kargaranamir
commited on
Commit
β’
b0e51d1
1
Parent(s):
2029420
add srp, snk, enm and edit abk, ady, che, mon, oss.
Browse files- languages/abk_Cyrl.json +4 -4
- languages/ady_Cyrl.json +4 -4
- languages/che_Cyrl.json +4 -4
- languages/enm_Latn.json +19 -0
- languages/mon_Mong.json +2 -2
- languages/oss_Cyrl.json +4 -4
- languages/snk_Latn.json +19 -0
- languages/srp_Latn.json +28 -0
languages/abk_Cyrl.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"Language Name": "Abkhazian",
|
3 |
-
"Family": "",
|
4 |
-
"Subgrouping": "",
|
5 |
-
"Number of Speakers": "",
|
6 |
-
"Supported by allenai/MADLAD-400 or facebook/flores": "",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
|
|
1 |
{
|
2 |
"Language Name": "Abkhazian",
|
3 |
+
"Family": "Abkhaz-Adyge",
|
4 |
+
"Subgrouping": "Abkhaz-Abaza",
|
5 |
+
"Number of Speakers": "190_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
languages/ady_Cyrl.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"Language Name": "Adyghe",
|
3 |
-
"Family": "",
|
4 |
-
"Subgrouping": "",
|
5 |
-
"Number of Speakers": "",
|
6 |
-
"Supported by allenai/MADLAD-400 or facebook/flores": "",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
|
|
1 |
{
|
2 |
"Language Name": "Adyghe",
|
3 |
+
"Family": "Abkhaz-Adyge",
|
4 |
+
"Subgrouping": "Circassian",
|
5 |
+
"Number of Speakers": "610_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π¨",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
languages/che_Cyrl.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"Language Name": "Chechen",
|
3 |
-
"Family": "",
|
4 |
-
"Subgrouping": "",
|
5 |
-
"Number of Speakers": "",
|
6 |
-
"Supported by allenai/MADLAD-400 or facebook/flores": "",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
|
|
1 |
{
|
2 |
"Language Name": "Chechen",
|
3 |
+
"Family": "Nakh-Daghestanian",
|
4 |
+
"Subgrouping": "Nakh",
|
5 |
+
"Number of Speakers": "1_800_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π¨",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
languages/enm_Latn.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Middle English",
|
3 |
+
"Family": "Indo-European",
|
4 |
+
"Subgrouping": "Germanic",
|
5 |
+
"Number of Speakers": "0",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "quod.lib.umich.edu/c/cme/AHA2727/",
|
10 |
+
"Site URL": "https://quod.lib.umich.edu/c/cme/AHA2727/",
|
11 |
+
"Category": "book",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "confirmed by glotlid",
|
14 |
+
"Possible Parallel Languages": "",
|
15 |
+
"Links": []
|
16 |
+
}
|
17 |
+
]
|
18 |
+
}
|
19 |
+
|
languages/mon_Mong.json
CHANGED
@@ -1,7 +1,7 @@
|
|
1 |
{
|
2 |
-
"Language Name": "Mongolian",
|
3 |
"Family": "Mongolic",
|
4 |
-
"Subgrouping": "
|
5 |
"Number of Speakers": "6_000_000",
|
6 |
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
"Sites": [
|
|
|
1 |
{
|
2 |
+
"Language Name": "Mongolian (Inner Mongolia)",
|
3 |
"Family": "Mongolic",
|
4 |
+
"Subgrouping": "",
|
5 |
"Number of Speakers": "6_000_000",
|
6 |
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
"Sites": [
|
languages/oss_Cyrl.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"Language Name": "Ossetian",
|
3 |
-
"Family": "",
|
4 |
-
"Subgrouping": "",
|
5 |
-
"Number of Speakers": "",
|
6 |
-
"Supported by allenai/MADLAD-400 or facebook/flores": "",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
|
|
1 |
{
|
2 |
"Language Name": "Ossetian",
|
3 |
+
"Family": "Indo-European",
|
4 |
+
"Subgrouping": "Iranian",
|
5 |
+
"Number of Speakers": "490_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π¨",
|
7 |
"Sites": [
|
8 |
{
|
9 |
"Site Name": "jinepsgazetesi.com",
|
languages/snk_Latn.json
ADDED
@@ -0,0 +1,19 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Soninke",
|
3 |
+
"Family": "Mande",
|
4 |
+
"Subgrouping": "Soninke-Bozo",
|
5 |
+
"Number of Speakers": "2_300_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "soninkara.com/snk/",
|
10 |
+
"Site URL": "http://www.soninkara.com/snk/",
|
11 |
+
"Category": "news,forums",
|
12 |
+
"Confidence": "π©",
|
13 |
+
"Info": "confirmed by webpage metadata",
|
14 |
+
"Possible Parallel Languages": "fra_Latn, eng_Latn",
|
15 |
+
"Links": []
|
16 |
+
}
|
17 |
+
]
|
18 |
+
}
|
19 |
+
|
languages/srp_Latn.json
ADDED
@@ -0,0 +1,28 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"Language Name": "Serbian (Latin)",
|
3 |
+
"Family": "Indo-European",
|
4 |
+
"Subgrouping": "Slavic",
|
5 |
+
"Number of Speakers": "12_000_000",
|
6 |
+
"Supported by allenai/MADLAD-400 or facebook/flores": "π₯",
|
7 |
+
"Sites": [
|
8 |
+
{
|
9 |
+
"Site Name": "turizamiputovanja.com",
|
10 |
+
"Site URL": "https://www.turizamiputovanja.com/",
|
11 |
+
"Category": "tourist blog",
|
12 |
+
"Confidence": "π¨",
|
13 |
+
"Info": "confirmed based on the content provided",
|
14 |
+
"Possible Parallel Languages": "",
|
15 |
+
"Links": []
|
16 |
+
},
|
17 |
+
{
|
18 |
+
"Site Name": "vijesti.me",
|
19 |
+
"Site URL": "https://www.vijesti.me/",
|
20 |
+
"Category": "news",
|
21 |
+
"Confidence": "π¨",
|
22 |
+
"Info": "confirmed based on the flag.",
|
23 |
+
"Possible Parallel Languages": "eng_Latn",
|
24 |
+
"Links": []
|
25 |
+
}
|
26 |
+
]
|
27 |
+
}
|
28 |
+
|