wassname/alignment-handbook
1
0
Fork 0
mirror of https://github.com/wassname/alignment-handbook.git synced 2026-09-11 12:00:25 +08:00
Code Issues Packages Projects Releases Wiki Activity

better formating

Browse Source
This commit is contained in:
wassname
2025-06-03 22:21:18 +00:00
parent 097e4e0b01
commit a264efaa4c
10 changed files with 752 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files
scripts/run_sft.py
+6 -3
View File
@@ -121,10 +121,13 @@ def main():
)
model = model_args.model_name_or_path
# For ChatML we need to add special tokens and resize the embedding layer
if "<|im_start|>" in tokenizer.chat_template and "gemma-tokenizer-chatml" not in tokenizer.name_or_path:
if tokenizer.chat_template is None:
# it conflated having chatml with adding chatml
# For ChatML we need to add special tokens and resize the embedding layer
logger.info(f"No chat template found on the tokenizer, setting up ChatML format for {model_args.model_name_or_path}.")
model = AutoModelForCausalLM.from_pretrained(model_args.model_name_or_path, **model_kwargs)
tokenizer.chat_template = None # Not quite sure why I have to do this, but if we end up with chatml for all models, and it's saved in the output tokeniser, that's fine
model, tokenizer = setup_chat_format(model, tokenizer)
model_kwargs = None