GEO MCTS by Quante Carlo
Home
How it works
Retrieval as a Bayesian Optimization Problem
Generative Engine Optimization using Monte Carlo Tree Search
Build your own
Contact us
GEO MCTS
by Quante Carlo
This demo uses Monte Carlo Tree Search to rewrite articles to optimize relevance.
Describe the athletic differences between football and baseball.
Optional: you can start with your own article but, the topic is still required.
Depth of rollout:
3
5
8
10
Number of initial trees:
3
5
8
10
How it works
Write:
Writes stories based on your topic and then looks for a score
Performs a roll out: sequentially rewrites the story.
Scores eaach story based on Cohere's
reranker
Uses Bayesian Optimization based on the embeddings of the stories to estimate the best place for a new node.
Performs rollouts in parallel
Rewrite:
Same as before but instead of writing from scratch, it starts by rewriting the story you provide.