Comment by dspoka

Looks cool! Any advantages to the mini-lm model - it seems better on most mteb tasks but wondering if maybe inference or something is better.