proof that the wrong tokenizer template can screw up your responses
Browse files- tokenizer_config.json +1 -1
tokenizer_config.json
CHANGED
@@ -3211,7 +3211,7 @@
|
|
3211 |
"PI:USER"
|
3212 |
],
|
3213 |
"bos_token": "[BOS]",
|
3214 |
-
"chat_template": "{% for message in messages %}{% if loop.first and message['role'] != 'system' %}
|
3215 |
"clean_up_tokenization_spaces": true,
|
3216 |
"eos_token": "[|endofturn|]",
|
3217 |
"extra_special_tokens": {},
|
|
|
3211 |
"PI:USER"
|
3212 |
],
|
3213 |
"bos_token": "[BOS]",
|
3214 |
+
"chat_template": "{% for message in messages %}{% if loop.first and message['role'] != 'system' %}[|system|][|endofturn|]{% endif %}[|{{ message['role'] }}|]{{ message['content'] }}[|endofturn|]{% endfor %}{% if add_generation_prompt %}[|assistant|]{% endif %}{%- if add_reasoning_prompt %}[|reasoning|]{% endif %}",
|
3215 |
"clean_up_tokenization_spaces": true,
|
3216 |
"eos_token": "[|endofturn|]",
|
3217 |
"extra_special_tokens": {},
|