Locutusque
commited on
Commit
•
bbce778
1
Parent(s):
ec8b4d2
Upload tokenizer
Browse files- tokenizer_config.json +8 -1
tokenizer_config.json
CHANGED
@@ -2052,11 +2052,18 @@
|
|
2052 |
"bos_token": "<|begin_of_text|>",
|
2053 |
"clean_up_tokenization_spaces": true,
|
2054 |
"eos_token": "<|end_of_text|>",
|
|
|
2055 |
"model_input_names": [
|
2056 |
"input_ids",
|
2057 |
"attention_mask"
|
2058 |
],
|
2059 |
"model_max_length": 1000000000000000019884624838656,
|
|
|
2060 |
"pad_token": "<|end_of_text|>",
|
2061 |
-
"
|
|
|
|
|
|
|
|
|
|
|
2062 |
}
|
|
|
2052 |
"bos_token": "<|begin_of_text|>",
|
2053 |
"clean_up_tokenization_spaces": true,
|
2054 |
"eos_token": "<|end_of_text|>",
|
2055 |
+
"max_length": 512,
|
2056 |
"model_input_names": [
|
2057 |
"input_ids",
|
2058 |
"attention_mask"
|
2059 |
],
|
2060 |
"model_max_length": 1000000000000000019884624838656,
|
2061 |
+
"pad_to_multiple_of": null,
|
2062 |
"pad_token": "<|end_of_text|>",
|
2063 |
+
"pad_token_type_id": 0,
|
2064 |
+
"padding_side": "right",
|
2065 |
+
"stride": 0,
|
2066 |
+
"tokenizer_class": "PreTrainedTokenizerFast",
|
2067 |
+
"truncation_side": "right",
|
2068 |
+
"truncation_strategy": "longest_first"
|
2069 |
}
|