Beam Search
Keeping several candidate sequences alive and extending the best.
Overview
Beam search expands several partial sequences at once and keeps the highest-scoring ones, rather than committing to the token chosen at each step. It solves the problem that greedy decoding can be locally optimal and globally poor. It exists as the default for translation and summarisation, and it tends to produce bland text for open-ended generation.
Technology Graph
How Beam Search connects to the rest of the stack.
Alternatives
-
Top-p Sampling
technique
An alternative to Beam Search · source
Comparisons
Side-by-side breakdowns with the alternatives above.