fix: tokenizer chat template sys msg
#1
by
legraphista
- opened
- tokenizer_config.json +1 -1
tokenizer_config.json
CHANGED
@@ -1744,7 +1744,7 @@
|
|
1744 |
"<end_of_turn>"
|
1745 |
],
|
1746 |
"bos_token": "<bos>",
|
1747 |
-
"chat_template": "{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{
|
1748 |
"clean_up_tokenization_spaces": false,
|
1749 |
"eos_token": "<eos>",
|
1750 |
"model_max_length": 1000000000000000019884624838656,
|
|
|
1744 |
"<end_of_turn>"
|
1745 |
],
|
1746 |
"bos_token": "<bos>",
|
1747 |
+
"chat_template": "{{ '<bos>' }}{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] %}{{ system_message }}{% endif %}{% for message in messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ '<start_of_turn>user\\n' + content + '<end_of_turn>\\n<start_of_turn>model\\n' }}{% elif message['role'] == 'assistant' %}{{ content + '<end_of_turn>\\n' }}{% endif %}{% endfor %}",
|
1748 |
"clean_up_tokenization_spaces": false,
|
1749 |
"eos_token": "<eos>",
|
1750 |
"model_max_length": 1000000000000000019884624838656,
|