Skip to content

Latest commit

 

History

History
118 lines (88 loc) · 3.58 KB

README.md

File metadata and controls

118 lines (88 loc) · 3.58 KB
Together Open Deep Research Logo

Together Open Deep Research

AI-powered comprehensive research on complex topics

Blog Post • Overview • Features • Installation • Usage • Disclaimer

✨ Overview

Together Open Deep Research is an agentic LLM workflow that delivers in-depth research on complex topics requiring multi-hop reasoning. It enhances traditional web search by producing comprehensive, well-cited content that mimics the human research process - planning, searching, evaluating information, and iterating until completion.

🎯 Features

  • Comprehensive Research Reports - Generates long-form, well-cited content on complex topics
  • Multi-Stage Process - Uses multiple self-reflection stages for quality information gathering
  • Source Verification - Provides citations for all information sources
  • Extensible Architecture - Designed with a flexible foundation for community extension

🔧 Installation

Prerequisites

Before installing, ensure you have Python 3.12+ and the following tools:

Tool macOS Ubuntu/Debian Windows
Pandoc brew install pandoc sudo apt-get install pandoc Download installer
pdfLaTeX brew install basictex sudo apt-get install texlive-xetex Download MiKTeX

Setup Environment

# Install uv (faster alternative to pip)
curl -LsSf https://astral.sh/uv/install.sh | sh

# Create and activate virtual environment
uv venv --python=3.12
source .venv/bin/activate

# Install project dependencies
uv pip install -r pyproject.toml
uv lock --check

# Optional: install with open-deep-research package (for langgraph evals)
uv pip install -e ".[with-open-deep-research]"

Configure API Keys

export TOGETHER_API_KEY=your_key_here
export TAVILY_API_KEY=your_key_here
export HUGGINGFACE_TOKEN=your_token_here

🚀 Usage

Run the deep research workflow:

# Set Python path
export PYTHONPATH=$PYTHONPATH:$(pwd)/src

# Run with default options
python src/together_open_deep_research.py --config configs/open_deep_researcher_config.yaml

Or run the gradio webapp:

python src/webapp.py

Options

  • --write-pdf - Generate a PDF document of the report
  • --write-html - Create an HTML version of the report
  • --write-podcast - Create a Podcast of the entire artcle
  • --add-toc-image - Add a visual table of contents image
  • --config PATH - Specify a custom configuration file (default: configs/open_deep_researcher_config.yaml)

⚠️ Disclaimer

As an LLM-based system, this tool may occasionally:

  • Generate hallucinations or fabricate information that appears plausible
  • Contain biases present in its training data
  • Misinterpret complex queries or provide incomplete analyses
  • Present outdated information

Always verify important information from generated reports with primary sources.