Horizon-Labs's picture
Multilingual prompt task & complexity classifier base v1.0
aa93893 verified
Raw History Blame Contribute Delete
994 Bytes
{
"nvidia": {
"n": 4200,
"task_acc": 0.45880952380952383,
"task_macro_f1": 0.3176954339403373,
"task_acc_lang_mean": 0.4588095238095239,
"task_acc_en": 0.655,
"spearman_creativity": 0.4777647258895101,
"spearman_reasoning": 0.2497500089936578,
"spearman_contextual_knowledge": 0.2302563300742145,
"spearman_domain_knowledge": 0.38518510108812076,
"spearman_constraints": 0.6480432086980401,
"spearman_complexity": 0.3471741488551528,
"spearman_mean": 0.3896955872664494
},
"teacher": {
"n": 4200,
"task_acc": 0.8557142857142858,
"task_macro_f1": 0.8071549940704531,
"task_acc_lang_mean": 0.8557142857142856,
"task_acc_en": 0.915,
"spearman_creativity": 0.8556580154707426,
"spearman_reasoning": 0.8302711408392133,
"spearman_contextual_knowledge": 0.6876942315551514,
"spearman_domain_knowledge": 0.808322633505938,
"spearman_constraints": 0.9146826907015201,
"spearman_complexity": 0.8191655790270074,
"spearman_mean": 0.8192990485165955
}
}