Muenzaniso Quantization Quiz

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

1. Quantizing a model mainly reduces what?
   • Iyo nhamba chaiyo inoshandiswa kuchengetedza huremu uye activation
   • Huwandu hwema layer mune network
   • Iyo dataset yekudzidzisa iyo modhi yakavakirwa pairi
   • Mutauro uyo muenzaniso unogona kunzwisisa
2. Ignoring scale metadata and other runtime tensors, what is the weight-payload ratio from sixteen bits to four bits per weight?
   • The payload is unchanged
   • It is one hundredth of the original payload
   • The four-bit payload is one quarter of the sixteen-bit payload
   • The payload becomes four times larger
3. Chii chinonyanya kuderera chehasha-bit quantization?
   • Iyo modhi inokanganwa mutauro wayo wekudzidzisa
   • Kugona kurasikirwa kwechokwadi sezvo kukosha kunosvinwa mumabhaketi mashoma
   • Inogona chete kumhanya pamakomputa makuru
   • Iyo inobvisa zvachose maseru
4. Ko quantization-aware training inosiyana sei nepost-training quantization?
   • Izvo zvinoda hapana data zvachose
   • Inoshanda chete pamifananidzo
   • Inowedzera nhamba yebhiti nekufamba kwenguva
   • Inotevedzera kutenderera panguva yekudzidziswa kuitira kuti modhi idzidze kushivirira kushomeka chaiko
5. Ndeipi nzira inoshandisa 4-bit quantization kuita zvakanaka-tuning hombe mhando dzinokwanisika paGPU imwe chete?
   • Dropout
   • Batch normalization
   • QLoRA
   • Kukamura Mazwi
6. Sei pa-channel kana pa-boka-chemwe kuyera ichibatsira mamodheru akaenzana?
   • Inopa zvakaparadzana zviyero zvimedu zvehuremu matrix kuchengetedza iko chaiko
   • Inobvisa kudiwa kwehuremu hwese
   • Inoita kuti modhi ive yakakura
   • Inoshandura zvinyorwa kuita audio
7. A four-bit model’s raw weights fit in available VRAM, but serving fails out of memory. What could explain this?
   • Every four-bit weight expands permanently to a gigabyte
   • Quantization guarantees no runtime allocations
   • Only the model’s filename consumes the remaining memory
   • Caches, scales, unquantized tensors, or workspace need additional memory
8. A team quantizes to INT8 but sees no inference speedup. What should it check?
   • Whether the training data are alphabetized
   • Whether the workload uses efficient supported kernels on its hardware
   • Whether lower bit width always guarantees faster execution
   • Whether the model must have changed languages