Marco

@aurorarossi thanks for the comments!

Yes, it would be better, if that is something that should actually be done. I don't know if there are any weird considerations that need to be addressed when summing...

Hi, thanks for the background info. IMO, the main reason one would want a validation-based early-stopping mechanism that isn't tied directly to the learning rate is when you are training...

Understood, ofc I will defer to your judgement. However, do you mind if I open (and then immediately close) a PR here linked to this issue so that others can...

Ah, yes an absolute path would solve this. However, are there other instances where you force an absolute rather than relative path in JoeyNMT? If not, maybe it is too...

Sorry for the slow reply. Hmm, it is interesting that it works on your machine but not mine (actually, I was able to reproduce it on three separate machines). I...

I also was not able to convert llama2-7B despite having 32GB ram (closed everything out, and htop reported only 2GB being used by other processes, and swp was totally clear)....

Hi @taha-yassine I recovered my old laptop and was able to find the dataset that was in the (now defunct) MEGA link. It is about 4.4Gb on disk when compressed...

I've sent you a message on Twitter (I don't want to share the link here, though I am not sure if it would actually cause me any issues). I will...