OllamaFunctions

caution

This was an experimental wrapper that bolted-on tool calling support to models that do not natively support it. The primary Ollama integration now supports tool calling, and should be used instead. See example usage in LangChain v0.2 documentation here.

This notebook shows how to use an experimental wrapper around Ollama that gives it the same API as OpenAI Functions.

Note that more powerful and capable models will perform better with complex schema and/or multiple functions. The examples below use llama3 and phi3 models. For a complete list of supported models and model variants, see the Ollama model library.

Setup

Follow these instructions to set up and run a local Ollama instance.

Usage

You can initialize OllamaFunctions in a similar way to how you'd initialize a standard ChatOllama instance:

from langchain_experimental.llms.ollama_functions import OllamaFunctions

model = OllamaFunctions(model="llama3", format="json")

API Reference:

• OllamaFunctions

You can then bind functions defined with JSON Schema parameters and a function_call parameter to force the model to call the given function:

model = model.bind_tools(
    tools=[
        {
            "name": "get_current_weather",
            "description": "Get the current weather in a given location",
            "parameters": {
                "type": "object",
                "properties": {
                    "location": {
                        "type": "string",
                        "description": "The city and state, " "e.g. San Francisco, CA",
                    },
                    "unit": {
                        "type": "string",
                        "enum": ["celsius", "fahrenheit"],
                    },
                },
                "required": ["location"],
            },
        }
    ],
    function_call={"name": "get_current_weather"},
)

Calling a function with this model then results in JSON output matching the provided schema:

from langchain_core.messages import HumanMessage

model.invoke("what is the weather in Boston?")

API Reference:

• HumanMessage

AIMessage(content='', additional_kwargs={'function_call': {'name': 'get_current_weather', 'arguments': '{"location": "Boston, MA"}'}}, id='run-1791f9fe-95ad-4ca4-bdf7-9f73eab31e6f-0')

Structured Output

One useful thing you can do with function calling using with_structured_output() function is extracting properties from a given input in a structured format:

from langchain_core.prompts import PromptTemplate
from langchain_core.pydantic_v1 import BaseModel, Field


# Schema for structured response
class Person(BaseModel):
    name: str = Field(description="The person's name", required=True)
    height: float = Field(description="The person's height", required=True)
    hair_color: str = Field(description="The person's hair color")


# Prompt template
prompt = PromptTemplate.from_template(
    """Alex is 5 feet tall. 
Claudia is 1 feet taller than Alex and jumps higher than him. 
Claudia is a brunette and Alex is blonde.

Human: {question}
AI: """
)

# Chain
llm = OllamaFunctions(model="phi3", format="json", temperature=0)
structured_llm = llm.with_structured_output(Person)
chain = prompt | structured_llm

API Reference:

• PromptTemplate

Extracting data about Alex

alex = chain.invoke("Describe Alex")
alex

Person(name='Alex', height=5.0, hair_color='blonde')

Extracting data about Claudia

claudia = chain.invoke("Describe Claudia")
claudia

Person(name='Claudia', height=6.0, hair_color='brunette')