Update README.md
Browse files
README.md
CHANGED
@@ -8,8 +8,12 @@ license: mit
|
|
8 |
|
9 |
# Dataset Card for HLSNKIOPUS100
|
10 |
|
11 |
-
This dataset contains English-Persian translation pairs from the OPUS-100 corpus.
|
12 |
-
|
|
|
|
|
|
|
|
|
13 |
|
14 |
## Dataset Details
|
15 |
|
|
|
8 |
|
9 |
# Dataset Card for HLSNKIOPUS100
|
10 |
|
11 |
+
This dataset contains English-Persian translation pairs from the OPUS-100 corpus.
|
12 |
+
The original corpus of OPUS-100 is movie dialogues and gets repetitive really fast,
|
13 |
+
it also contains duplicates, false translations and huge chunks of unrelated data,
|
14 |
+
making the dataset very noisy and almost unsuitable for any usage.
|
15 |
+
We have deduplicated, denoised and cleaned the dataset, reviewed the translations,
|
16 |
+
retranslated and cleared the meaning across all the entries to reach a standard version suitable for deep learning tasks.
|
17 |
|
18 |
## Dataset Details
|
19 |
|