Customize a Workflow#

Prerequisites#

  1. Set up your environment by following the instructions in the Install From Source section of the install guide.

  2. Install NVIDIA NeMo Agent toolkit and the Simple example workflow.

    uv pip install -e .
uv pip install -e examples/getting_started/simple_web_query

This tutorial assumes familiarity with workflows and the command line interface.

Customizing the examples/getting_started/simple_web_query Workflow#

The examples/getting_started/simple_web_query workflow is defined by the examples/getting_started/simple_web_query/configs/config.yml configuration file, which you can examine in the configuration file contents.

examples/getting_started/simple_web_query/configs/config.yml:

functions:
  webpage_query:
    _type: webpage_query
    webpage_url: https://docs.smith.langchain.com
    description: "Search for information about LangSmith. For any questions about LangSmith, you must use this tool!"
    embedder_name: nv-embedqa-e5-v5
    chunk_size: 512
  current_datetime:
    _type: current_datetime

llms:
  nim_llm:
    _type: nim
    model_name: meta/llama-3.1-70b-instruct
    temperature: 0.0

embedders:
  nv-embedqa-e5-v5:
    _type: nim
    model_name: nvidia/nv-embedqa-e5-v5

workflow:
  _type: react_agent
  tool_names: [webpage_query, current_datetime]
  llm_name: nim_llm
  verbose: true
  parse_agent_response_max_retries: 3

The workflow contains two tools: one that queries the LangSmith User Guide, and another that returns the current date and time. It also contains two models: an embedding model and an LLM model. After running the workflow, you can query it for information about LangSmith. This tutorial demonstrates how to customize this workflow.

Each workflow contains several configuration parameters that can be modified to customize the workflow. While copying and modifying the file is possible, it is not always necessary as some parameters can be overridden using the --override flag.

Examining the examples/getting_started/simple_web_query/configs/config.yml file, the llms section is as follows:

llms:
  nim_llm:
    _type: nim
    model_name: meta/llama-3.1-70b-instruct
    temperature: 0.0

To override the temperature parameter for the nim_llm, the following command can be used:

nat run --config_file examples/getting_started/simple_web_query/configs/config.yml --input "What is LangSmith?"  \
  --override llms.nim_llm.temperature 0.7

When successful, the output contains the following line:

nat.cli.cli_utils.config_override - INFO - Successfully set override for llms.nim_llm.temperature with value: 0.7

The --override flag can be specified multiple times, allowing the ability to override multiple parameters. For example, the llama-3.1-70b-instruct model can be replaced with the llama-3.3-70b-instruct using:

nat run --config_file examples/getting_started/simple_web_query/configs/config.yml --input "What is LangSmith?"  \
  --override llms.nim_llm.temperature 0.7 \
  --override llms.nim_llm.model_name meta/llama-3.3-70b-instruct

Note

Not all parameters are specified in the workflow YAML. For each tool, there are potentially multiple optional parameters with default values that can be overridden. The nat info components command can be used to list all available parameters. In this case, to list all available parameters for the LLM nim type run:

nat info components -t llm_provider -q nim