Saluru Durga Sandeep

icon location Amherst, MA icon location LLM Researcher at BioNLP Lab | MS in UMass Amherst | Dual Degree in IIT Madras

Results 1 comments of Saluru Durga Sandeep

BioMistral Generating \n repeatedly after training

I have tried bad_word_ids to not generate "\n". But I am still not able to understand why the SFT trained model is generating that way (raw BioMistral is working fine)