Ollama fixes

[dharmapurikar]

GPT Researcher - Fix Issues when Running in Docker Container

Issue Description

The GPT Researcher tool was unable to complete research tasks with Ollama when run inside a Docker container on WSL2. This issue has been resolved by addressing two key problems.

Issue 1: Incorrect Ollama Base URL

By default, the tool attempted to connect to localhost instead of using the provided base URL for Ollama. We have updated the code to use the provided base URL when running in a Docker container.

Issue 2: Error in Generating Detailed Report

When generating detailed reports, we encountered an error because get_subtopics was sending a list instead of a dictionary. This has been corrected by updating the get_subtopics function to return a dictionary.

Changes Made

The following changes have been made to address these issues:

• Updated base.py to use the provided Ollama base URL.
• Modified the get_subtopics function in report.py to return a dictionary instead of a list.

Steps to Reproduce the Fix

To reproduce the fix, follow these steps:

1. Without fixes, run the gpt-researcher in using docker compose on WSL2

Code Changes

The code changes made to address these issues include updating the following files:

• base.py
• report.py

The updated code can be found in the attached code snippet.

[assafelovic]

This is great, thank you so much for the contribution.