quiz bu lëkkale ak tegtal · Diggante Niveau

Modelu làkku filigran

Saytu limu filigran yu wert yi, lim yiñ gis, seeding caabi sekkere yi, àppu dëgëraay yi, njaayum kalite mbind yi ak SynthID-Text.

Yooni njiit yu jëm ciModelu làkku filigran
Laaj 1 de 8

Ci filigran bu am limu wert/limu xonk, naka lañuy dugal siñaal bi?

Xoolaatal laaj bu nekk8 laaj

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

  1. Ci filigran bu am limu wert/limu xonk, naka lañuy dugal siñaal bi?
    • Ni ñuy jëlee sampling token ci limu wert buñ tànn ci nëbbëtu
    • Sooy dugal ay arafu Unicode yuñudul gis ci diggante kàddu yi
    • Sooy yokk footer bu nëbbu ci mbind mi
    • Ni ñuy encrypte lépp luy génn
  2. Lan moo waral bind bu am watermark di gëna jafe gis sudee dafa gàtt lool?
    • Bind bu gàtt dafay jëfandikoo vocabulaire bu wuute
    • Surplus token bu wert bi tuuti lool ngir nekk lu am solo ci wàllu lim
    • Detektër yi duñu bàyyi xel ci mbind yu néew 10 baat ci sàrt yi
    • Mbind mu gàtt nit ñi ñoo ko bind
  3. Lan mooy wane ndax token bi dafay wàcci ci limu wert wala xonk?
    • Alfabe bu token bi
    • Adres IP jëfandikukat bi
    • Hash bu token yi njëkk boole ci caabi sekkere
    • Waxtu ci bis bi ñu defaree mbind mi
  4. Ban jëf moo gëna mëna dindi wala néewal doole filigran bind?
    • Liggéey mbind mi ak baat bu kawe
    • Sanc fichier bi ci beneen formaa
    • Soppi font bi
    • Tekki làkk bu diis
  5. Lan mooy kompromis bi gëna am solo sooy yokk dooley green-token bias?
    • Dafay yokk dëgëraay bi waaye mën na wàññi kalite mbind mi ak bariwaayu mbind mi
    • Dafay tax gis gi gëna yeex waaye bind bi gëna gaaw
    • Dafay gëna baaxal gis-gis bi ak kalite bi te doo fay
    • Dina dindi soxla caabi sekkere
  6. Ban sistem ci àdduna bi moo indi watermarking modelu làkk ci liggéey bi?
    • Adobe fotoshop
    • _AIU_AAR_11__ Bindu ID-Synth bu DeepMind
    • ab téere bu mag
    • Done DNS
  7. As a watermarked passage gets longer, how does the detector’s green-token z-statistic typically scale?
    • Linearly with the number of model parameters
    • As the inverse square root of the passage length
    • Roughly with the square root of the number of tokens
    • It remains exactly constant
  8. Why is the secret watermark key not embedded directly in the generated text?
    • The vocabulary has no token IDs
    • Embedding a key would disable tokenization
    • The model cannot store character strings
    • The signal is encoded in token choices, while keeping the key secret prevents others from easily reproducing or forging the split