search

LFM-bot / IOCRec

Pytorch implementation for paper: Multi-Intention Oriented Contrastive Learning for Sequential Recommendation (WSDM23)
13 stars 0 forks source link

Code Request Letter #2

Open dddddFm opened 4 months ago

dddddFm commented 4 months ago 
I am a first-year master's student, currently focusing on recommender systems. I am studying your IOCRec code, but the results I obtained during reproduction differ significantly from those in the README file. Could you please advise on the possible reasons for this discrepancy?

Thank you very much for your work and assistance. I look forward to your guidance on this issue.

LFM-bot commented 4 months ago

Sorry for the late response.  According to your description, I guess the discrepancy may be caused by random seeds, the Pytorch version, and other hyper-parameters (e.g., hidden dimension for the Transformer, named ffn_hidden in the runIOCRec.py). To produce the results, I rerun this code with different random seeds and hidden sizes using Pytorch 2.3.1. However, I haven't encountered the discrepancy you described, and the newly obtained results are quite similar to those in the README. Therefore, I upload the running script (toys.sh) and corresponding logs (in the log folder) to provide more details. Wish this could help you get the right results.  Thanks for your interest in my repository. :)