reasoning-on-graphs plug-and-play.sh bugs

There appears to be an issue with the plug-and-play.sh script where ${MODEL_NAME} is referenced before it's defined. It seems like RoG was intended to be the default model. Should the line be updated to: RULE_PATH=results/gen_rule_path/${DATA_NAME}/RoG/test/predictions_${N_BEAM}_False.jsonl?

Regarding other LLMs, the default model path specified in https://github.com/RManLuo/reasoning-on-graphs/blob/master/src/llms/language_models/alpaca.py uses an in-house absolute path. Should this be updated to a Hugging Face URL instead? Would wxjiao/alpaca-7b be an appropriate substitute?

Thank you for your diligent work and the code you've shared; it's been incredibly useful.

Jan 25 '25 16:01 DongzhuoranZhou

Thanks for pointing out. I have updated the script. The alpaca model we used is from the official weight diff. I am not sure it is okay to replace it with the HF version. Thus, I leave as it is now.

Mar 05 '25 11:03 RManLuo

Greetings, I hope you are all doing well.

The link of pre-trained weights: https://huggingface.co/rmanluo/RoG can not be found. I hope you can update it.

Apr 28 '25 13:04 YRVGFO9588