Maswali yanayounganishwa na mwongozo · Ngumu Kiwango

Maswali ya Kusimbua Hatari ya Chini ya Bayes

Majaribio ya matumizi yanayotarajiwa, sampuli za marejeleo ya uwongo, ulinganisho wa watahiniwa wa quadratic, vikomo vya metri, na vizuizi vya pamoja vya usimbaji wa MBR.

Njia za mwongozo zinazohusianaUsimbuaji wa Kiwango cha Chini cha Hatari ya Bayes
Swali 1 ya 8

Je, kusimbua kwa MBR huchagua nini badala ya mlolongo mmoja unaowezekana zaidi?

Hakiki kila swali8 maswali

Majibu na maelezo huonekana unapojibu maswali.

  1. Je, kusimbua kwa MBR huchagua nini badala ya mlolongo mmoja unaowezekana zaidi?
    • Mtahiniwa aliye na matumizi bora zaidi yanayotarajiwa dhidi ya matokeo mengine yanayowezekana
    • Mlolongo mfupi zaidi
    • Sampuli ya nasibu
    • Mlolongo mrefu zaidi
  2. Kwa kuwa marejeleo ya kweli hayapatikani, MBR hutumia nini kama marejeleo ya uwongo?
    • Shirika la nje lisilobadilika
    • Mkusanyiko wa matokeo ya sampuli ya wagombea wenyewe
    • Maelezo ya kibinadamu
    • Seti ya mafunzo
  3. Kwa nini MBR isiyo na maana ni ghali kikokotoa?
    • Inahitaji kufundisha tena mfano
    • Inahitaji msamiati mkubwa zaidi
    • Ulinganisho wa jozi wa watahiniwa wa N ni simu za kipimo cha O(N za mraba).
    • Haitumii GPU
  4. What has COMET been used for in minimum Bayes risk machine-translation experiments?
    • As a universal guarantee of better translation quality
    • As a tokenizer for the candidate pool
    • As an exact reference translation available at inference
    • As a learned utility metric in evaluated translation settings
  5. Kujitosheleza katika hoja kunahusiana na MBR kwa sababu:
    • Sampuli za minyororo mingi na huchagua jibu lililokubaliwa zaidi
    • Daima huchukua jibu la kwanza la sampuli
    • Hutumia utafutaji wa boriti ndani
    • Hupuuza usambazaji wa muundo
  6. How is risk generally expressed in minimum Bayes risk decision-making?
    • Always one minus a normalized similarity score
    • Expected loss under a chosen loss function and reference distribution
    • The model’s parameter count divided by accuracy
    • The number of candidates sampled
  7. If the chosen utility metric disagrees with human judgments, what will MBR primarily optimize?
    • Human preference automatically overrides the metric
    • The model’s raw next-token likelihood
    • The expected utility under that selected metric
    • The shortest candidate in the pool
  8. If a high-quality answer is missing from the sampled candidate pool, what can MBR do?
    • It guarantees the missing answer is generated
    • It changes the model weights to add the answer
    • It replaces the utility metric with a human reference
    • It can only select or rerank candidates that are in the pool