LLM4AD

Large Language Model for Algorithm Design

LLM4AD harnesses the power of large language models to streamline and enhance the process of algorithm design. Our platform enables users to quickly generate, test, and refine algorithms using natural language inputs, making complex algorithm tasks more accessible and efficient.

Document GitHub

Get started with LLM4AD

Large language model for algorithm design (LLM4AD) platform

An efficient, large language model-based framework for algorithm design, aimed at assisting researchers and related users in this field to conduct experimental exploration and industrial applications more quickly and conveniently.

# Import necessary libraries
import numpy as np

def heuristic(item, bins):
    diff = bins - item  # remaining capacity
    exp = np.exp(diff)
    sqrt = np.sqrt(diff)
    ulti = 1 - diff / bins 
    comb = ulti * sqrt  # combination of utilization and square root
    adjust = np.where(diff > (item * 3), comb + 0.8, comb + 0.3)

    # hybrid score based on exponent term
    hybrid_exp = bins / ((exp + 0.7) * exp) 
    # sum of hybrid score and adjustment
    scores = hybrid_exp + adjust  

    return scores
Run quickstart
Video Example

Solve Traveling Salesman Problem (TSP) with LLM4AD

What's new in LLM4AD

Design Image

LLM4AD Paper Online

2024-12-24

The paper introducing the LLM4AD platform has now been made available on arXiv.

Design Image

Hello, LLM4AD!

2024-11-11

An open-source platform using Large Language Models (LLMs) for Algorithm Design (AD).

Design Image

Systematic Survey

2024-11-10

A Systematic Survey on Large Language Models for Algorithm Design (LLM4AD).