quiz bu lëkkale ak tegtal · Dëgër Niveau
Reglementation bu baax ak Quiz Graders
Dafay saytu ni grader yi di joxee poñ ci model yi, ban liggéey moo méngoo ak fine-tuning bi, ak ni pirate neexal ak siñaal tàggat yaram yu néew doole di feeñee.
Laaj 1 de 8
Lan mooy wuutale ajustement bu am doole ak ajustement buñ saytu?
Xoolaatal laaj bu nekk8 laaj
Sooy def quiz bi, tontu yi ak leeral yi dañuy feeñ.
- Lan mooy wuutale ajustement bu am doole ak ajustement buñ saytu?
- It trains the model to imitate exact target outputs
- It rewards sampled outputs a grader scores highly instead of copying examples
- It needs no training prompts
- It only changes the system prompt
- Ban xeetu grader ngay jëfandikoo ngir xool ndax kode biñ defar dafay dox?
- A programmatic grader that runs unit tests
- A string-match grader comparing to one reference function
- A grader that counts words
- No grader is needed for code
- Ban liggéey moo gëna ñakk njariñ ci RFT buñu sukkandikoo ci tegtal bi?
- Classifying items into defined category codes
- Solving maths problems with known answers
- Judging subjective quality where expert graders disagree
- Structured data extraction with a reference answer
- Ci Fine-Tuning ak Graders: luy piratage neexal?
- Stealing another team's grader
- The model learning to satisfy the grader rather than the real intent
- Using too little compute during training
- A grader giving partial credit
- Lan moo waral laaj yi nga xamni misaal bu nekk am benn poñ, duñu joxe siñaal bu bari ci njàng ci pexe yu jëm ci kuréel yu melni GRPO?
- Because the grader crashes on them
- Because learning comes from score differences within the group
- Because they are too long
- Because GRPO ignores scores entirely
- What is the purpose of a KL penalty against the original model during RFT?
- To speed up grading
- To limit how far the model drifts from its starting behaviour
- To increase output length
- To replace the grader
- A model grader over-rewards confident, lengthy replies. What is the likely outcome of training against it?
- Outputs drift toward longer, more confident wording
- Modèle bi dafay taxawal génne
- Accuracy is guaranteed to improve
- The grader becomes stricter automatically
- Which is the best way to check that rising training reward reflects real improvement?
- Trust the training reward curve
- Increase the learning rate
- Spot-check outputs by hand or with an independent grader on held-out data
- Remove the validation set