From edc9711a229940cbaad1f86dede79ac7b469eb45 Mon Sep 17 00:00:00 2001 From: Jan Janssen Date: Sat, 14 Sep 2024 00:32:42 +0200 Subject: [PATCH] Rename lessons learned to summary --- book/energy_volume_curve.ipynb | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/book/energy_volume_curve.ipynb b/book/energy_volume_curve.ipynb index 209ab3d..b95fd5c 100644 --- a/book/energy_volume_curve.ipynb +++ b/book/energy_volume_curve.ipynb @@ -340,7 +340,7 @@ "id": "555d589b-8607-410f-8a37-12aefd2fec59", "metadata": {}, "source": [ - "## Lessons learned\n", + "## Summary\n", "The important point in this example is that at no point the order in which the functions should be executed. Instead the LLM automatically determines that `plot_equation_of_state()` needs an `AtomsDict()` object as input and that `get_equilibirum_lattice()` returns such a `AtomsDict()` object, so it makes sense to call `get_equilibirum_lattice()` first and `plot_equation_of_state()` second. The same principles apply to a LLM agents with a larger number of python functions. \n", "\n", "The limiting point at the moment is that the LLMs are web services, so all Python objects have to be converted to JSON to be communicated to the LLMs. This restricts the choice of Python objects and requires the development of specialized data classes to construct those interfaces between different Python functions. "