Upload tokenizer_config.json with huggingface_hub
Browse files- tokenizer_config.json +21 -0
tokenizer_config.json
ADDED
|
@@ -0,0 +1,21 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 50000,
|
| 3 |
+
"special_tokens": [
|
| 4 |
+
"[PAD]",
|
| 5 |
+
"[UNK]",
|
| 6 |
+
"[CLS]",
|
| 7 |
+
"[SEP]",
|
| 8 |
+
"[MASK]",
|
| 9 |
+
"[BOS]",
|
| 10 |
+
"[EOS]",
|
| 11 |
+
"<|im_start|>",
|
| 12 |
+
"<|im_end|>",
|
| 13 |
+
"<|system|>",
|
| 14 |
+
"<|user|>",
|
| 15 |
+
"<|assistant|>"
|
| 16 |
+
],
|
| 17 |
+
"model_type": "BPE",
|
| 18 |
+
"training_corpus": "/content/drive/MyDrive/turkish_nlp_project/FINAL_TURKISH_CORPUS.txt",
|
| 19 |
+
"training_date": "2025-09-06 13:37:54",
|
| 20 |
+
"corpus_size_mb": 19.955299377441406
|
| 21 |
+
}
|