fixed cnr and mk
Browse files- chunking.py +2 -1
chunking.py
CHANGED
@@ -16,7 +16,7 @@ def get_max_word_length(target_languages: list[str]) -> int:
|
|
16 |
"da": 140,
|
17 |
"de": 150,
|
18 |
"pl": 120, # New Languages Start Here Need To Test
|
19 |
-
"mk":
|
20 |
"es": 50,
|
21 |
"it": 60,
|
22 |
"is": 90,
|
@@ -25,6 +25,7 @@ def get_max_word_length(target_languages: list[str]) -> int:
|
|
25 |
"lb": 140,
|
26 |
"sq": 140,
|
27 |
"sl": 140,
|
|
|
28 |
}
|
29 |
|
30 |
max_word_length = 700 # Default Chunk Size
|
|
|
16 |
"da": 140,
|
17 |
"de": 150,
|
18 |
"pl": 120, # New Languages Start Here Need To Test
|
19 |
+
"mk": 20,
|
20 |
"es": 50,
|
21 |
"it": 60,
|
22 |
"is": 90,
|
|
|
25 |
"lb": 140,
|
26 |
"sq": 140,
|
27 |
"sl": 140,
|
28 |
+
"cnr": 60,
|
29 |
}
|
30 |
|
31 |
max_word_length = 700 # Default Chunk Size
|