Star Trek III: The Search for Spock

How do we search for the best sequence in a sea of possibilities?

Imagine you're trying to find the perfect recipe for a new dish. You're experimenting with different ingredients and cooking times, but you want to avoid wasting time on bad combinations.

Beam search is like trying out recipes. You keep picking the best ones until you can't find any better, stopping before you exhaust all options.

Example

You try 5 recipes, then 3 more, and finally 2 more after noticing the first 8 aren't improving your dish.

Remember this

Beam search stops exploring once you can't find a better sequence, saving time and effort.

Text adapted from Wikipedia, licensed under CC BY-SA 4.0.

Related concepts

Greedy vs beam search decoding: greedy picks best token, beam maintains k candidates

Ever wondered why Google Search sometimes shows you the top results first?

Hierarchical navigable small world

HNSW is an efficient ANN search algorithm

weight tying does in language models: shares embedding and output projection matrices

Ever wonder how machines understand the sequence of words in a sentence?

Overlapping subproblems

Ever calculated a huge Fibonacci sequence by hand?

the Y combinator does: enables recursion in languages without named functions

Can a computer program call itself to solve problems?

batch size affects generalization: larger batches find sharper minima

Larger batch sizes lead to sharper minima, enhancing generalization by providing more accurate gradient estimates

Swipe through 100 ML concepts daily

Open Pocket Polymath