Bird
0
0

How does combining temperature scaling with top-k sampling improve text generation?

hard📝 Application Q9 of 15
NLP - Text Generation
How does combining temperature scaling with top-k sampling improve text generation?
ATemperature and top-k have no combined effect
BTemperature selects top-k words; top-k scales probabilities
CBoth methods randomly select words ignoring probabilities
DTemperature smooths probabilities; top-k limits choices to likely words
Step-by-Step Solution
Solution:

  1. Step 1: Understand temperature scaling

    Temperature adjusts probability sharpness, controlling randomness.

  2. Step 2: Understand top-k sampling

    Top-k restricts sampling to top k probable words, avoiding rare words.

  3. Step 3: Combined effect

    Temperature smooths probabilities; top-k limits choices, improving quality and diversity.

  4. Final Answer:

    Temperature smooths probabilities; top-k limits choices to likely words -> Option D

  5. Quick Check:

    Temperature + top-k = smoother, focused sampling [OK]
Quick Trick: Temperature smooths; top-k filters choices [OK]
Common Mistakes:
MISTAKES
  • Confusing roles of temperature and top-k
  • Thinking both ignore probabilities
  • Assuming no combined effect

Want More Practice?

15+ quiz questions · All difficulty levels · Free

Free Signup - Practice All Questions
More NLP Quizzes
Sequence Models for NLP - Attention mechanism basics - Quiz 4medium Sequence Models for NLP - Embedding layer usage - Quiz 2easy Text Generation - Beam search decoding - Quiz 12easy Text Similarity and Search - Semantic similarity with embeddings - Quiz 10hard Topic Modeling - Latent Dirichlet Allocation (LDA) - Quiz 6medium Topic Modeling - Choosing number of topics - Quiz 5medium Topic Modeling - LDA with Gensim - Quiz 11easy Topic Modeling - Topic coherence evaluation - Quiz 12easy Word Embeddings - GloVe embeddings - Quiz 13medium Word Embeddings - FastText embeddings - Quiz 4medium