Develop an LLM agent which can predict the crystal structure of a chemical element, by accessing the reference database of the Atomic Simulation Environment. While the corresponding python function is simple and only requires a few lines of code, this example already limits the hallucinations of the LLM, by using the langchain framework to interface a python function with the LLM.
-In particular we follow the Custom Agent tutorial from the Langchain documentation.
+As a first demonstration the langchain team decided to develop a LLM agent, which can predict the crystal structure of a chemical element, by accessing the reference database of the Atomic Simulation Environment. While the corresponding python function is simple and only requires a few lines of code, this example already limits the hallucinations of the LLM, by using the langchain framework to interface the LLM with the python function.
+In particular, we follow the Custom Agent tutorial from the Langchain documentation.
For this first example we use OpenAI as LLM provider but the example can also be adjusted to work with other LLM providers. We store the OpenAI API key in the OPENAI_API_KEY variable:
For this first example, we use OpenAI as LLM provider but the example can also be adjusted to work with other LLM providers, for more details check the langchain documentation. We store the OpenAI API key in the OPENAI_API_KEY variable:
As a next step we import the corresponding functionality from ASE and the tool decorator from langchain:
As a next step, we import the corresponding functionality from ASE and the tool decorator from langchain:
from ase.data import reference_states, atomic_numbers
@@ -408,7 +408,7 @@ Python Functiontype hints and documentation based on a Docstring for the LLM to understand the functionality of the function. Finally, all data types used as input or output of the function need to have a JSON representation so they can be communicated to the LLM. For example numpy arrays have to be converted to standard python lists.
+For the python function, it is important to include type hints and documentation based on a Docstring for the LLM to understand the functionality of the function. Finally, all data types used as input or output of the function need to have a JSON representation so they can be communicated to the LLM. For example, numpy arrays have to be converted to standard python lists.
@tool
@@ -451,7 +451,7 @@ Python Function
Define Agent#
-After the definition of the Python function, the next step is the definition of the agent which the LLM uses to interact with the Python function. In this example the ChatOpenAI interface of the langchain_openai package is used. Depending on your configuration it might be necessary to install this using:
+After the definition of the Python function, the next step is the definition of the agent which the LLM uses to interact with the Python function. In this example the ChatOpenAI interface of the langchain_openai package is used. Depending on your configuration, it might be necessary to install the langchain_openai package using the following command:
conda install -c conda-forge langchain-openai
@@ -536,11 +536,11 @@ Conversationverbose=True parameter the internal steps of the LLM agent are printed in green. As a first step the Agent calls the get_crystal_structure() already with the converted input parameter, rather than using gold as input it uses the chemical symbol Au. The function returns fcc and the LLM converts this answer in a sentence a human can understand:
+With the verbose=True parameter the internal steps of the LLM agent are printed in green. As a first step the agent calls the get_crystal_structure() already with the converted input parameter, rather than using gold as input it uses the chemical symbol Au. The function returns fcc and the LLM converts this answer in a sentence a human can understand:
The crystal structure of gold is face-centered cubic (fcc).
-This example highlights how easy it is these days to make a python function accessible via an LLM for all kinds of users to interact with this python funtion.
+This example highlights how easy it is these days to make a python function accessible via a LLM for all kinds of users to interact with this python funtion.
Prompt Engineering#
@@ -618,11 +618,11 @@ Prompt Engineering
Summary#
-By following the Custom Agent tutorial from the Langchain documentation we were able to create a first simulation agent, which calls specialized python frameworks like ASE to address material-science specific questions. Still it is important to carefully engineer the prompt of the agent, otherwise even these LLMs with access to specialized agents tend to hallucinat.
+By following the Custom Agent tutorial from the Langchain documentation, we were able to create a first simulation agent, which calls specialized python frameworks like ASE to address material-science specific questions. Still, it is important to carefully engineer the prompt of the agent, otherwise even these LLMs with access to specialized agents tend to hallucinate.
@@ -666,7 +666,7 @@ Summary#<