hello on Nostr: i thought distance-weighted kNN would be more resillient to OOD than finetuning BERT ...

i thought distance-weighted kNN would be more resillient to OOD than finetuning BERT with the classification head. i am not even sure if i know what i am talking about because i don't know the math behind those magic. but my whole obsession and hate is about abstention... i want to trust classifiers to do actions automatically without me having to babysitting them. i want to cry ;-;