Spaces:
Paused
Paused
nroggendorff
commited on
Commit
•
5c9b987
1
Parent(s):
b06da2c
Update train.py
Browse files
train.py
CHANGED
@@ -105,9 +105,10 @@ def configure_tokenizer(tokenizer):
|
|
105 |
"eos_token": "</s>",
|
106 |
"unk_token": "<unk>",
|
107 |
"pad_token": "<pad>",
|
108 |
-
"mask_token": "<mask>"
|
109 |
-
"additional_special_tokens": ["<|user|>", "<|bot|>", "<|end|>"]
|
110 |
}
|
|
|
|
|
111 |
tokenizer.add_special_tokens(special_tokens)
|
112 |
|
113 |
if INSTRUCT_FINETUNE_BOOL:
|
|
|
105 |
"eos_token": "</s>",
|
106 |
"unk_token": "<unk>",
|
107 |
"pad_token": "<pad>",
|
108 |
+
"mask_token": "<mask>"
|
|
|
109 |
}
|
110 |
+
if INSTRUCT_FINETUNE_BOOL:
|
111 |
+
special_tokens["additional_special_tokens"] = ["<|user|>", "<|bot|>", "<|end|>"]
|
112 |
tokenizer.add_special_tokens(special_tokens)
|
113 |
|
114 |
if INSTRUCT_FINETUNE_BOOL:
|