Datasets:

Modalities:
Text
Formats:
parquet
DOI:
Libraries:
Datasets
pandas
License:
fdelucaf commited on
Commit
e2c8c4d
1 Parent(s): a47addb

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +6 -10
README.md CHANGED
@@ -70,15 +70,10 @@ The dataset contains a single split: `train`.
70
 
71
  ### Source Data
72
 
73
- The dataset is a combination of the following authentic datasets:
74
-
75
- | Dataset | Sentences |
76
- |:-------|-------:|
77
-
78
  The Catalan-Chinese data collected from the web was a combination of the following datasets:
79
 
80
  | Dataset | Sentences before cleaning |
81
- |-------------------|----------------|
82
  | WikiMatrix | 90.643 |
83
  | XLENT | 535.803 |
84
  | GNOME | 78|
@@ -90,7 +85,7 @@ The Catalan-Chinese data collected from the web was a combination of the followi
90
  The 6.658.607 sentence pairs of synthetic parallel data were created from the following Spanish-Chinese datasets:
91
 
92
  | Dataset | Sentences before cleaning |
93
- |-------------------|----------------|
94
  | UNPC |17.599.223|
95
  | CCMatrix | 24.051.233 |
96
  | MultiParacrawl| 3410087|
@@ -98,10 +93,11 @@ The 6.658.607 sentence pairs of synthetic parallel data were created from the fo
98
 
99
  ### Data preparation
100
 
101
- The Chinese side of all datasets are passed through the [fastlangid](https://github.com/currentslab/fastlangid) language detector and any sentences which are not identified
102
- as simplified Chinese are discarded. The datasets are then also deduplicated and filtered to remove any sentence pairs with a cosine similarity of less than 0.75.
 
103
  This is done using sentence embeddings calculated using [LaBSE](https://huggingface.co/sentence-transformers/LaBSE).
104
- The filtered datasets are then concatenated to form a final corpus of 6.833.114 parallel sentences.
105
 
106
  ### Personal and Sensitive Information
107
 
 
70
 
71
  ### Source Data
72
 
 
 
 
 
 
73
  The Catalan-Chinese data collected from the web was a combination of the following datasets:
74
 
75
  | Dataset | Sentences before cleaning |
76
+ |:------------------|---------------:|
77
  | WikiMatrix | 90.643 |
78
  | XLENT | 535.803 |
79
  | GNOME | 78|
 
85
  The 6.658.607 sentence pairs of synthetic parallel data were created from the following Spanish-Chinese datasets:
86
 
87
  | Dataset | Sentences before cleaning |
88
+ |:------------------|---------------:|
89
  | UNPC |17.599.223|
90
  | CCMatrix | 24.051.233 |
91
  | MultiParacrawl| 3410087|
 
93
 
94
  ### Data preparation
95
 
96
+ The Chinese side of all datasets are passed through the [fastlangid](https://github.com/currentslab/fastlangid) language detector and any sentences which are not
97
+ identified as simplified Chinese are discarded.
98
+ The datasets are then also deduplicated and filtered to remove any sentence pairs with a cosine similarity of less than 0.75.
99
  This is done using sentence embeddings calculated using [LaBSE](https://huggingface.co/sentence-transformers/LaBSE).
100
+ The filtered datasets are then concatenated to form a final corpus of **6.833.114** parallel sentences.
101
 
102
  ### Personal and Sensitive Information
103