Potential model duplication with monot5-large-msmarco

#1
by Meranti - opened

Hi,

I just compared performances of monot5-large-msmarco-10k with monot5-large-msmarco on TREC iKAT track 23, and they end up yielding same numbers.

I am quite sure that nothing went wrong from my evaluation pipeline, so I am wondering if it could be because the 2 weights are the same by mistake?

If you could check it would be appreciated. Thank you so much!

Best regards,

Yuchen Hui

Sign up or log in to comment