Austin Davis
commited on
Commit
·
04a3e2b
1
Parent(s):
97550e1
Upload tokenizer
Browse files- special_tokens_map.json +0 -1
- tokenizer_config.json +1 -8
special_tokens_map.json
CHANGED
|
@@ -1,6 +1,5 @@
|
|
| 1 |
{
|
| 2 |
"bos_token": "<s>",
|
| 3 |
-
"eos_token": "<|endoftext|>",
|
| 4 |
"pad_token": "<pad>",
|
| 5 |
"unk_token": "<pad>"
|
| 6 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"bos_token": "<s>",
|
|
|
|
| 3 |
"pad_token": "<pad>",
|
| 4 |
"unk_token": "<pad>"
|
| 5 |
}
|
tokenizer_config.json
CHANGED
|
@@ -2,15 +2,8 @@
|
|
| 2 |
"add_bos_token": true,
|
| 3 |
"bos_token": "<s>",
|
| 4 |
"clean_up_tokenization_spaces": true,
|
| 5 |
-
"max_length": 512,
|
| 6 |
"model_max_length": 1000000000000000019884624838656,
|
| 7 |
-
"pad_to_multiple_of": null,
|
| 8 |
"pad_token": "<pad>",
|
| 9 |
-
"
|
| 10 |
-
"padding_side": "right",
|
| 11 |
-
"stride": 0,
|
| 12 |
-
"tokenizer_class": "PreTrainedTokenizerFast",
|
| 13 |
-
"truncation_side": "right",
|
| 14 |
-
"truncation_strategy": "longest_first",
|
| 15 |
"unk_token": "<pad>"
|
| 16 |
}
|
|
|
|
| 2 |
"add_bos_token": true,
|
| 3 |
"bos_token": "<s>",
|
| 4 |
"clean_up_tokenization_spaces": true,
|
|
|
|
| 5 |
"model_max_length": 1000000000000000019884624838656,
|
|
|
|
| 6 |
"pad_token": "<pad>",
|
| 7 |
+
"tokenizer_class": "ChessTokenizer",
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8 |
"unk_token": "<pad>"
|
| 9 |
}
|