Text Generation
Transformers
Safetensors
Japanese
English
mistral
conversational
text-generation-inference
leonardlin commited on
Commit
47f0a0e
·
1 Parent(s): aedf730

BOS token fix - was not being emitted when there was no system prompt, leading to bad output

Browse files
Files changed (1) hide show
  1. tokenizer_config.json +1 -1
tokenizer_config.json CHANGED
@@ -467,7 +467,7 @@
467
  },
468
  "additional_special_tokens": [],
469
  "bos_token": "<s>",
470
- "chat_template": "{%- for idx in range(0, messages|length) -%}\n{%- if messages[idx]['role'] == 'user' -%}\n{%- if idx > 1 -%}\n{{- bos_token + '[INST] ' + messages[idx]['content'] + ' [/INST]' -}}\n{%- else -%}\n{{- messages[idx]['content'] + ' [/INST]' -}}\n{%- endif -%}\n{% elif messages[idx]['role'] == 'system' %}\n{{- bos_token + '[INST] <<SYS>>\\n' + messages[idx]['content'] + '\\n<</SYS>>\\n\\n' -}}\n{%- elif messages[idx]['role'] == 'assistant' -%}\n{{- ' ' + messages[idx]['content'] + ' ' + eos_token -}}\n{% endif %}\n{% endfor %}\n",
471
  "clean_up_tokenization_spaces": false,
472
  "eos_token": "</s>",
473
  "legacy": true,
 
467
  },
468
  "additional_special_tokens": [],
469
  "bos_token": "<s>",
470
+ "chat_template": "{%- for idx in range(0, messages|length) -%}\n{%- set message = messages[idx] -%}\n{%- if message['role'] == 'system' -%}\n{{- bos_token + '[INST] <<SYS>>\\n' + message['content'] + '\\n<</SYS>>\\n\\n' -}}\n{%- elif message['role'] == 'user' -%}\n{%- if idx > 0 and messages[idx-1]['role'] == 'system' -%}\n{{- message['content'] + ' [/INST]' -}}\n{%- else -%}\n{{- bos_token + '[INST] ' + message['content'] + ' [/INST]' -}}\n{%- endif -%}\n{%- elif message['role'] == 'assistant' -%}\n{{- ' ' + message['content'] + ' ' + eos_token -}}\n{%- endif -%}\n{% endfor %}\n",
471
  "clean_up_tokenization_spaces": false,
472
  "eos_token": "</s>",
473
  "legacy": true,