Beta. Content is under active construction and has not been peer-reviewed. Report errors on
GitHub
.
Disclaimer
Theorem
Path
Curriculum
Paths
Demos
Diagnostic
Search
Quiz Hub
/
Decoding Strategies
Decoding Strategies
3 questions
Difficulty 3-5
View topic
Foundation
0 / 3
2 foundation
1 intermediate
Adapts to your performance
1 / 3
foundation (3/10)
compare
Greedy decoding picks the highest-probability next token at each step; beam search keeps the top-
k
hypotheses. When does beam search beat greedy?
Hide and think first
A.
Beam search handles out-of-vocabulary tokens more gracefully than greedy decoding by default
B.
Beam search is strictly better than greedy for all generation tasks, in both quality and computational cost
C.
When locally-optimal token choices lead to globally-suboptimal sequences, beam search's lookahead finds better overall sequences
D.
Beam search is required for models with more than 1 billion parameters; greedy only works for smaller models
Submit Answer