Ollama memory embeddings

[jmrichardson]

Description

When using ollama memory embeddings, i am getting the following errors:

Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Give Feedback / Get Help: https://github.com/BerriAI/litellm/issues/new
LiteLLM.Info: If you need to debug this error, use `litellm.set_verbose=True'.
Provider List: https://docs.litellm.ai/docs/providers
Failed to add to long term memory: Failed to convert text into a Pydantic model due to the following error: litellm.APIConnectionError: OllamaException - [WinError 10061] No connection could be made because the target machine actively refused it

Steps to Reproduce

Here is my config:

crew = Crew(
        agents=[agent],
        tasks=[task],
        process=Process.sequential,
        embedder={
          "provider": "ollama",
          "config": {
            "model": "nomic-embed-text",
            "url": "http://10.23.50.101:11434/api/embeddings",
          }
        },
        memory=True, # Enable memory for the crew - this activates short-term, long-term, entity, contextual memory
        verbose=True,
        llm=crew_llm # Explicitly set the Crew's LLM to ollama_llm
    )

Expected behavior

I am able to get embeddings by testing with Curl:

curl http://10.23.50.101:11434/api/embed -d "{ \"model\": \"nomic-embed-text\", \"input\": \"Llamas are members of the camelid family\" }"

Operating System

Ubuntu 20.04

Python Version

3.10

crewAI Version

crewai 0.100.1

crewAI Tools Version

crewai-tools 0.33.0

Virtual Environment

Conda

Evidence

None

Possible Solution

I have tried many different configs but none seem to work

Additional context

None

[jmrichardson]

I was able to resolve the problem by updating this line to:

            self._client = instructor.from_litellm(
                completion,
                model=self.llm.model,
                api_base=self.llm.base_url,
            )

the litellm was defaulting to OpenAI provider which I don't use. Hopefully someone can make the edit in main if it looks ok.