Calaamadaynta iyo Kediska Codaynta Labada Byte

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. Maxay calaamadeyntu u soo saartaa qaabka luqadda si loo akhriyo?
   • Calaamadaha, kuwaas oo ah aqoonsiyo wada jir ah oo lagu sameeyay qaybo qoraal ah
   • Qaababka mowjadaha codka qaydhiin
   • Xargaha Pixel
   • Koodhka mashiinka la soo ururiyey
2. Sidee bay Bayte Pair Encoding u dhistaa ereybixinteeda?
   • Adigoo ku meeleynaya hal calaamad eray kasta oo dhan qaamuuska
   • Iyaga oo si isdaba joog ah isugu daraya calaamadaha ugu badan ee isku xiga
   • Iyadoo si aan kala sooc lahayn loo kooxeeyay jilayaasha
   • Adigoo u turjumaya qoraalka luqad kale marka hore
3. Dhibaato noocee ah ayay hababka kelmad-hoosaadka sida BPE xalliyaan marka la barbar dhigo calaamadaynta heerka erayga?
   • Waxay jumlad kasta ka dhigaan hal calaamad
   • Waxay meesha ka saaraan baahida eray kasta
   • Waxay qabtaan kelmado naadir ah ama aan la arkin iyagoo u kala qaybinaya qaybo dib loo isticmaali karo
   • Waxay u shaqeeyaan Ingiriisiga oo keliya
4. Waa maxay faa'iidada heerka-byte BPE, sida loo isticmaalo moodooyinka GPT?
   • Waxay u isticmaashaa calaamado ka yar hababka kale ee dhammaan luqadaha
   • Waxay meesha ka saaraysaa baahida loo qabo in lagu tababaro qaabka
   • Waxa kaliya oo ay dejisaa xarfaha Ingiriisiga
   • Waxay matali kartaa qoraal kasta oo Unicode ah, oo ay ku jiraan emoji iyo luqad kasta, iyada oo aan lahayn guul-darrooyin ka baxsan erayada
5. Maxay tahay sababta tirinta calaamada tusaalaha ay inta badan uga duwan tahay tirada erayada jumlada?
   • Meelaha bannaan, weyneynta iyo erayada naadirka ah waxay beddelaan sida qoraalku u kala qaybsamo, hal kelmadna wuxuu noqon karaa calaamado dhawr ah
   • Calaamadaha waxaa lagu tiriyaa luqad aan toos ahayn
   • Erey kastaa had iyo jeer waa laba calaamadood
   • Tirooyinka calaamaduhu waxay kaliya ku xiran yihiin xarakaynta
6. Goobtee ayaa markii hore laga soo amaahday Algoorithm-ka Isku-dhafka Byte?
   • Samaynta sawirka
   • Cadaadiska xogta
   • Tilmaanta database-ka
   • Cryptography
7. After BPE training creates its vocabulary, what does an encoder use to tokenize new text?
   • A new merge ranking learned separately for each sentence
   • A dictionary of whole words only
   • The learned vocabulary and ordered merge rules applied to the input symbols
   • A translation table that changes the text language
8. Why can byte-level BPE encode a character it never saw as a whole token during training?
   • The tokenizer invents new model parameters for each character
   • It drops every unseen character
   • The model automatically translates it to English first
   • The UTF-8 input can be represented as a sequence of bytes from a finite base alphabet