Self-Discover LlamaPack

This LlamaPack implements Self-Discover: Large Language Models Self-Compose Reasoning Structures paper.

It has two stages for the given task:

1. STAGE-1:

   a. SELECT: Selects subset of reasoning Modules.

   b. ADAPT: Adapts selected reasoning modules to the task.

   c. IMPLEMENT: It gives reasoning structure for the task.

2. STAGE-2: Uses the generated reasoning structure for the task to generate an answer.

The implementation is inspired from the codebase

CLI Usage

You can download llamapacks directly using llamaindex-cli, which comes installed with the llama-index python package:

llamaindex-cli download-llamapack SelfDiscoverPack --download-dir ./self_discover_pack

You can then inspect the files at ./self_discover_pack and use them as a template for your own project!

Code Usage

There are two ways using LlamaPack:

1. Do download_llama_pack to load the Self-Discover LlamaPack.
2. Directly use SelfDiscoverPack

Using download_llama_pack

from llama_index.core.llama_pack import download_llama_pack

# download and install dependencies
SelfDiscoverPack = download_llama_pack(
    "SelfDiscoverPack", "./self_discover_pack"
)

self_discover_pack = SelfDiscoverPack(verbose=True, llm=llm)

Directly use SelfRAGPack

from llama_index.packs.self_discover import SelfDiscoverPack

self_discover_pack = SelfRAGPack(llm=llm, verbose=True)

The run() function serves as a concise wrapper that implements the logic outlined in the "self-discover" paper, applying it to a sample task as illustrated below.

Emma needs to prepare 50 invitations for her upcoming birthday party. She can handwrite 10 invitations in an hour. After working for 2 hours, she takes a break for 30 minutes. If she resumes writing at the same pace, how long will it take her to complete all 50 invitations?

output = pack.run("<task>")