Kedis ku xidhan hagaha · Dhexdhexaad Heerka

Nooca degdega ah iyo Kediska Imtixaanka Dib-u-celinta

Wuxuu tijaabiyaa waxa ku jira qaabaynta degdega ah ee la habeeyey, sida qolalka dib-u-celinta u maareeyaan go'aan la'aan, iyo sababta moodeelka sawir-qaadista iyo garsoorayaasha LLM ugu baahan yihiin dhejis.

Wadooyinka haga ee la xidhiidhaTijaabinta Soo Celinta Degdegga ah iyo Dib-u-celinta
Su'aal 1 ee 8

Which set of items should be versioned together as one prompt configuration?

Horudhac su'aal kasta8 su'aalo

Jawaabaha iyo sharraxaadaha waxay soo baxayaan markaad qaadanayso kediska.

  1. Which set of items should be versioned together as one prompt configuration?
    • Only the prompt wording
    • Kaliya magaca modelka
    • Prompt template, model identifier, sampling parameters and tool schemas
    • Only the user inputs from production
  2. Why pin a dated model snapshot instead of a generic latest alias?
    • So model changes happen only when you deliberately switch and re-test
    • Because aliases are always slower
    • Because snapshots are free
    • Because aliases do not support system prompts
  3. What does the guide say about temperature 0?
    • It guarantees identical outputs every time
    • It makes regression testing unnecessary
    • It only affects the length of outputs
    • It reduces randomness but does not guarantee identical outputs across runs
  4. How should a regression suite handle nondeterministic outputs?
    • Require exact string matches
    • Use thresholds and repeated samples instead of exact equality
    • Ignore any failing cases
    • Only test with a single example
  5. What is a strong source of new test cases for a golden set?
    • Past production failures turned into test cases
    • Random text from the web
    • Only cases the current prompt already passes
    • Outputs copied from a competitor
  6. If an LLM judge grades outputs, what should you do?
    • Let the judge model change freely
    • Trust every verdict without review
    • Pin the judge's model and prompt and check it against human labels
    • Use the same prompt under test as the judge
  7. Why log the prompt version ID with each production request?
    • To reduce token costs
    • To improve model accuracy
    • To satisfy the model provider's terms
    • To tie problems to a specific deploy and roll back quickly
  8. Why list individual cases that flipped from pass to fail rather than only the average score?
    • Averages are always wrong
    • An unchanged average can hide new breakage offset by gains elsewhere
    • Flipped cases reduce API costs
    • Only flipped cases are graded