Upload tokenizer_config.json with huggingface_hub
Browse files- tokenizer_config.json +22 -0
tokenizer_config.json
ADDED
|
@@ -0,0 +1,22 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"vocab_size": 8192,
|
| 3 |
+
"special_tokens": [
|
| 4 |
+
"<pad>",
|
| 5 |
+
"<unk>",
|
| 6 |
+
"<bos>",
|
| 7 |
+
"<eos>"
|
| 8 |
+
],
|
| 9 |
+
"pad_token": "<pad>",
|
| 10 |
+
"unk_token": "<unk>",
|
| 11 |
+
"bos_token": "<bos>",
|
| 12 |
+
"eos_token": "<eos>",
|
| 13 |
+
"pad_id": 0,
|
| 14 |
+
"unk_id": 1,
|
| 15 |
+
"bos_id": 2,
|
| 16 |
+
"eos_id": 3,
|
| 17 |
+
"min_frequency": 2,
|
| 18 |
+
"model_max_length": 128,
|
| 19 |
+
"padding_side": "right",
|
| 20 |
+
"truncation_side": "right",
|
| 21 |
+
"tokenizer_class": "PreTrainedTokenizerFast"
|
| 22 |
+
}
|