Giga Researcher

Russian description can be accessed here.

✨ A literature analysis tool that will help researchers study articles and write code faster.

✨ Giga Researcher is an autonomous system based on AI agents designed for comprehensive online research on a variety of tasks. The agent can produce detailed, factual, and unbiased research reports, with customization options for focusing on relevant resources.

😎 It is powered by GigaChat and GigaChain.

System design:

The workflow runs "planner", "execution", and "report generator" agents to perform the literature analysis and create a report. The planner generates research questions (RQ). Then, multiple execution agents seek the most relevant information based on RQs. Finally, the report generator filters and aggregates all crawled information and creates a research report.

✨ How to use it

📖 Note: Make sure to put GITHUB and GIGACHAIN API keys in secret_key.env.

👉There are arxiv and web crawl search retrievers. Both generate reports for asked questions or prompts with paper titles for the former and URL references for the latter retrievers type.

👉Additionally, there is a custom link search that lets the user query provided URL to generate a report based on a prompt. This also includes code links from GitHub.

✨ Code structure and installations for arxiv and web search

Install all the dependencies using requirements.txt by navigating to the parent directory, and creating a virtual environment (recommended).

pip install -r requirements.txt

💻arxiv_search_git → Contains the files to run the arxiv retriever along with GitHub repo links. The github_retrieve directory, handles the GitHub part of the agent. main.py is the entry point. Navigate to the directory arxiv_search_git and run the following:

cd arxiv_search_git/
python main.py

💻web_search_git → Contains the files to run the web retriever along with GitHub repo links. The github_retrieve directory, handles the GitHub part of the agent. main.py is the entry point. Navigate to the directory web_search_git and run the following:

cd web_search_git/
python main.py    

✨ The solution can be accessed locally via the route below:

http://localhost:8000/giga_research/playground/

📖 Note: Both the arxiv and web search should be hosted on the same address, hence, either change the ports or use them separately. Also, put your own GITHUB and GIGACHAIN API keys in secret_key.env.

✨ Code structure and installations for custom link search

Install all the dependencies using requirements.txt by navigating to the parent directory, and creating a virtual environment (recommended).

pip install -r requirements.txt

main.py is the entry point. Navigate to the directory custom_link_search and run the following:

cd custom_link_search/
python main.py    

✨ The solution can be accessed locally via the route below:

http://localhost:4000/research-assistant/playground/

📖 Note: Put your own GITHUB and GIGACHAIN API keys in secret_key.env. This service would be hosted on port:4000, hence can be simultaneously used with arxiv or web agent.

Authors

Jayveersinh Raj

Made Oka Resia Wedamarta

Vladimir Kalabukhov