ZhankuiHe commited on
Commit
c27c033
·
verified ·
1 Parent(s): cfc7440

Upload tokenizer

Browse files
dialog/tokenizer_config.json CHANGED
@@ -20,7 +20,7 @@
20
  }
21
  },
22
  "bos_token": "<|endoftext|>",
23
- "chat_template": "{% for message in messages %}{{ message.role + '\n' + message.content | replace('<e>', '') | replace('</e>', '') | replace('_', ' ') }}{{ eos_token }}{% endfor %}{{ 'assistant\n' }}",
24
  "clean_up_tokenization_spaces": true,
25
  "eos_token": "<|endoftext|>",
26
  "errors": "replace",
 
20
  }
21
  },
22
  "bos_token": "<|endoftext|>",
23
+ "chat_template": "{% for message in messages %}{{ message.role + '\n' + message.content + '\n' | replace('<e>', '') | replace('</e>', '') | replace('_', ' ') }}{{ eos_token }}{% endfor %}{{ 'assistant\n' }}",
24
  "clean_up_tokenization_spaces": true,
25
  "eos_token": "<|endoftext|>",
26
  "errors": "replace",
word/tokenizer_config.json CHANGED
@@ -51,7 +51,7 @@
51
  }
52
  },
53
  "bos_token": "<s>",
54
- "chat_template": "{% for message in messages %}{{ message.role + '\n' + message.content | replace('<e>', '') | replace('</e>', '') | replace('_', ' ') }}{{ eos_token }}{% endfor %}",
55
  "clean_up_tokenization_spaces": true,
56
  "cls_token": "<s>",
57
  "eos_token": "</s>",
 
51
  }
52
  },
53
  "bos_token": "<s>",
54
+ "chat_template": "{% for message in messages %}{{ message.role + '\n' + message.content + '\n' | replace('<e>', '') | replace('</e>', '') | replace('_', ' ') }}{{ eos_token }}{% endfor %}",
55
  "clean_up_tokenization_spaces": true,
56
  "cls_token": "<s>",
57
  "eos_token": "</s>",