PDFSearchTool fails with Ollama

CrewAI Community Support
1

Hi,

Why does the following (copied from the CrewAI docs) not work, and results in complaints about missing OPENAI_API_KEY due to trying to use the default embedder. I hoped this problem had been addressed by the 0.102 update. How to be able to use Ollama for pdf, doclings etc, as every attempt ends up with missing OpenAI API keys? I have seen many having this kind of problems, but no agreed solutions. Is this not possible to resolve?

Thanks in advance,
magnus

from crewai import Agent, Task, Crew, Process

from crewai_tools import PDFSearchTool

# Initialize the tool allowing for any PDF content search if the path is provided during execution
tool = PDFSearchTool()

tool = PDFSearchTool(
    config=dict(
        llm=dict(
            provider="ollama", # or google, openai, anthropic, llama2, ...
            config=dict(
                model="ollama/llama3.1:latest",
                api_key=""
                # temperature=0.5,
                # top_p=1,
                # stream=true,
            ),
        ),
        embedder=dict(
            provider="ollama", # or openai, ollama, ...
            config=dict(
                model="ollama/nomic-embed-text:latest",
                task_type="retrieval_document",
                api_key=""
                # title="Embeddings",
            ),
        ),
    ),
    pdf='"D:/Crewai/Projects/test/AI_Maturity_Assessment.pdf"'
)

# Create an agent with the knowledge store
agent = Agent(
    role="About AI Maturity Assessments",
    goal="You know everything about AI Maturity Assessments .",
    backstory="""You are a master at summarizing research.""",
    verbose=True,
    allow_delegation=False,
    tools=[tool]
)
task = Task(
    description="Answer the following questions about AI Maturity Assessments: {question}",
    expected_output="An answer to the question.",
    agent=agent,
)

crew = Crew(
    agents=[agent],
    tasks=[task],
    verbose=True,
    process=Process.sequential,
)

result = crew.kickoff(inputs={"question": "What are the key conclustions?"})