quiz bu lëkkale ak tegtal · Diggante Niveau

Adaptatëru couche ngir quiz toxal

Saytu ni adaptatëru bottleneck yi di denc modelu base bu liw, may liggéey modular wala toxal domen, ba noppi méngale ko ak LoRA.

Yooni njiit yu jëm ciAdaptatër Layer ngir toxal
Laaj 1 de 8

Lan mooy dall poid yiñ njëkka tàggat suñuy tàggat ak adaptatër?

Xoolaatal laaj bu nekk8 laaj

Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.

  1. Lan mooy dall poid yiñ njëkka tàggat suñuy tàggat ak adaptatër?
    • Dañuy gelé ci diir bi ñuy tàggatee ci parametru adaptëer yi kese
    • Dañu leen efaase
    • Dañu leen di indiwaat ci anam wu mucc
    • Dañu yokk ñaari yoon
  2. Lan mooy jumtukaayu biir bu adaptatëru bottleneck?
    • Benn lalu dense bu rëy
    • Projection buy wàcci, ab non-lignearite, ginaaw ga ab projection buy yéeg, ak ab lëkkaloo bu des
    • Bopp buy bàyyi xel kese
    • Seggal nataal buñ jagleel
  3. Lu tollu ci ban paccu parametre la adaptatër yi di faral di tàggat?
    • 100%
    • 50% ci dëgg
    • Lu bari ci suufu 5%
    • Amul benn
  4. Lan moo waral adaptër yi baax ci liggéey yu bari?
    • Liggéey bu nekk dañu soxla ab kopi model bu mat te wuute
    • Mënu leen boole
    • Dañuy efaase modelu base bi ginaaw liggéey bu nekk
    • Yaa ngi tëye benn model bu njëkk ba noppi nga weccoo adaptatër yu ndaw ci liggéey bu nekk
  5. Ban famiy pexe la adaptatër yi bokk?
    • Reglage bu baax ci paramet (PEFT)
    • Yokkateg done
    • Kompresioŋ nataal
    • Xeetu tokenisasioŋ
  6. Lan mooy njariñu LoRA ci adaptatër buñ dugal ci waxtu inference?
    • Mënu ñu ko tàggat
    • poids yiñ jàng mën nañu leen boole ci matrices yu njëkk yi, moytu latency bu gëna bari ci couche bu nekk
    • Dafay laaj ñu tàggataat model bi yépp
    • Dina dindi lëkkaloo bi des ba fàww
  7. What does a residual connection contribute to a classic bottleneck adapter?
    • A second full pretrained model copy for each task
    • A mandatory external retrieval database
    • A path that adds the adapter’s transformed representation back to the original hidden state
    • A guarantee that no added computation occurs
  8. What is one benefit of stacking a domain adapter with a task adapter?
    • Merge all user data into the frozen base weights automatically
    • Avoid using any adapter modules
    • Convert a language model into a vision encoder
    • Reuse a domain specialization together with a separate task specialization