rajammanabrolu
commited on
Commit
•
2177a48
1
Parent(s):
fe7c483
Upload tokenizer
Browse files- special_tokens_map.json +1 -0
- tokenizer_config.json +1 -1
special_tokens_map.json
CHANGED
@@ -5,5 +5,6 @@
|
|
5 |
],
|
6 |
"bos_token": "<|endoftext|>",
|
7 |
"eos_token": "<|endoftext|>",
|
|
|
8 |
"unk_token": "<|endoftext|>"
|
9 |
}
|
|
|
5 |
],
|
6 |
"bos_token": "<|endoftext|>",
|
7 |
"eos_token": "<|endoftext|>",
|
8 |
+
"pad_token": "<|pad|>",
|
9 |
"unk_token": "<|endoftext|>"
|
10 |
}
|
tokenizer_config.json
CHANGED
@@ -18,7 +18,7 @@
|
|
18 |
"eos_token": "<|endoftext|>",
|
19 |
"model_max_length": 8192,
|
20 |
"model_name": "gpt-4",
|
21 |
-
"pad_token":
|
22 |
"tokenizer_class": "TiktokenTokenizerWrapper",
|
23 |
"unk_token": "<|endoftext|>"
|
24 |
}
|
|
|
18 |
"eos_token": "<|endoftext|>",
|
19 |
"model_max_length": 8192,
|
20 |
"model_name": "gpt-4",
|
21 |
+
"pad_token": "<|pad|>",
|
22 |
"tokenizer_class": "TiktokenTokenizerWrapper",
|
23 |
"unk_token": "<|endoftext|>"
|
24 |
}
|