a repo of examples of using local llm models with langgraph containing two examples.
-
local_llm_langgraph/testing_ollama_llm_w_function_calling.py
- a mwe of langraph with a single custom tool for evaluating the success of local llms for function calling.
- edit the local llm settings for langchain for ollama, vllm, or llama cpp and add the local llm you are testing
- run
python3 testing_ollama_llm_w_function_calling.py
-
local_llm_langgraph/ollama_llm_lats.py
- the end goal final example of using a custom reflexion tool and duckduckgo search tool to implment language agent tree search for local llm models
- to test update the environment variables to for your hosted openai api server/framework and local llm and add a test prompt.
- run
python3 ollama_llm_lats.py.py
Note: ollama openai server implmetnation doesn't include function calling / tool use. Langgraph is missing a bind_tools method for everything other than the default OpenAi ChatOpenAI
llm method for the lats example