Yakareba-Context Modelling Quiz

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

1. Chii chinonzi 'context window' yemuenzaniso?
   • Huwandu hwema tokens hwaanogona kuita mune imwe pass
   • Saizi yescreen yechat interface
   • Nhamba yevashandisi inobvumirwa kamwechete
   • Inguva yakareba sei iyo modhi yakadzidziswa
2. Sei kuzvitarisira kwakajairwa kuchidhura kune zvigadzirwa zvenguva refu?
   • Inokura mutsetse uye inogara yakachipa
   • Iyo compute uye ndangariro inokura quadratically nekutevedzana kureba
   • Haikwanise kuverenga zvinopfuura chiratidzo chimwe
   • Inoshanda chete pamifananidzo
3. Dambudziko re'kurasika pakati' nderei?
   • Models vanokanganwa kudzidziswa kwavo zvachose
   • Mienzaniso inogona kuverenga pakati pegwaro chete
   • Mienzaniso inorangarira ruzivo pakutanga uye kumagumo kweiyo refu yekuisa zviri nani pane chokwadi chakavigwa pakati
   • Mienzaniso inoshandura yepakati ichienda kune mumwe mutauro
4. Chii chinonzi FlashAttention inonyanya kuvandudza?
   • Inoita kuti modhi inobuda irebe mavara
   • Iyo inowedzera imwe data yekudzidziswa otomatiki
   • Inobvisa kudiwa kwekutariswa kupi zvako
   • Iyo inoderedza ndangariro bhodhoro neIO-inoziva, mataira computation
5. Ibasa rei rinoitwa nerotary position embeddings (RoPE) mumhando dzenguva refu?
   • Ivo vanokodha zvinzvimbo zvetokeni uye, nekududzira, vanobatsira kuita kutevedzana kwakareba
   • Vanodzima ma tokens kubva pakuiswa
   • Vanoshandura mavara pakati pemitauro
   • Vanochengeta zviyero zvemuenzaniso
6. What does prompt-cache research reuse across compatible requests?
   • The final answer regardless of the new query
   • Computed attention state for reusable prompt segments
   • A fixed set of token IDs without model computation
   • The model’s training data
7. What does the RULER benchmark add beyond locating one isolated “needle” fact?
   • Kuiswa kwemifananidzo chete
   • Audio transcription from waveforms
   • Tasks such as multi-hop tracing and aggregation over long contexts
   • GPU memory allocation benchmarks
8. Why test several task types when measuring a model’s effective context length?
   • Every task uses the same computation
   • The advertised window guarantees all tasks work
   • Task type never affects context use
   • A model may handle simple retrieval yet fail to combine or track information across a long input