Scaling Laws for Neural Networks Quiz

Answers and explanations appear as you take the quiz.

1. What pattern did Kaplan et al. report for language-model loss as model size, data, or compute increased?
   • Approximate power-law relationships over the tested ranges
   • A sudden guaranteed drop to zero
   • A fixed increase in loss for every larger model
   • Random oscillation with no measurable trend
2. Which three quantities do the original scaling laws relate to loss?
   • Batch size, dropout, and epochs
   • Model parameters, dataset size, and compute
   • Learning rate, weight decay, and depth
   • GPU count, memory, and latency
3. How can a scaling-law fit help when planning a larger training run?
   • It guarantees the final benchmark score
   • It eliminates the need for validation
   • It can inform an extrapolation from smaller experiments, subject to the tested setup
   • It predicts results exactly for any dataset or architecture
4. What does the constant (irreducible) term in a scaling law loss formula represent?
   • A software bug
   • The learning rate
   • The number of GPUs
   • The intrinsic entropy of the data that no model can beat
5. Why can absolute loss reductions become smaller as compute keeps increasing under a power-law fit?
   • The reducible loss term shrinks while the loss approaches its floor
   • The data is removed after each compute increase
   • The model stops using parameters
   • Power laws require loss to increase over time
6. Which trend is pushing the field beyond simple 'bigger is better' scaling?
   • GPUs becoming free
   • Scarcity of high-quality data plus energy and cost limits
   • Unlimited high-quality text appearing
   • Loss curves turning into straight vertical lines
7. What did the original Kaplan study say about architecture details such as width and depth within its tested range?
   • They always dominated model size, data, and compute
   • They made loss independent of data
   • They had relatively small effects compared with the studied scaling variables
   • They guaranteed the same performance for all architectures
8. What compute-allocation tendency did the original Kaplan paper report?
   • Every model should be trained to zero loss
   • Small models always use less compute at equal loss
   • Training data has no effect on compute allocation
   • Large models could be more sample-efficient and compute-efficient training could stop before convergence