Came here to say this. It's behind the 14b Phi-reasoning-plus (which is self-reported). I don't understand why "TIGER-LAb"-sourced scores are 'unknown' in term

Came here to say this. It's behind the 14b Phi-reasoning-plus (which is self-reported). I don't understand why "TIGER-LAb"-sourced scores are 'unknown' in terms of model size?