Update README.md

This commit is contained in:
Zachary Huang 2025-01-09 13:38:11 -05:00 committed by GitHub
parent 55f8ac0bac
commit b0322b099a
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
1 changed files with 8 additions and 4 deletions

View File

@ -1,8 +1,16 @@
<h1 align="center">Pocket Flow - LLM Framework in 100 Lines</h1>
![License: MIT](https://img.shields.io/badge/License-MIT-yellow.svg)
[![Docs](https://img.shields.io/badge/docs-latest-blue)](https://minillmflow.github.io/PocketFlow/)
<div align="center">
<img src="/assets/minillmflow.jpg" width="400"/>
</div>
<br>
A [100-line](pocketflow/__init__.py) minimalist LLM framework for ([Multi-](https://minillmflow.github.io/PocketFlow/multi_agent.html))[Agents](https://minillmflow.github.io/PocketFlow/agent.html), [task decomposition](https://minillmflow.github.io/PocketFlow/decomp.html), [RAG](https://minillmflow.github.io/PocketFlow/rag.html), etc.
- Install via ```pip install pocketflow```, or just copy the [source codes](pocketflow/__init__.py) (only 100 lines)
@ -43,10 +51,6 @@ The ideal framework for LLMs should (1) **strip away low-level implementation de
Pocket Flow is also a *learning resource*, as current frameworks abstract too much away.
<div align="center">
<img src="/assets/minillmflow.jpg" width="400"/>
</div>
## How Does it Work?
The [100 lines](minillmflow/__init__.py) capture what we see as the core abstraction of most LLM frameworks: a **nested directed graph** that breaks down tasks into multiple (LLM) steps, with branching and recursion for agent-like decision-making. From there, its easy to layer on more complex features.