Datasets:
Commit
·
bf70c74
1
Parent(s):
7d5b0ab
Added initial files
Browse files- .gitattributes +1 -0
- README.md +190 -0
- data/squad_bn.tar.bz2 +3 -0
- squad_bn.py +120 -0
.gitattributes
CHANGED
|
@@ -36,3 +36,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
|
|
| 36 |
*.mp3 filter=lfs diff=lfs merge=lfs -text
|
| 37 |
*.ogg filter=lfs diff=lfs merge=lfs -text
|
| 38 |
*.wav filter=lfs diff=lfs merge=lfs -text
|
|
|
|
|
|
| 36 |
*.mp3 filter=lfs diff=lfs merge=lfs -text
|
| 37 |
*.ogg filter=lfs diff=lfs merge=lfs -text
|
| 38 |
*.wav filter=lfs diff=lfs merge=lfs -text
|
| 39 |
+
data/squad_bn.tar.bz2 filter=lfs diff=lfs merge=lfs -text
|
README.md
ADDED
|
@@ -0,0 +1,190 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
annotations_creators:
|
| 3 |
+
- machine-generated
|
| 4 |
+
language_creators:
|
| 5 |
+
- found
|
| 6 |
+
multilinguality:
|
| 7 |
+
- monolingual
|
| 8 |
+
size_categories:
|
| 9 |
+
- 100K<n<1M
|
| 10 |
+
source_datasets:
|
| 11 |
+
- extended
|
| 12 |
+
task_categories:
|
| 13 |
+
- text-classification
|
| 14 |
+
task_ids:
|
| 15 |
+
- natural-language-inference
|
| 16 |
+
languages:
|
| 17 |
+
- bn
|
| 18 |
+
licenses:
|
| 19 |
+
- cc-by-nc-sa-4.0
|
| 20 |
+
---
|
| 21 |
+
|
| 22 |
+
# Dataset Card for `xnli_bn`
|
| 23 |
+
|
| 24 |
+
## Table of Contents
|
| 25 |
+
- [Dataset Card for `xnli_bn`](#dataset-card-for-xnli_bn)
|
| 26 |
+
- [Table of Contents](#table-of-contents)
|
| 27 |
+
- [Dataset Description](#dataset-description)
|
| 28 |
+
- [Dataset Summary](#dataset-summary)
|
| 29 |
+
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
|
| 30 |
+
- [Languages](#languages)
|
| 31 |
+
- [Usage](#usage)
|
| 32 |
+
- [Dataset Structure](#dataset-structure)
|
| 33 |
+
- [Data Instances](#data-instances)
|
| 34 |
+
- [Data Fields](#data-fields)
|
| 35 |
+
- [Data Splits](#data-splits)
|
| 36 |
+
- [Dataset Creation](#dataset-creation)
|
| 37 |
+
- [Curation Rationale](#curation-rationale)
|
| 38 |
+
- [Source Data](#source-data)
|
| 39 |
+
- [Initial Data Collection and Normalization](#initial-data-collection-and-normalization)
|
| 40 |
+
- [Who are the source language producers?](#who-are-the-source-language-producers)
|
| 41 |
+
- [Annotations](#annotations)
|
| 42 |
+
- [Annotation process](#annotation-process)
|
| 43 |
+
- [Who are the annotators?](#who-are-the-annotators)
|
| 44 |
+
- [Personal and Sensitive Information](#personal-and-sensitive-information)
|
| 45 |
+
- [Considerations for Using the Data](#considerations-for-using-the-data)
|
| 46 |
+
- [Social Impact of Dataset](#social-impact-of-dataset)
|
| 47 |
+
- [Discussion of Biases](#discussion-of-biases)
|
| 48 |
+
- [Other Known Limitations](#other-known-limitations)
|
| 49 |
+
- [Additional Information](#additional-information)
|
| 50 |
+
- [Dataset Curators](#dataset-curators)
|
| 51 |
+
- [Licensing Information](#licensing-information)
|
| 52 |
+
- [Citation Information](#citation-information)
|
| 53 |
+
- [Contributions](#contributions)
|
| 54 |
+
|
| 55 |
+
## Dataset Description
|
| 56 |
+
|
| 57 |
+
- **Repository:** [https://github.com/csebuetnlp/banglabert](https://github.com/csebuetnlp/banglabert)
|
| 58 |
+
- **Paper:** [**"BanglaBERT: Combating Embedding Barrier in Multilingual Models for Low-Resource Language Understanding"**](https://arxiv.org/abs/2101.00204)
|
| 59 |
+
- **Point of Contact:** [Tahmid Hasan](mailto:[email protected])
|
| 60 |
+
|
| 61 |
+
### Dataset Summary
|
| 62 |
+
|
| 63 |
+
This is a Natural Language Inference (NLI) dataset for Bengali, curated using the subset of
|
| 64 |
+
MNLI data used in XNLI and state-of-the-art English to Bengali translation model introduced **[here](https://aclanthology.org/2020.emnlp-main.207/).**
|
| 65 |
+
|
| 66 |
+
|
| 67 |
+
### Supported Tasks and Leaderboards
|
| 68 |
+
|
| 69 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 70 |
+
|
| 71 |
+
### Languages
|
| 72 |
+
|
| 73 |
+
* `Bengali`
|
| 74 |
+
|
| 75 |
+
### Usage
|
| 76 |
+
```python
|
| 77 |
+
from datasets import load_dataset
|
| 78 |
+
dataset = load_dataset("csebuetnlp/xnli_bn")
|
| 79 |
+
```
|
| 80 |
+
## Dataset Structure
|
| 81 |
+
|
| 82 |
+
### Data Instances
|
| 83 |
+
|
| 84 |
+
One example from the dataset is given below in JSON format.
|
| 85 |
+
```
|
| 86 |
+
{
|
| 87 |
+
"sentence1": "আসলে, আমি à¦à¦®à¦¨à¦•ি à¦à¦‡ বিষয়ে চিনà§à¦¤à¦¾à¦“ করিনি, কিনà§à¦¤à§ আমি à¦à¦¤ হতাশ হয়ে পড়েছিলাম যে, শেষ পরà§à¦¯à¦¨à§à¦¤ আমি আবার তার সঙà§à¦—ে কথা বলতে শà§à¦°à§ করেছিলাম",
|
| 88 |
+
"sentence2": "আমি তার সাথে আবার কথা বলিনি।",
|
| 89 |
+
"label": "contradiction"
|
| 90 |
+
}
|
| 91 |
+
```
|
| 92 |
+
|
| 93 |
+
### Data Fields
|
| 94 |
+
|
| 95 |
+
The data fields are as follows:
|
| 96 |
+
|
| 97 |
+
- `sentence1`: a `string` feature indicating the premise.
|
| 98 |
+
- `sentence2`: a `string` feature indicating the hypothesis.
|
| 99 |
+
- `label`: a classification label, where possible values are `contradiction` (0), `entailment` (1), `neutral` (2) .
|
| 100 |
+
|
| 101 |
+
### Data Splits
|
| 102 |
+
| split |count |
|
| 103 |
+
|----------|--------|
|
| 104 |
+
|`train`| 381449 |
|
| 105 |
+
|`validation`| 2419 |
|
| 106 |
+
|`test`| 4895 |
|
| 107 |
+
|
| 108 |
+
|
| 109 |
+
|
| 110 |
+
|
| 111 |
+
## Dataset Creation
|
| 112 |
+
|
| 113 |
+
The dataset curation procedure was the same as the [XNLI](https://aclanthology.org/D18-1269/) dataset: we translated the [MultiNLI](https://aclanthology.org/N18-1101/) training data using the English to Bangla translation model introduced [here](https://aclanthology.org/2020.emnlp-main.207/). Due to the possibility of incursions of error during automatic translation, we used the [Language-Agnostic BERT Sentence Embeddings (LaBSE)](https://arxiv.org/abs/2007.01852) of the translations and original sentences to compute their similarity. All sentences below a similarity threshold of 0.70 were discarded.
|
| 114 |
+
|
| 115 |
+
### Curation Rationale
|
| 116 |
+
|
| 117 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 118 |
+
|
| 119 |
+
### Source Data
|
| 120 |
+
|
| 121 |
+
[XNLI](https://aclanthology.org/D18-1269/)
|
| 122 |
+
|
| 123 |
+
#### Initial Data Collection and Normalization
|
| 124 |
+
|
| 125 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 126 |
+
|
| 127 |
+
|
| 128 |
+
#### Who are the source language producers?
|
| 129 |
+
|
| 130 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 131 |
+
|
| 132 |
+
|
| 133 |
+
### Annotations
|
| 134 |
+
|
| 135 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 136 |
+
|
| 137 |
+
|
| 138 |
+
#### Annotation process
|
| 139 |
+
|
| 140 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 141 |
+
|
| 142 |
+
#### Who are the annotators?
|
| 143 |
+
|
| 144 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 145 |
+
|
| 146 |
+
### Personal and Sensitive Information
|
| 147 |
+
|
| 148 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 149 |
+
|
| 150 |
+
## Considerations for Using the Data
|
| 151 |
+
|
| 152 |
+
### Social Impact of Dataset
|
| 153 |
+
|
| 154 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 155 |
+
|
| 156 |
+
### Discussion of Biases
|
| 157 |
+
|
| 158 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 159 |
+
|
| 160 |
+
### Other Known Limitations
|
| 161 |
+
|
| 162 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 163 |
+
|
| 164 |
+
## Additional Information
|
| 165 |
+
|
| 166 |
+
### Dataset Curators
|
| 167 |
+
|
| 168 |
+
[More information needed](https://github.com/csebuetnlp/banglabert)
|
| 169 |
+
|
| 170 |
+
### Licensing Information
|
| 171 |
+
|
| 172 |
+
Contents of this repository are restricted to only non-commercial research purposes under the [Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)](https://creativecommons.org/licenses/by-nc-sa/4.0/). Copyright of the dataset contents belongs to the original copyright holders.
|
| 173 |
+
### Citation Information
|
| 174 |
+
|
| 175 |
+
If you use the dataset, please cite the following paper:
|
| 176 |
+
```
|
| 177 |
+
@misc{bhattacharjee2021banglabert,
|
| 178 |
+
title={BanglaBERT: Combating Embedding Barrier in Multilingual Models for Low-Resource Language Understanding},
|
| 179 |
+
author={Abhik Bhattacharjee and Tahmid Hasan and Kazi Samin and Md Saiful Islam and M. Sohel Rahman and Anindya Iqbal and Rifat Shahriyar},
|
| 180 |
+
year={2021},
|
| 181 |
+
eprint={2101.00204},
|
| 182 |
+
archivePrefix={arXiv},
|
| 183 |
+
primaryClass={cs.CL}
|
| 184 |
+
}
|
| 185 |
+
```
|
| 186 |
+
|
| 187 |
+
|
| 188 |
+
### Contributions
|
| 189 |
+
|
| 190 |
+
Thanks to [@abhik1505040](https://github.com/abhik1505040) and [@Tahmid](https://github.com/Tahmid04) for adding this dataset.
|
data/squad_bn.tar.bz2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1cb33684f2ba0afd68bc0c4e9ec86d5960daa0bb2434c37e062b4758bbc3d6b9
|
| 3 |
+
size 8432345
|
squad_bn.py
ADDED
|
@@ -0,0 +1,120 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""SQuAD Bengali Dataset"""
|
| 2 |
+
|
| 3 |
+
import os
|
| 4 |
+
import json
|
| 5 |
+
|
| 6 |
+
import datasets
|
| 7 |
+
from datasets.tasks import QuestionAnsweringExtractive
|
| 8 |
+
|
| 9 |
+
|
| 10 |
+
_CITATION = """\
|
| 11 |
+
@misc{bhattacharjee2021banglabert,
|
| 12 |
+
title={BanglaBERT: Combating Embedding Barrier in Multilingual Models for Low-Resource Language Understanding},
|
| 13 |
+
author={Abhik Bhattacharjee and Tahmid Hasan and Kazi Samin and Md Saiful Islam and M. Sohel Rahman and Anindya Iqbal and Rifat Shahriyar},
|
| 14 |
+
year={2021},
|
| 15 |
+
eprint={2101.00204},
|
| 16 |
+
archivePrefix={arXiv},
|
| 17 |
+
primaryClass={cs.CL}
|
| 18 |
+
}
|
| 19 |
+
"""
|
| 20 |
+
|
| 21 |
+
_DESCRIPTION = """\
|
| 22 |
+
SQuAD-bn is derived from the SQuAD-2.0 and TyDI-QA datasets.
|
| 23 |
+
"""
|
| 24 |
+
|
| 25 |
+
_HOMEPAGE = "https://github.com/csebuetnlp/banglabert"
|
| 26 |
+
_LICENSE = "Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License (CC BY-NC-SA 4.0)"
|
| 27 |
+
_URL = "https://huggingface.co/datasets/csebuetnlp/squad_bn/resolve/main/data/squad_bn.tar.bz2"
|
| 28 |
+
_VERSION = datasets.Version("0.0.1")
|
| 29 |
+
|
| 30 |
+
|
| 31 |
+
|
| 32 |
+
class SquadBn(datasets.GeneratorBasedBuilder):
|
| 33 |
+
"""SQuAD Bengali Dataset"""
|
| 34 |
+
|
| 35 |
+
BUILDER_CONFIGS = [
|
| 36 |
+
datasets.BuilderConfig(
|
| 37 |
+
name="squad_bn",
|
| 38 |
+
version=_VERSION,
|
| 39 |
+
description=_DESCRIPTION,
|
| 40 |
+
)
|
| 41 |
+
]
|
| 42 |
+
|
| 43 |
+
def _info(self):
|
| 44 |
+
return datasets.DatasetInfo(
|
| 45 |
+
description=_DESCRIPTION,
|
| 46 |
+
features=datasets.Features(
|
| 47 |
+
{
|
| 48 |
+
"id": datasets.Value("string"),
|
| 49 |
+
"title": datasets.Value("string"),
|
| 50 |
+
"context": datasets.Value("string"),
|
| 51 |
+
"question": datasets.Value("string"),
|
| 52 |
+
"answers": datasets.features.Sequence(
|
| 53 |
+
{
|
| 54 |
+
"text": datasets.Value("string"),
|
| 55 |
+
"answer_start": datasets.Value("int32"),
|
| 56 |
+
}
|
| 57 |
+
),
|
| 58 |
+
}
|
| 59 |
+
),
|
| 60 |
+
supervised_keys=None,
|
| 61 |
+
homepage=_HOMEPAGE,
|
| 62 |
+
license=_LICENSE,
|
| 63 |
+
citation=_CITATION,
|
| 64 |
+
task_templates=[
|
| 65 |
+
QuestionAnsweringExtractive(
|
| 66 |
+
question_column="question", context_column="context", answers_column="answers"
|
| 67 |
+
)
|
| 68 |
+
],
|
| 69 |
+
)
|
| 70 |
+
|
| 71 |
+
def _split_generators(self, dl_manager):
|
| 72 |
+
"""Returns SplitGenerators."""
|
| 73 |
+
data_dir = os.path.join(dl_manager.download_and_extract(_URL), "squad_bn")
|
| 74 |
+
return [
|
| 75 |
+
datasets.SplitGenerator(
|
| 76 |
+
name=datasets.Split.TRAIN,
|
| 77 |
+
gen_kwargs={
|
| 78 |
+
"filepath": os.path.join(data_dir, "train.json"),
|
| 79 |
+
},
|
| 80 |
+
),
|
| 81 |
+
datasets.SplitGenerator(
|
| 82 |
+
name=datasets.Split.TEST,
|
| 83 |
+
gen_kwargs={
|
| 84 |
+
"filepath": os.path.join(data_dir, "test.json"),
|
| 85 |
+
},
|
| 86 |
+
),
|
| 87 |
+
datasets.SplitGenerator(
|
| 88 |
+
name=datasets.Split.VALIDATION,
|
| 89 |
+
gen_kwargs={
|
| 90 |
+
"filepath": os.path.join(data_dir, "validation.json"),
|
| 91 |
+
},
|
| 92 |
+
),
|
| 93 |
+
]
|
| 94 |
+
|
| 95 |
+
def _generate_examples(self, filepath):
|
| 96 |
+
"""Yields examples as (key, example) tuples."""
|
| 97 |
+
|
| 98 |
+
with open(filepath, encoding="utf-8") as f:
|
| 99 |
+
data = json.load(f)
|
| 100 |
+
for example in data["data"]:
|
| 101 |
+
title = example.get("title", "")
|
| 102 |
+
for paragraph in example["paragraphs"]:
|
| 103 |
+
context = paragraph["context"].strip()
|
| 104 |
+
for qa in paragraph["qas"]:
|
| 105 |
+
question = qa["question"].strip()
|
| 106 |
+
id_ = qa["id"]
|
| 107 |
+
|
| 108 |
+
answer_starts = [answer["answer_start"] for answer in qa["answers"]]
|
| 109 |
+
answers = [answer["text"].strip() for answer in qa["answers"]]
|
| 110 |
+
|
| 111 |
+
yield id_, {
|
| 112 |
+
"title": title,
|
| 113 |
+
"context": context,
|
| 114 |
+
"question": question,
|
| 115 |
+
"id": id_,
|
| 116 |
+
"answers": {
|
| 117 |
+
"answer_start": answer_starts,
|
| 118 |
+
"text": answers,
|
| 119 |
+
},
|
| 120 |
+
}
|