Home

Awesome

๐Ÿฆ™๐Ÿงช Llama Lab ๐Ÿงฌ๐Ÿฆ™

Llama Lab is a repo dedicated to building cutting-edge projects using LlamaIndex.

LlamaIndex is an interface for LLM data augmentation. It provides easy-to-use and flexible tools to index various types of data. At its core, it can be used to index a knowledge corpus. But it can also be used to index tasks, and provide memory-like capabilities for any outer agent abstractions.

Here's an overview of some of the amazing projects we're exploring:

Each folder is a stand-alone project. See below for a description of each project along with usage examples.

Contributing: We're very open to contributions! This can include the following:

Current Labs

llama_agi (v0.1.0)

Inspired from babyagi and AutoGPT, using LlamaIndex as a task manager and LangChain as a task executor.

The current version of this folder will start with an overall objective ("solve world hunger" by default), and create/prioritize the tasks needed to achieve that objective. LlamaIndex is used to create and prioritize tasks, while LangChain is used to guess the "result" of completing each action.

Using LangChain and LlamaIndex, llama_agi has access to the following tools: google-search, webpage reading, and note-taking. Note that the google-search tool requires a Google API key and a CSE ID.

This will run in a loop until the task list is empty (or maybe you run out of OpenAI credits ๐Ÿ˜‰).

For more info, see the README in the llama_agi folder or the pypi page.

auto_llama

Inspired by autogpt. This implement its own Agent system similar to AutoGPT. Given a user query, this system has the capability to search the web and download web pages, before analyzing the combined data and compiling a final answer to the user's prompt.

Example usage:

cd auto_llama
pip install -r requirements.txt
python -m auto_llama
Enter what you would like AutoLlama to do:
Summarize the financial news from the past week.

Conversational Agents

This is a fun conversational simulator between different agents. You can choose to provide some details about the context/setting, and watch as the conversation between different agents evolves.

A sample notebook is provided in the convo_agents folder. Usage:

cd convo_agents

jupyter notebook ConvoAgents.ipynb

External projects

We also provide references to other project repos using LlamaIndex in novel ways.

These repos are hosted as submodules in our external folder.

Check it out here: https://github.com/run-llama/llama-lab/tree/main/external

Ecosystem

Llama Lab is part of the broader Llama ecosystem.

Community: