.acl-main.768", pages = "8690--8705", abstract = "Natural language inference (NLI) is an increasingly important task for natural language understanding, which requires one to infer whether a sentence entails another. However, the ability of NLI models to make pragmatic inferences remains understudied. We create an IMPlicature and PRESupposition diagnostic dataset (IMPPRES), consisting of 32K semi-automatically generated sentence pairs illustrating well-studied pragmatic inference types. We use IMPPRES to evaluate whether BERT, InferSent, and BOW NLI models trained on MultiNLI (Williams et al., 2018) learn to make pragmatic inferences. Although MultiNLI appears to contain very few pairs illustrating these inference types, we find that BERT learns to draw pragmatic inferences. It reliably treats scalar implicatures triggered by {``}some{''} as entailments. For some presupposition triggers like {``}only{''}, BERT reliably recognizes the presupposition as an entailment, even when the trigger is embedded under an entailment canceling operator like negation. BOW and InferSent show weaker evidence of pragmatic reasoning. We conclude that NLI training encourages models to learn some, but not all, pragmatic inferences.", } ``` ### Contributions Thanks to [@aclifton314](https://github.com/aclifton314) for adding this dataset.
inquisitive_qg
2022-11-18T20:09:50.000Z
[ "task_categories:text2text-generation", "annotations_creators:crowdsourced", "language_creators:crowdsourced", "multilinguality:monolingual", "size_categories:10K<n<100K", "source_datasets:original", "language:en", "license:unknown", "question-generation", "region:us" ]
null
A dataset of about 20k questions that are elicited from readers as they naturally read through a document sentence by sentence. Compared to existing datasets, INQUISITIVE questions target more towards high-level (semantic and discourse) comprehension of text. Because these questions are generated while the readers are processing the information, the questions directly communicate gaps between the reader’s and writer’s knowledge about the events described in the text, and are not necessarily answered in the document itself. This type of question reflects a real-world scenario: if one has questions during reading, some of them are answered by the text later on, the rest are not, but any of them would help further the reader’s understanding at the particular point when they asked it. This resource could enable question generation models to simulate human-like curiosity and cognitive processing, which may open up a new realm of applications.
@InProceedings{ko2020inquisitive, author = {Ko, Wei-Jen and Chen, Te-Yuan and Huang, Yiyan and Durrett, Greg and Li, Junyi Jessy}, title = {Inquisitive Question Generation for High Level Text Comprehension}, booktitle = {Proceedings of EMNLP}, year = {2020}, }
null
1
3
--- pretty_name: InquisitiveQg annotations_creators: - crowdsourced language_creators: - crowdsourced language: - en license: - unknown multilinguality: - monolingual size_categories: - 10K<n<100K source_datasets: - original task_categories: - text2text-generation task_ids: [] paperswithcode_id: inquisitive tags: - question-generation dataset_info: features: - name: id dtype: int32 - name: article_id dtype: int32 - name: article dtype: string - name: sentence_id dtype: int32 - name: sentence dtype: string - name: span dtype: string - name: question dtype: string - name: span_start_position dtype: int32 - name: span_end_position dtype: int32 config_name: plain_text splits: - name: train num_bytes: 66099232 num_examples: 15931 - name: validation num_bytes: 8904329 num_examples: 1991 - name: test num_bytes: 7167203 num_examples: 1894 download_size: 7085941 dataset_size: 82170764 --- # Dataset Card for InquisitiveQg ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [Add homepage URL here if available (unless it's a GitHub repository)]() - **Repository:** [If the dataset is hosted on github or has a github homepage, add URL here]() - **Paper:** [If the dataset was introduced by a paper or there was a paper written describing the dataset, add URL here (landing page for Arxiv paper preferred)]() - **Leaderboard:** [If the dataset supports an active leaderboard, add link here]() - **Point of Contact:** [If known, name and email of at least one person the reader can contact for questions about the dataset.]() ### Dataset Summary [More Information Needed] ### Supported Tasks and Leaderboards [More Information Needed] ### Languages [More Information Needed] ## Dataset Structure ### Data Instances [More Information Needed] ### Data Fields [More Information Needed] ### Data Splits [More Information Needed] ## Dataset Creation ### Curation Rationale [More Information Needed] ### Source Data [More Information Needed] #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? [More Information Needed] ### Annotations [More Information Needed] #### Annotation process [More Information Needed] #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators [More Information Needed] ### Licensing Information [More Information Needed] ### Citation Information [More Information Needed] ### Contributions Thanks to [@patil-suraj](https://github.com/patil-suraj) for adding this dataset.
isizulu_ner_corpus
2023-01-25T14:33:13.000Z
[ "task_categories:token-classification", "task_ids:named-entity-recognition", "annotations_creators:expert-generated", "language_creators:expert-generated", "multilinguality:monolingual", "size_categories:10K<n<100K", "source_datasets:original", "language:zu", "license:other", "region:us" ]
null
Named entity annotated data from the NCHLT Text Resource Development: Phase II Project, annotated with PERSON, LOCATION, ORGANISATION and MISCELLANEOUS tags.
@inproceedings{isizulu_ner_corpus, author = {A.N. Manzini and Roald Eiselen}, title = {NCHLT isiZulu Named Entity Annotated Corpus}, booktitle = {Eiselen, R. 2016. Government domain named entity recognition for South African languages. Proceedings of the 10th Language Resource and Evaluation Conference, Portorož, Slovenia.}, year = {2016}, url = {https://repo.sadilar.org/handle/20.500.12185/319}, }
null
0
3
--- annotations_creators: - expert-generated language_creators: - expert-generated language: - zu license: - other multilinguality: - monolingual size_categories: - 10K<n<100K source_datasets: - original task_categories: - token-classification task_ids: - named-entity-recognition pretty_name: Isizulu Ner Corpus license_details: Creative Commons Attribution 2.5 South Africa dataset_info: features: - name: id dtype: string - name: tokens sequence: string - name: ner_tags sequence: class_label: names: '0': OUT '1': B-PERS '2': I-PERS '3': B-ORG '4': I-ORG '5': B-LOC '6': I-LOC '7': B-MISC '8': I-MISC config_name: isizulu_ner_corpus splits: - name: train num_bytes: 4038876 num_examples: 10956 download_size: 25097584 dataset_size: 4038876 --- # Dataset Card for Isizulu Ner Corpus ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** [Isizulu Ner Corpus Homepage](https://repo.sadilar.org/handle/20.500.12185/319) - **Repository:** - **Paper:** - **Leaderboard:** - **Point of Contact:** [Martin Puttkammer](mailto:Martin.Puttkammer@nwu.ac.za) ### Dataset Summary The isizulu Ner Corpus is a Zulu dataset developed by [The Centre for Text Technology (CTexT), North-West University, South Africa](http://humanities.nwu.ac.za/ctext). The data is based on documents from the South African goverment domain and crawled from gov.za websites. It was created to support NER task for Zulu language. The dataset uses CoNLL shared task annotation standards. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages The language supported is Zulu. ## Dataset Structure ### Data Instances A data point consists of sentences seperated by empty line and tab-seperated tokens and tags. {'id': '0', 'ner_tags': [7, 8, 0, 0, 0], 'tokens': ['Lesi', 'sigaba', 'se-website', ',', 'esikhonjiswe'] } ### Data Fields - `id`: id of the sample - `tokens`: the tokens of the example text - `ner_tags`: the NER tags of each token The NER tags correspond to this list: ``` "OUT", "B-PERS", "I-PERS", "B-ORG", "I-ORG", "B-LOC", "I-LOC", "B-MISC", "I-MISC", ``` The NER tags have the same format as in the CoNLL shared task: a B denotes the first item of a phrase and an I any non-initial word. There are four types of phrases: person names (PER), organizations (ORG), locations (LOC) and miscellaneous names (MISC). (OUT) is used for tokens not considered part of any named entity. ### Data Splits The data was not split. ## Dataset Creation ### Curation Rationale The data was created to help introduce resources to new language - zulu. [More Information Needed] ### Source Data #### Initial Data Collection and Normalization The data is based on South African government domain and was crawled from gov.za websites. #### Who are the source language producers? The data was produced by writers of South African government websites - gov.za ### Annotations #### Annotation process [More Information Needed] #### Who are the annotators? The data was annotated during the NCHLT text resource development project. [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations [More Information Needed] ## Additional Information ### Dataset Curators The annotated data sets were developed by the Centre for Text Technology (CTexT, North-West University, South Africa). See: [more information](http://www.nwu.ac.za/ctext) ### Licensing Information The data is under the [Creative Commons Attribution 2.5 South Africa License](http://creativecommons.org/licenses/by/2.5/za/legalcode) ### Citation Information ``` @inproceedings{isizulu_ner_corpus, author = {A.N. Manzini and Roald Eiselen}, title = {NCHLT isiZulu Named Entity Annotated Corpus}, booktitle = {Eiselen, R. 2016. Government domain named entity recognition for South African languages. Proceedings of the 10th Language Resource and Evaluation Conference, Portorož, Slovenia.}, year = {2016}, url = {https://repo.sadilar.org/handle/20.500.12185/319}, } ``` ### Contributions Thanks to [@yvonnegitau](https://github.com/yvonnegitau) for adding this dataset.
makhzan
2022-11-03T16:07:47.000Z
[ "task_categories:text-generation", "task_categories:fill-mask", "task_ids:language-modeling", "task_ids:masked-language-modeling", "annotations_creators:expert-generated", "language_creators:expert-generated", "multilinguality:monolingual", "size_categories:1K<n<10K", "source_datasets:original", "...
null
An Urdu text corpus for machine learning, natural language processing and linguistic analysis.
null
null
0
3
--- annotations_creators: - expert-generated language_creators: - expert-generated language: - ur license: - other multilinguality: - monolingual size_categories: - 1K<n<10K source_datasets: - original task_categories: - text-generation - fill-mask task_ids: - language-modeling - masked-language-modeling paperswithcode_id: null pretty_name: makhzan dataset_info: features: - name: file_id dtype: string - name: metadata dtype: string - name: title dtype: string - name: num-words dtype: int64 - name: contains-non-urdu-languages dtype: string - name: document_body dtype: string splits: - name: train num_bytes: 35637310 num_examples: 5522 download_size: 15187763 dataset_size: 35637310 --- # Dataset Card for makhzan ## Table of Contents - [Dataset Description](#dataset-description) - [Dataset Summary](#dataset-summary) - [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards) - [Languages](#languages) - [Dataset Structure](#dataset-structure) - [Data Instances](#data-instances) - [Data Fields](#data-fields) - [Data Splits](#data-splits) - [Dataset Creation](#dataset-creation) - [Curation Rationale](#curation-rationale) - [Source Data](#source-data) - [Annotations](#annotations) - [Personal and Sensitive Information](#personal-and-sensitive-information) - [Considerations for Using the Data](#considerations-for-using-the-data) - [Social Impact of Dataset](#social-impact-of-dataset) - [Discussion of Biases](#discussion-of-biases) - [Other Known Limitations](#other-known-limitations) - [Additional Information](#additional-information) - [Dataset Curators](#dataset-curators) - [Licensing Information](#licensing-information) - [Citation Information](#citation-information) - [Contributions](#contributions) ## Dataset Description - **Homepage:** https://matnsaz.net/en/makhzan - **Repository:** https://github.com/zeerakahmed/makhzan - **Paper:** [More Information Needed] - **Leaderboard:** [More Information Needed] - **Point of Contact:** Zeerak Ahmed ### Dataset Summary An Urdu text corpus for machine learning, natural language processing and linguistic analysis. ### Supported Tasks and Leaderboards [More Information Needed] ### Languages ur ## Dataset Structure ### Data Instances ``` { "contains-non-urdu-languages": "No", "document_body": " <body> <section> <p>بنگلہ دیش کی عدالتِ عالیہ نے طلاق کے ایک مقدمے کا فیصلہ کرتے ہوئے علما کے فتووں کو غیر قانونی قرار دیا ہے۔ عدالت نے پارلیمنٹ سے یہ درخواست کی ہے کہ وہ جلد ایسا قانون وضع کرے کہ جس کے بعد فتویٰ بازی قابلِ دست اندازیِ پولیس جرم بن جائے۔ بنگلہ دیش کے علما نے اس فیصلے پر بھر پور ردِ عمل ظاہرکرتے ہوئے اس کے خلاف ملک گیر تحریک چلانے کا اعلان کیا ہے۔ اس ضمن میں علما کی ایک تنظیم ”اسلامک یونٹی الائنس“ نے متعلقہ ججوں کو مرتد یعنی دین سے منحرف اور دائرۂ اسلام سے خارج قرار دیا ہے۔</p> <p>فتوے کا لفظ دو موقعوں پر استعمال ہوتا ہے۔ ایک اس موقع پر جب کوئی صاحبِ علم شریعت کے کسی مئلے کے بارے میں اپنی رائے پیش کرتا ہے۔ دوسرے اس موقع پر جب کوئی عالمِ دین کسی خاص واقعے کے حوالے سے اپنا قانونی فیصلہ صادر کرتا ہے۔ ایک عرصے سے ہمارے علما کے ہاں اس دوسرے موقعِ استعمال کا غلبہ ہو گیا ہے۔ اس کا نتیجہ یہ نکلا ہے کہ اس لفظ کا رائے یا نقطۂ نظر کے مفہوم میں استعمال کم و بیش متروک ہو گیا ہے۔ چنانچہ اب فتوے کا مطلب ہی علما کی طرف سے کسی خاص مألے یا واقعے کے بارے میں حتمی فیصلے کا صدور سمجھا جاتا ہے۔ علما اسی حیثیت سے فتویٰ دیتے ہیں اور عوام الناس اسی اعتبار سے اسے قبول کرتے ہیں۔ اس صورتِ حال میں ہمارے نزدیک، چند مسائل پیدا ہوتے ہیں۔ اس سے پہلے کہ ہم مذکورہ فیصلے کے بارے میں اپنا تاثر بیان کریں، یہ ضروری معلوم ہوتا ہے کہ مختصر طور پر ان مسائل کا جائزہ لے لیا جائے۔</p> <p>پہلا مألہ یہ پیدا ہوتا ہے کہ قانون سازی اور شرعی فیصلوں کا اختیار ایسے لوگوں کے ہاتھ میں آجاتا ہے جو قانون کی رو سے اس کے مجاز ہی نہیں ہوتے۔ کسی میاں بیوی کے مابین طلاق کے مألے میں کیا طلاق واقع ہوئی ہے یا نہیں ہوئی؟ ان کا نکاح قائم ہے یا باطل ہو گیا ہے؟ رمضان یا عید کا چاند نظر آیا ہے یا نہیں آیا؟کوئی مسلمان اپنے کسی قول یا اقدام کی وجہ سے کہیں دائرۂ اسلام سے خارج اورنتیجۃً مسلم شہریت کے قانونی حقوق سے محروم تو نہیں ہو گیا؟ یہ اور اس نوعیت کے بہت سے دوسرے معاملات سر تا سر قانون اور عدالت سے متعلق ہوتے ہیں۔ علما کی فتویٰ سازی کے نتیجے میںیہ امور گویا حکومت اورعدلیہ کے ہاتھ سے نکل کر غیر متعلق افراد کے ہاتھوں میں آجاتے ہیں۔</p> <p>دوسرا مألہ یہ پیدا ہوتا ہے کہ قانون کی حاکمیت کا تصور مجروح ہوتا ہے اور لوگوں میں قانون سے روگردانی کے رجحانات کو تقویت ملتی ہے۔ اس کی وجہ یہ ہے کہ قانون اپنی روح میں نفاذ کا متقاضی ہوتا ہے۔ اگر اسے نفاذ سے محروم رکھا جائے تو اس کی حیثیت محض رائے اور نقطۂ نظر کی سی ہوتی ہے۔ غیر مجاز فرد سے صادر ہونے والا فتویٰ یا قانون حکومت کی قوتِ نافذہ سے محروم ہوتا ہے۔ اس کی خلاف ورزی پر کسی قسم کی سزا کا خوف نہیں ہوتا۔ چنانچہ فتویٰ اگر مخاطب کی پسند کے مطابق نہ ہو تو اکثر وہ اسے ماننے سے انکار کر دیتا ہے۔ اس طرح وہ فتویٰ یا قانون بے توقیر ہوتا ہے۔ ایسے ماحول میں رہنے والے شہریوں میں قانون ناپسندی کا رجحان فروغ پاتا ہے اور جیسے ہی انھیں موقع ملتا ہے وہ بے دریغ قانون کی خلاف ورزی کر ڈالتے ہیں۔</p> <p>تیسرامسئلہ یہ پیدا ہوتا ہے کہ اگرغیر مجاز افراد سے صادر ہونے والے فیصلوں کو نافذ کرنے کی کوشش کی جائے تو ملک میں بد نظمی اور انارکی کا شدید اندیشہ پیدا ہو جاتا ہے۔ جب غیر مجازافراد سے صادر ہونے والے قانونی فیصلوں کو حکومتی سرپرستی کے بغیر نافذ کرنے کی کوشش کی جاتی ہے تو اپنے عمل سے یہ اس بات کا اعلان ہوتا ہے کہ مرجعِ قانون و اقتدارتبدیل ہو چکا ہے۔ جب کوئی عالمِ دین مثال کے طور پر، یہ فتویٰ صادر کرتا ہے کہ سینما گھروں اور ٹی وی اسٹیشنوں کو مسمار کرنامسلمانوں کی ذمہ داری ہے، یا کسی خاص قوم کے خلاف جہاد فرض ہو چکا ہے، یا فلاں کی دی گئی طلاق واقع ہو گئی ہے اور فلاں کی نہیں ہوئی، یا فلاں شخص یا گروہ اپنا اسلامی تشخص کھو بیٹھا ہے تو وہ درحقیقت قانونی فیصلہ جاری کر رہا ہوتا ہے۔ دوسرے الفاظ میں، وہ ریاست کے اندر اپنی ایک الگ ریاست بنانے کا اعلان کر رہا ہوتا ہے۔ اس کا نتیجہ سوائے انتشار اور انارکی کے اور کچھ نہیں نکلتا۔ یہی وجہ ہے کہ جن علاقوں میں حکومت کی گرفت کمزور ہوتی ہے وہاں اس طرح کے فیصلوں کا نفاذ بھی ہو جاتا ہے اور حکومت منہ دیکھتی رہتی ہے۔</p> <p>چوتھا مسئلہ یہ پیدا ہوتا ہے کہ مختلف مذہبی مسالک کی وجہ سے ایک ہی معاملے میں مختلف اور متضاد فتوے منظرِ عام پر آتے ہیں۔ یہ تو ہمارے روز مرہ کی بات ہے کہ ایک ہی گروہ کو بعض علماے دین کافر قرار دیتے ہیں اور بعض مسلمان سمجھتے ہیں۔ کسی شخص کے منہ سے اگر ایک موقع پر طلاق کے الفاظ تین بار نکلتے ہیں تو بعض علما اس پر ایک طلاق کا حکم لگا کر رجوع کا حق باقی رکھتے ہیں اور بعض تین قرار دے کررجوع کو باطل قرار دیتے ہیں۔ یہ صورتِ حال ایک عام آدمی کے لیے نہایت دشواریاں پیدا کر دیتی ہے۔</p> <p>پانچواں مسئلہ یہ پیدا ہوتا ہے کہ حکمران اگر دین و شریعت سے کچھ خاص دلچسپی نہ رکھتے ہوں تو وہ اس صورتِ حال میں شریعت کی روشنی میں قانون سازی کی طرف متوجہ نہیں ہوتے۔ کام چل رہا ہے کے اصول پر وہ اس طریقِ قانون سازی سے سمجھوتاکیے رہتے ہیں۔ اس کا نتیجہ یہ نکلتا ہے کہ حکومتی ادارے ضروری قانون سازی کے بارے میں بے پروائی کا رویہ اختیار کرتے ہیں اور قوانین اپنے فطری ارتقا سے محروم رہتے ہیں۔</p> <p>چھٹا مألہ یہ پیدا ہوتا ہے کہ رائج الوقت قانون اور عدالتوں کی توہین کے امکانات پیدا ہو جاتے ہیں۔ جب کسی مسئلے میں عدالتیں اپنا فیصلہ سنائیں اور علما اسے باطل قرار دیتے ہوئے اس کے برعکس اپنا فیصلہ صادر کریں تو اس سے عدالتوں کا وقار مجروح ہوتا ہے۔ اس کا مطلب یہ ہوتا ہے کہ کوئی شہری عدلیہ کو چیلنج کرنے کے لیے کھڑا ہو گیا ہے۔</p> <p>ان مسائل کے تناظر میں بنگلہ دیش کی عدالتِ عالیہ کا فیصلہ ہمارے نزدیک، امت کی تاریخ میں ایک عظیم فیصلہ ہے۔ جناب جاوید احمد صاحب غامدی نے اسے بجا طور پر صدی کا بہترین فیصلہ قرار دیا ہے۔ بنگلہ دیش کی عدالت اگر علما کے فتووں اور قانونی فیصلوں پر پابندی لگانے کے بجائے، ان کے اظہارِ رائے پر پابندی عائدکرتی تو ہم اسے صدی کا بدترین فیصلہ قرار دیتے اور انھی صفحات میں بے خوفِ لومۃ و لائم اس پر نقد کر رہے ہوتے۔</p> <p>موجودہ زمانے میں امتِ مسلمہ کا ایک بڑا المیہ یہ ہے کہ اس کے علما اپنی اصل ذمہ داری کو ادا کرنے کے بجائے ان ذمہ داریوں کو ادا کرنے پر مصر ہیں جن کے نہ وہ مکلف ہیں اور نہ اہل ہیں۔ قرآن و سنت کی رو سے علما کی اصل ذمہ داری دعوت و تبلیغ، انذار و تبشیر اور تعلیم و تحقیق ہے۔ ان کا کام سیاست نہیں، بلکہ سیاست دانوں کو دین کی رہنمائی سے آگاہی ہے؛ ان کا کام حکومت نہیں، بلکہ حکمرانوں کی اصلاح کی کوشش ہے؛ ان کا کام جہاد و قتال نہیں، بلکہ جہادکی تعلیم اور جذبۂ جہاد کی بیداری ہے؛ اسی طرح ان کا کام قانون سازی اور فتویٰ بازی نہیں بلکہ تحقیق و اجتہاد ہے۔ گویا انھیں قرآنِ مجیدکامفہوم سمجھنے، سنتِ ثابتہ کا مدعا متعین کرنے اور قولِ پیغمبر کا منشامعلوم کرنے کے لیے تحقیق کرنی ہے اور جن امور میں قرآن و سنت خاموش ہیں ان میں اپنی عقل و بصیرت سے اجتہادی آراقائم کرنی ہیں۔ ان کی کسی تحقیق یا اجتہاد کو جب عدلیہ یا پارلیمنٹ قبول کرے گی تو وہ قانون قرار پائے گا۔ اس سے پہلے اس کی حیثیت محض ایک رائے کی ہوگی۔ اس لیے اسے اسی حیثیت سے پیش کیا جائے گا۔</p> <p>اس کا مطلب یہ ہے کہ کوئی حکم نہیں لگایا جائے گا، کوئی فیصلہ نہیں سنایا جائے گا، کوئی فتویٰ نہیں دیا جائے گا، بلکہ طالبِ علمانہ لب و لہجے میں محض علم و استدلال کی بنا پر اپنا نقطۂ نظر پیش کیا جائے گا۔ یہ نہیں کہا جائے گا کہ فلاں شخص کافر ہے، بلکہ اس کی اگر ضرورت پیش آئے تو یہ کہا جائے گا کہ فلاں شخص کا فلاں عقیدہ کفر ہے۔ یہ نہیں کہا جائے گا کہ فلاں آدمی دائرۂ اسلام سے خارج ہو گیا ہے، بلکہ یہ کہا جائے گا کہ فلاں آدمی کا فلاں نقطۂ نظر اسلام کے دائرے میں نہیں آتا۔ یہ نہیں کہا جائے گا فلاں آدمی مشرک ہے، بلکہ یہ کہا جائے گا فلاں نظریہ یا فلاں طرزِ عمل شرک ہے۔ یہ نہیں کہا جائے گا کہ زید کی طرف سے دی گئی ایک وقت کی تین طلاقیں واقع ہو گئی ہیں، بلکہ یہ کہا جائے گا کہ ایک وقت کی تین طلاقیں واقع ہو نی چاہییں۔</p> <p>حکم لگانا، فیصلہ سنانا، قانون وضع کرنا اورفتویٰ جاری کرنا درحقیقت، عدلیہ اور حکومت کا کام ہے کسی عالمِ دین یا کسی اور غیر مجاز فرد کی طرف سے اس کام کو انجام دینے کی کوشش سراسر تجاوز ہے۔ خلافتِ راشدہ کے زمانے میں اس اصول کو ہمیشہ ملحوظ رکھا گیا۔ شاہ ولی اللہ محدث دہلوی اپنی کتاب ”ازالتہ الخفا ء“ میں لکھتے ہیں:</p> <blockquote> <p>”اس زمانے تک وعظ اور فتویٰ خلیفہ کی رائے پر موقوف تھا۔ خلیفہ کے حکم کے بغیر نہ وعظ کہتے تھے اور نہ فتویٰ دیتے تھے۔ بعد میں خلیفہ کے حکم کے بغیر وعظ کہنے اور فتویٰ دینے لگے اور فتویٰ کے معاملے میں جماعت (مجلسِ شوریٰ) کے مشورہ کی جو صورت پہلے تھی وہ باقی نہ رہی——- (اس زمانے میں) جب کوئی اختلافی صورت نمودار ہوتی، خلیفہ کے سامنے معاملہ پیش کرتے، خلیفہ اہلِ علم و تقویٰ سے مشورہ کرنے کے بعد ایک رائے قائم کرتا اور وہی سب لوگوں کی رائے بن جاتی۔ حضرت عثمان کی شہادت کے بعد ہر عالم بطورِ خود فتویٰ دینے لگا اور اس طرح مسلمانوں میں اختلاف برپا ہوا۔“ (بحوالہ ”اسلامی ریاست میں فقہی اختلافات کا حل“، مولاناامین احسن اصلاحی، ص۳۲)</p> </blockquote> </section> </body> ", "file_id": "0001.xml", "metadata": " <meta> <title>بنگلہ دیش کی عدالت کا تاریخی فیصلہ</title> <author> <name>سید منظور الحسن</name> <gender>Male</gender> </author> <publication> <name>Mahnama Ishraq February 2001</name> <year>2001</year> <city>Lahore</city> <link>https://www.javedahmedghamidi.org/#!/ishraq/5adb7341b7dd1138372db999?articleId=5adb7452b7dd1138372dd6fb&amp;year=2001&amp;decade=2000</link> <copyright-holder>Al-Mawrid</copyright-holder> </publication> <num-words>1694</num-words> <contains-non-urdu-languages>No</contains-non-urdu-languages> </meta> ", "num-words": 1694, "title": "بنگلہ دیش کی عدالت کا تاریخی فیصلہ" } ``` ### Data Fields ```file_id (str)```: Document file_id corresponding to filename in repository. ```metadata(str)```: XML formatted string containing metadata on the document such as the document's title, information about the author and publication, as well as other potentially useful facts such as the number of Urdu words in the document and whether the document contains text in any other languages. ```title (str)```: Title of the document. ```num-words (int)```: Number of words in document. ```contains-non-urdu-languages (str)```: ```Yes``` if document contains words other than urdu, ```No``` otherwise. ```document_body```: XML formatted body of the document. Details below: The document is divided into ```<section>``` elements. In general the rule is that a clear visual demarkation in the original text (such as a page break, or a horizontal rule) is used to indicate a section break. A heading does not automatically create a new section. Each paragraph is a ```<p>``` element. Headings are wrapped in an ```<heading>``` element. Blockquotes are wrapped in a ```<blockquote>``` element. Blockquotes may themselves contain other elements. Lists are wrapped in an ```<list>```. Individual items in each list are wrapped in an ```<li>``` element. Poetic verses are wrapped in a ```<verse>``` element. Each verse is on a separate line but is not wrapped in an individual element. Tables are wrapped in a ```<table>``` element. A table is divided into rows marked by ```<tr>``` and columns marked by ```<td>```. Text not in the Urdu language is wrapped in an ```<annotation>``` tag (more below). ```<p>, <heading>, <li>, <td>``` and ```<annotation>``` tags are inline with the text (i.e. there is no new line character before and after the tag). Other tags have a new line after the opening and before the closing tag. Due to the use of XML syntax, ```<```, ```>``` and ```&``` characters have been escaped as ```&lt```;, ```&gt```;, and ```&amp```; respectively. This includes the use of these characters in URLs inside metadata. ### Data Splits All the data is in one split ```train``` ## Dataset Creation ### Curation Rationale All text in this repository has been selected for quality of language, upholding high editorial standards. Given the poor quality of most published Urdu text in digital form, this selection criteria allows the use of this text for natural language processing, and machine learning applications without the need to address fundamental quality issues in the text. We have made efforts to ensure this text is as broadly representative as possible. Specifically we have attempted to select for as many authors as possible, and diversity in the gender of the author, as well as years and city of publication. This effort is imperfect, and we appreciate any attempts at pointing us to resources that can help diversify this text further. ### Source Data #### Initial Data Collection and Normalization [More Information Needed] #### Who are the source language producers? Makhzan has been started with generous initial donations of text from two renowned journals  Bunyad, from the Gurmani Center of Literature and Languages at the Lahore University of Management Sciences (LUMS), and Ishraq, from the Al-Mawrid Institute. This choice of sources allowed us to get a diversity of voices even in a small initial corpus, while ensuring the highest editorial standards available in published Urdu text. As a result your models can also maintain high linguistic standards. ### Annotations #### Annotation process Text is structured and annotated using XML syntax. The ontology of elements used is loosely based around HTML, with simplifications made when HTML's specificity is not needed, and additions made to express common occurences in this corpus that would be useful for linguistic analysis. The semantic tagging of text is editorial in nature, which is to say that another person semantically tagging the text may do so differently. Effort has been made however to ensure consistency, and to retain the original meaning of the text while making it easy to parse through linguistically different pieces of text for analysis. Annotations have been made inline using an ```<annotation>``` element. A language (lang) attribute is added to the ```<annotation>``` element to indicate text in other languages (such as quoted text or technical vocabulary presented in other languages and scripts). The attribute value a two-character ISO 639-1 code. So the resultant annotation for an Arabic quote for example, will be ```<annotation lang="ar"></annotation>```. A type (type) attributed is added to indicate text that is not in a language per se but is not Urdu text. URLs for example are wrapped in an ```<annotation type="url">``` tag. #### Who are the annotators? [More Information Needed] ### Personal and Sensitive Information [More Information Needed] ## Considerations for Using the Data ### Social Impact of Dataset [More Information Needed] ### Discussion of Biases [More Information Needed] ### Other Known Limitations A few of the files do not have valid XML and cannot be loaded. This issue is tracked [here](https://github.com/zeerakahmed/makhzan/issues/28) ## Additional Information ### Dataset Curators Zeerak Ahmed ### Licensing Information [More Information Needed] ### Citation Information ``` @misc{makhzan, title={Maḵẖzan}, howpublished = "\url{https://github.com/zeerakahmed/makhzan/}", } ``` ### Contributions Thanks to [@arkhalid](https://github.com/arkhalid) for adding this dataset.