LLMs can Schedule

Henrik Abgaryan,Ararat Harutyunyan,Tristan Cazenave

2024-08-13

Abstract:The job shop scheduling problem (JSSP) remains a significant hurdle in optimizing production processes. This challenge involves efficiently allocating jobs to a limited number of machines while minimizing factors like total processing time or job delays. While recent advancements in artificial intelligence have yielded promising solutions, such as reinforcement learning and graph neural networks, this paper explores the potential of Large Language Models (LLMs) for JSSP. We introduce the very first supervised 120k dataset specifically designed to train LLMs for JSSP. Surprisingly, our findings demonstrate that LLM-based scheduling can achieve performance comparable to other neural approaches. Furthermore, we propose a sampling method that enhances the effectiveness of LLMs in tackling JSSP.

Artificial Intelligence

What problem does this paper attempt to address?

The paper attempts to address the task allocation optimization problem in the **Job Shop Scheduling Problem (JSSP)**. Specifically: - **Main Objective**: The paper aims to optimize task scheduling in the production process by leveraging Large Language Models (LLMs), particularly exploring the effectiveness of LLMs in small-scale JSSP problems. - **Specific Challenges**: JSSP involves efficiently allocating jobs on a limited number of machines while minimizing factors such as total processing time or job delays. Traditional methods like mathematical programming techniques and heuristic algorithms have limitations in large-scale problems or complex job-machine priority relationships. - **Solution**: The paper introduces a brand-new supervised dataset containing 120,000 natural language descriptions of JSSP problem instances, specifically designed for training LLMs for scheduling tasks. Experimental results show that with minimal fine-tuning, LLMs can find high-quality solutions, sometimes even matching or surpassing traditional neural network methods. - **Additional Contribution**: A sampling method is proposed to further enhance the effectiveness of LLMs in solving JSSP problems. In summary, the paper demonstrates the potential application of LLMs in the JSSP field by introducing a new dataset and improved sampling techniques, and provides a comparative analysis with existing neural network methods.

LLMs can Schedule

Hybrid Deep Neural Network Scheduler for Job-Shop Problem Based on Convolution Two-Dimensional Transformation

Efficient LLM Scheduling by Learning to Rank

Learning to Solve Job Shop Scheduling under Uncertainty

Fast Inference for Augmented Large Language Models

Efficient Interactive LLM Serving with Proxy Model-based Sequence Length Prediction

Learning to schedule (L2S): adaptive job shop scheduling using double deep Q network

Fast Approximations for Job Shop Scheduling: A Lagrangian Dual Deep Learning Method

Is the GPU Half-Empty or Half-Full? Practical Scheduling Techniques for LLMs

Investigating the Potential of Using Large Language Models for Scheduling

Gated‐Attention Model with Reinforcement Learning for Solving Dynamic Job Shop Scheduling Problem

Solving Flexible Job-shop Problem with Sequence-dependent Setup Times by Using Selection Hyper-heuristics.

Slice-Level Scheduling for High Throughput and Load Balanced LLM Serving

Residual Scheduling: A New Reinforcement Learning Approach to Solving Job Shop Scheduling Problem

The Effect of Scheduling and Preemption on the Efficiency of LLM Inference Serving

Learning to schedule job-shop problems: representation and policy learning using graph neural network and reinforcement learning

Hhs/Lns: an Integrated Search Method for Flexible Job Shop Scheduling

Learning to generalize Dispatching rules on the Job Shop Scheduling

Graph Neural Networks for Job Shop Scheduling Problems: A Survey

Scheduling with Job-Splitting Considering Learning and the Vital-Few Law

An actor-critic algorithm with policy gradients to solve the job shop scheduling problem using deep double recurrent agents