Upload data/processed/shared_vocab_info.json
Browse files
data/processed/shared_vocab_info.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 32002,
|
| 3 |
+
"sos_id": 1,
|
| 4 |
+
"eos_id": 2,
|
| 5 |
+
"pad_id": 0,
|
| 6 |
+
"unk_id": 3,
|
| 7 |
+
"en2vi_id": 32000,
|
| 8 |
+
"vi2en_id": 32001,
|
| 9 |
+
"tokenizer_path": "/home/alida/Documents/Cursor/NLP/data/processed/tokenizer_shared.json"
|
| 10 |
+
}
|