Mirostat Perplexity Xakamaynta Kediska

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

1. What does Mirostat try to control during generation?
   • A target level of output perplexity or surprise
   • Tirada qiyaasaha moodeelka
   • The training dataset size
   • Heerkulka GPU
2. Sidee Mirostat asal ahaan uga duwan tahay heerka sare-k ama muunada sare-p?
   • Marna ma isticmaasho randomness
   • Waxay si firfircoon u hagaajisaa goynta iyadoo adeegsanaysa jawaab celin halkii goyn go'an ahayd
   • Waxay kaliya ku shaqeysaa sawirada
   • Waxay u baahan tahay in dib loo tababaro qaabka
3. Mirostat gudaheeda, doorkee ayuu ka ciyaaraa parameter eta?
   • Waxay dejisaa cabbirka erayada
   • Waxay hagaajinaysaa dhererka wax soo saarka
   • Waa heerka waxbarashada xakameynaya sida ugu dhaqsaha badan ee heerku ula qabsanayo
   • Waxay curyaamisaa muunada
4. Waa maxay horumarka ugu weyn ee nooca Mirostat 2 ee ka sarreeya nooca 1?
   • Waxay soo saartaa sawirro sidoo kale
   • Waxay u baahan tahay xasuus badan
   • Waxay meesha ka saaraysaa cabbirka tau
   • Waxay hoos u dhigaysaa malo-awaalka qaybinta Zipfian si ay u fududaato, xakamayn adag
5. Haddii Mirostat uu ogaado in wax-soo-saarka uu noqonayaa mid la saadaalin karo, waxay:
   • Dabci goynta si aad u qirato calaamado badan oo kala duwan
   • Adkee goynta si aad u oggolaato calaamado yar
   • Jooji wax soo saarka isla markiiba
   • Kordhi cabbirka qaabka
6. Which Mirostat settings does current llama.cpp expose?
   • Tokenizer vocabulary and number of layers
   • Mode, target tau, and learning-rate eta
   • Only the prompt length
   • Training epochs and validation loss
7. What does eta control in Mirostat’s feedback update?
   • Cabbirka ereyada tusaalaha
   • The target number of output tokens
   • How strongly the threshold changes in response to the observed error
   • The probability of a correct factual answer
8. What does a Mirostat target not guarantee by itself?
   • That the sampler reads token probabilities
   • That the setting can influence surprise
   • That the output contains tokens
   • That the generated text is factually correct