Guide-yakabatana mibvunzo · Pakati Nhanho

Odds Ratio Preference Optimization Quiz

Inotsanangura ORPO's referensi-modhi-yemahara yekuda chinangwa, odds-ratio chirango, uye yakasanganiswa inotariswa uye yekuda kudzidziswa.

Nzira dzekutungamira dzinoenderanaOdds Ratio Preference Optimization
Mubvunzo 1 ye 8

Ndeipi yakakosha mukana weORPO pamusoro penzira dzakaita seDPO?

Tarisa mubvunzo wese8 mibvunzo

Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.

  1. Ndeipi yakakosha mukana weORPO pamusoro penzira dzakaita seDPO?
    • Inobvisa kudiwa kweimwe referensi modhi
    • Izvo zvinoda hapana data rekudzidziswa zvachose
    • Inoshanda chete pamifananidzo yemifananidzo
    • Inovimbisa zero hallucinations
  2. Chii chinofananidzwa neiyo 'odds ratio' muORPO?
    • Kumhanya kwekudzidzira maringe nekururama
    • Kusiyana kwemhinduro yakasarudzwa maringe nemhinduro yarambwa
    • Zviyero zviviri zvakasiyana zvekudzidza
    • Saizi yemuenzaniso maringe nesaizi yedataset
  3. What form does an ORPO preference example contain?
    • Only a single unlabeled response
    • A prompt with two identical completions
    • A prompt with one chosen response and one disfavored response
    • A reward score with no associated responses
  4. How does ORPO convert a response probability p into odds?
    • One minus p divided by p
    • The square root of p
    • The logarithm of the prompt length
    • p divided by one minus p
  5. Iko kurasikirwa kweORPO kunotsanangurwa zvakanyanya sekuti ndeipi musanganiswa?
    • SFT muchinjiko-entropy pamwe neinorema odds-ratio yekuda nguva
    • Chete mubairo-muenzaniso zvibodzwa
    • Kudzoka kwakachena kwekusimbisa-kudzidza
    • Random ruzha jekiseni
  6. Which response receives the supervised fine-tuning component of ORPO?
    • Only the rejected response
    • The chosen or preferred response
    • A response sampled after deployment
    • No response; ORPO has no language-modeling term
  7. Which parameter controls the weight of ORPO’s preference term relative to the supervised loss?
    • Saizi yemazwi tokenizer
    • Nhamba yekutarisisa misoro
    • The preference-loss coefficient
    • The prompt’s character count
  8. Why should an ORPO result report its preference data and evaluation setup?
    • ORPO results are identical across datasets
    • The objective removes the need for evaluation
    • The number of model layers determines every preference
    • Outcomes depend on the dataset, coefficient, base model, and evaluation