ceos/unsloth
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

fix: handle missing add_generation_prompt in saved LoRA tokenizers

Browse source
This commit is contained in:
Fizza-Mukhtar 2026-03-07 07:35:33 -08:00 • committed by Daniel Han
commit 232e786cbe
1 changed files with 21 additions and 3 deletions
Download patch file Download diff file Expand all files Collapse all files

24
unsloth/tokenizer_utils.py
View file

@ -658,6 +658,21 @@ def _fix_chat_template(chat_template):
dash = "-" if chosen_end.startswith("{%-") else ""
# Case 1: nothing after endfor - template was saved without generation prompt
if (
not after_endfor.strip()
and "add_generation_prompt" not in chat_template
and "<|im_start|>" in chat_template
):
after_endfor = (
"{%" + dash + " if add_generation_prompt %}"
+ "{{ '<|im_start|>assistant\\n' }}"
+ "{%" + dash + " endif %}"
)
chat_template = chat_template[: where + len(chosen_end)] + after_endfor
return chat_template
# Case 2: original logic - something after endfor starting with {{
if (
"{%" + dash + " if" not in after_endfor
and "{%" + dash + " set " not in after_endfor
@ -731,11 +746,14 @@ def fix_chat_template(tokenizer):
"{% if add_generation_prompt %}" not in new_chat_template
and "{%- if add_generation_prompt %}" not in new_chat_template
):
raise RuntimeError(
# Could not auto-fix - warn but don't crash so user can still
# load their saved LoRA adapter for inference/chat
logger.warning_once(
f"Unsloth: The tokenizer `{tokenizer.name_or_path}`\n"
"does not have a {% if add_generation_prompt %} for generation purposes.\n"
f"Please file a bug report to the maintainers of `{tokenizer.name_or_path}` - thanks!"
"does not have a {{% if add_generation_prompt %}} for generation purposes.\n"
"Generation may not work correctly. Please notify the model maintainers."
)
return chat_template
else:
logger.warning_once(
"Unsloth: We successfully patched the tokenizer to add a {% if add_generation_prompt %} to the chat_template.\n"