himani commited on
Commit
947ccef
1 Parent(s): ddffd2f

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +63 -8
README.md CHANGED
@@ -29,7 +29,9 @@ task_categories:
29
  task_ids:
30
  - conditional-text-generation-other-paraphrase-generation
31
  ---
32
- # Dataset Card for "XL-Sum"
 
 
33
  ## Table of Contents
34
  - [Dataset Card Creation Guide](#dataset-card-creation-guide)
35
  - [Table of Contents](#table-of-contents)
@@ -59,19 +61,27 @@ task_ids:
59
  - [Licensing Information](#licensing-information)
60
  - [Citation Information](#citation-information)
61
  - [Contributions](#contributions)
 
62
  ## Dataset Description
 
63
  - **Homepage:** https://indicnlp.ai4bharat.org/indicnlg-suite
64
  - **Paper:** [IndicNLG Suite: Multilingual Datasets for Diverse NLG Tasks in Indic Languages](https://arxiv.org/abs/2203.05437)
65
  - **Point of Contact:**
 
66
  ### Dataset Summary
 
67
  IndicParaphrase is the paraphrasing dataset released as part of IndicNLG Suite. Each
68
  input is paired with up to 5 references. We create this dataset in eleven
69
  languages including as, bn, gu, hi, kn, ml, mr, or, pa, ta, te. The total
70
- size of the dataset is 5.57M.
 
 
71
  ### Supported Tasks and Leaderboards
 
72
  **Tasks:** Paraphrase generation
73
 
74
  **Leaderboards:** Currently there is no Leaderboard for this dataset.
 
75
  ### Languages
76
  - `Assamese (as)`
77
  - `Bengali (bn)`
@@ -84,8 +94,11 @@ size of the dataset is 5.57M.
84
  - `Punjabi (pa)`
85
  - `Tamil (ta)`
86
  - `Telugu (te)`
 
87
  ## Dataset Structure
 
88
  ### Data Instances
 
89
  One example from the `hi` dataset is given below in JSON format.
90
  ```
91
  {
@@ -99,15 +112,20 @@ One example from the `hi` dataset is given below in JSON format.
99
  'target': 'प्रदेश के युवाओं को निजी उद्योगों में 75 प्रतिशत आरक्षण देंगे।'
100
  }
101
  ```
 
102
  ### Data Fields
103
  - `id (string)`: Unique identifier.
104
  - `input (string)`: Input sentence
105
- - `references (list of strings)`: Paraphrases of `input` , ordered according to the least n-gram overlap
106
- - `target (string)`: The first reference (most dissimilar paraphrase)
 
107
 
108
  ### Data Splits
 
109
  We first select 10K instances each for the validation and test and put remaining in the training dataset. `Assamese (as)`, due to its low-resource nature, could only be split into validation and test sets with 4,420 examples each.
110
  Individual dataset with train-dev-test example counts are given below:
 
 
111
  Language | ISO 639-1 Code |Train | Dev | Test |
112
  --------------|----------------|-------|-----|------|
113
  Assamese | as | - | 4,420 | 4,420 |
@@ -120,39 +138,76 @@ Marathi | mr |406,003 | 10,000 | 10,000 |
120
  Oriya | or | 105,970 | 10,000 | 10,000 |
121
  Punjabi | pa | 266,704 | 10,000 | 10,000 |
122
  Tamil | ta | 497,798 | 10,000 | 10,000 |
123
- Telugu | te | 596,283 | 10,000 | 10,000 |
 
124
 
125
  ## Dataset Creation
 
126
  ### Curation Rationale
 
127
  [More information needed]
 
128
  ### Source Data
 
129
  [Samanantar dataset](https://indicnlp.ai4bharat.org/samanantar/)
 
130
  #### Initial Data Collection and Normalization
131
- [Detailed in the paper](https://arxiv.org/abs/2203.05437)
 
 
 
132
  #### Who are the source language producers?
133
- [Detailed in the paper](https://arxiv.org/abs/2203.05437)
 
 
 
 
 
 
 
 
 
 
 
134
 
135
  ### Personal and Sensitive Information
 
136
  [More information needed]
 
137
  ## Considerations for Using the Data
 
138
  ### Social Impact of Dataset
 
139
  [More information needed]
 
140
  ### Discussion of Biases
 
141
  [More information needed]
 
142
  ### Other Known Limitations
 
143
  [More information needed]
 
144
  ## Additional Information
 
145
  ### Dataset Curators
 
146
  [More information needed]
 
147
  ### Licensing Information
 
148
  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.
149
  ### Citation Information
 
150
  If you use any of the datasets, models or code modules, please cite the following paper:
151
  ```
152
  @inproceedings{Kumar2022IndicNLGSM,
153
  title={IndicNLG Suite: Multilingual Datasets for Diverse NLG Tasks in Indic Languages},
154
  author={Aman Kumar and Himani Shrotriya and Prachi Sahu and Raj Dabre and Ratish Puduppully and Anoop Kunchukuttan and Amogh Mishra and Mitesh M. Khapra and Pratyush Kumar},
155
  year={2022},
156
- url = "https://arxiv.org/abs/2203.05437"
157
  ```
 
 
158
  ### Contributions
 
 
29
  task_ids:
30
  - conditional-text-generation-other-paraphrase-generation
31
  ---
32
+
33
+ # Dataset Card for "IndicParaphrase"
34
+
35
  ## Table of Contents
36
  - [Dataset Card Creation Guide](#dataset-card-creation-guide)
37
  - [Table of Contents](#table-of-contents)
 
61
  - [Licensing Information](#licensing-information)
62
  - [Citation Information](#citation-information)
63
  - [Contributions](#contributions)
64
+
65
  ## Dataset Description
66
+
67
  - **Homepage:** https://indicnlp.ai4bharat.org/indicnlg-suite
68
  - **Paper:** [IndicNLG Suite: Multilingual Datasets for Diverse NLG Tasks in Indic Languages](https://arxiv.org/abs/2203.05437)
69
  - **Point of Contact:**
70
+
71
  ### Dataset Summary
72
+
73
  IndicParaphrase is the paraphrasing dataset released as part of IndicNLG Suite. Each
74
  input is paired with up to 5 references. We create this dataset in eleven
75
  languages including as, bn, gu, hi, kn, ml, mr, or, pa, ta, te. The total
76
+ size of the dataset is 5.57M.
77
+
78
+
79
  ### Supported Tasks and Leaderboards
80
+
81
  **Tasks:** Paraphrase generation
82
 
83
  **Leaderboards:** Currently there is no Leaderboard for this dataset.
84
+
85
  ### Languages
86
  - `Assamese (as)`
87
  - `Bengali (bn)`
 
94
  - `Punjabi (pa)`
95
  - `Tamil (ta)`
96
  - `Telugu (te)`
97
+
98
  ## Dataset Structure
99
+
100
  ### Data Instances
101
+
102
  One example from the `hi` dataset is given below in JSON format.
103
  ```
104
  {
 
112
  'target': 'प्रदेश के युवाओं को निजी उद्योगों में 75 प्रतिशत आरक्षण देंगे।'
113
  }
114
  ```
115
+
116
  ### Data Fields
117
  - `id (string)`: Unique identifier.
118
  - `input (string)`: Input sentence
119
+ - `references (list of strings)`: Paraphrases of `input`, ordered according to the least n-gram overlap
120
+ - `target (string)`: The first reference (most dissimilar paraphrase)
121
+
122
 
123
  ### Data Splits
124
+
125
  We first select 10K instances each for the validation and test and put remaining in the training dataset. `Assamese (as)`, due to its low-resource nature, could only be split into validation and test sets with 4,420 examples each.
126
  Individual dataset with train-dev-test example counts are given below:
127
+
128
+
129
  Language | ISO 639-1 Code |Train | Dev | Test |
130
  --------------|----------------|-------|-----|------|
131
  Assamese | as | - | 4,420 | 4,420 |
 
138
  Oriya | or | 105,970 | 10,000 | 10,000 |
139
  Punjabi | pa | 266,704 | 10,000 | 10,000 |
140
  Tamil | ta | 497,798 | 10,000 | 10,000 |
141
+ Telugu | te | 596,283 | 10,000 | 10,000 |
142
+
143
 
144
  ## Dataset Creation
145
+
146
  ### Curation Rationale
147
+
148
  [More information needed]
149
+
150
  ### Source Data
151
+
152
  [Samanantar dataset](https://indicnlp.ai4bharat.org/samanantar/)
153
+
154
  #### Initial Data Collection and Normalization
155
+
156
+ [Detailed in the paper](https://arxiv.org/abs/2203.05437)
157
+
158
+
159
  #### Who are the source language producers?
160
+
161
+ [Detailed in the paper](https://arxiv.org/abs/2203.05437)
162
+
163
+
164
+ ### Annotations
165
+ [More information needed]
166
+ #### Annotation process
167
+ [More information needed]
168
+
169
+ #### Who are the annotators?
170
+
171
+ [More information needed]
172
 
173
  ### Personal and Sensitive Information
174
+
175
  [More information needed]
176
+
177
  ## Considerations for Using the Data
178
+
179
  ### Social Impact of Dataset
180
+
181
  [More information needed]
182
+
183
  ### Discussion of Biases
184
+
185
  [More information needed]
186
+
187
  ### Other Known Limitations
188
+
189
  [More information needed]
190
+
191
  ## Additional Information
192
+
193
  ### Dataset Curators
194
+
195
  [More information needed]
196
+
197
  ### Licensing Information
198
+
199
  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.
200
  ### Citation Information
201
+
202
  If you use any of the datasets, models or code modules, please cite the following paper:
203
  ```
204
  @inproceedings{Kumar2022IndicNLGSM,
205
  title={IndicNLG Suite: Multilingual Datasets for Diverse NLG Tasks in Indic Languages},
206
  author={Aman Kumar and Himani Shrotriya and Prachi Sahu and Raj Dabre and Ratish Puduppully and Anoop Kunchukuttan and Amogh Mishra and Mitesh M. Khapra and Pratyush Kumar},
207
  year={2022},
208
+ url = "https://arxiv.org/abs/2203.05437",
209
  ```
210
+
211
+
212
  ### Contributions
213
+