safikhan commited on
Commit
289f51a
1 Parent(s): d9ca2bb

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +19 -1
README.md CHANGED
@@ -182,7 +182,7 @@ Sangraha is the largest high-quality, cleaned Indic language pretraining data co
182
 
183
  ## Getting Started
184
 
185
- You can download the dataset using Hugging Face datasets:
186
 
187
  ```python
188
  from datasets import load_dataset
@@ -190,6 +190,24 @@ from datasets import load_dataset
190
  dataset = load_dataset("ai4bharat/sangraha")
191
  ```
192
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
193
  ## Background
194
 
195
  Sangraha contains three broad components:
 
182
 
183
  ## Getting Started
184
 
185
+ For downloading the entire Sangraha:
186
 
187
  ```python
188
  from datasets import load_dataset
 
190
  dataset = load_dataset("ai4bharat/sangraha")
191
  ```
192
 
193
+ For downloading a subset (Verified/Unverified) of Sangraha:
194
+
195
+ ```python
196
+ from datasets import load_dataset
197
+
198
+ dataset = load_dataset("ai4bharat/sangraha", data_dir="<subset_name>")
199
+ # for example: dataset = load_dataset("ai4bharat/sangraha", data_dir="verified")
200
+ ```
201
+
202
+ For downloading one language from a subset of Sangraha:
203
+
204
+ ```python
205
+ from datasets import load_dataset
206
+
207
+ dataset = load_dataset("ai4bharat/sangraha", data_dir="<subset_name>/<lang_code>")
208
+ # for example: dataset = load_dataset("ai4bharat/sangraha", data_dir="verified/asm")
209
+ ```
210
+
211
  ## Background
212
 
213
  Sangraha contains three broad components: