Guide-yakabatana mibvunzo · Zvakaoma Nhanho
Kusimbisa Zvakanaka-Tuning neGraders Quiz
Miedzo kuti magiredhi anokora sei mamodhi anobuda, ayo mabasa anoenderana nekusimbisa-tuning, uye kuti mibairo yekubira uye isina simba masaini ekudzidzisa anomuka sei.
Mubvunzo 1 ye 8
Ko yekusimbisa-tuning inosiyana sei neinotarisirwa tuning?
Tarisa mubvunzo wese8 mibvunzo
Mhinduro uye tsananguro zvinoonekwa paunenge uchitora bvunzo.
- Ko yekusimbisa-tuning inosiyana sei neinotarisirwa tuning?
- 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
- Ndeipi mhando yegiredhi yaungashandisa kutarisa kuti inogadzirwa kodhi inoshanda here?
- 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
- Ibasa ripi iro rakashata zvakanyanya kuRFT zvinoenderana negwara?
- 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
- Mukusimbisa Kwakanaka-Tuning neGraders: chii mubairo kubira?
- 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
- Sei kukurudzira uko sampuli yega yega inowana zvibodzwa zvakafanana zvichipa chiratidzo chidiki chekudzidza munzira dzeboka senge 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
- Iyo modhi inomira kuburitsa
- 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