distillation — Training a small, cheap model to imitate a big one's outputs, keeping much of the capability at a fraction of the cost.
Why it matters
EVIE-8B is the uncompressed teacher model behind Tencent's retrieval line, giving developers a highest-accuracy option (66.75 nDCG@10 on ViDoRe V3) when index size matters less than top retrieval quality for complex document layouts and tables.
