From b849e6c92ce3314ca680525798a4779ac2750af0 Mon Sep 17 00:00:00 2001 From: chinganc Date: Tue, 25 Jun 2024 10:44:07 -0700 Subject: [PATCH] update --- docs/index.html | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/docs/index.html b/docs/index.html index 1999bdf..4e0d611 100644 --- a/docs/index.html +++ b/docs/index.html @@ -506,7 +506,7 @@

Trace Graph

Showcases

-

We can extend the same idea of end-to-end optimization to train more complicated AI systems. Empirical studies showcase Trace's ability to optimize diverse problems, from numerical optimization, LLM agents, to robot control, often outperforming specialized optimizers.

+

We extend the same idea of end-to-end optimization seen in the Battleship example to train more complicated AI systems. Empirical studies showcase Trace's ability to use a single LLM-based optimizer (OptoPrime) to solve diverse problems, from numerical optimization, LLM agents, to robot control, often outperforming specialized optimizers. In these experiments, each iteration makes just one call to an LLM (GPT-4) to optimize graphs of tens of nodes.