​
Tutorial Overview

The next step in our tutorial is connecting our Nodes together to build a DAG.

​
Building a DAG

All we need to do is instantiate an instance of a DAG, add instances of our Nodes, and add edges between them.

​
Instantiate the DAG

First, we need to import the DAG class.

example_dag.py
from trellis_dag import DAG

Then, we can instantiate a DAG.

example_dag.py
dag = DAG()

DAG takes no arguments. We’ll set everything up through function calls to it.

​
Import our cat fact tool

We need to import our CatFactsAPITool class from the previous tutorial.

example_dag.py
from cat_fact_tool import CatFactsAPITool

cat_fact_tool = CatFactsAPITool("cat_fact_tool")

​
Add Nodes

Next, we’ll add our Nodes using the add_node function.

example_dag.py
dag.add_node(cat_fact_tool)
dag.add_node(generate_cat_fact_llm)
dag.add_node(distinguish_cat_fact_llm)

​
Add Edges

Finally, we’ll add edges between our Nodes using the add_edge function.

example_dag.py
dag.add_edge(cat_fact_tool, distinguish_cat_fact_llm)

The cat_fact_tool -> distinguish_cat_fact_llm edge is easy, because we already changed the output name of cat_fact_tool to cat_fact.

example_dag.py
dag.add_edge(
    generate_cat_fact_llm,
    distinguish_cat_fact_llm,
    fn=lambda x: {"cat_fact_2": x["choices"][0]["message"]["content"]},
)

The generate_cat_fact_llm -> distinguish_cat_fact_llm edge is a little more complicated. We need to specify a function that transforms and renames the output of the generate_cat_fact_llm into the input of the distinguish_cat_fact_llm. In this case, we’ll just take the first choice and rename it to cat_fact_2.

Every edge in Trellis allows you to optionally transform data between nodes when you add an edge. We designed it this way to reduce the amount of code you’d have to rewrite.

​
Putting it all together

That’s it! We’ve put our DAG together. In the next and final tutorial, we’ll run it. Visit the DAG reference to learn more. Here’s the full code for this tutorial.

example_dag.py
from cat_fact_tool import CatFactsAPITool
from trellis_dag import DAG

cat_fact_tool = CatFactsAPITool("cat_fact_tool")

dag = DAG()
dag.add_node(cat_fact_tool)
dag.add_node(generate_cat_fact_llm)
dag.add_node(distinguish_cat_fact_llm)

dag.add_edge(cat_fact_tool, distinguish_cat_fact_llm)
dag.add_edge(
    generate_cat_fact_llm,
    distinguish_cat_fact_llm,
    fn=lambda x: {"cat_fact_2": x["choices"][0]["message"]["content"]},
)
Instantiate LLM NodesRun your DAG