HPC-Coder-V2: Studying Code LLMs Across Low-Resource Parallel Languages

TL;DR

This paper investigates the challenges of fine-tuning large language models for high performance computing, focusing on parallel code generation, and presents a specialized HPC LLM that outperforms existing open-source models.

Contribution

It provides an in-depth analysis of fine-tuning HPC-specific LLMs and introduces a new specialized model that excels in parallel code generation tasks.

Findings

Fine-tuning improves parallel code generation capabilities.

The specialized HPC LLM outperforms previous open-source models.

Identifies key hurdles in adapting LLMs for HPC domains.

Abstract

Large Language Model (LLM) based coding tools have been tremendously successful as software development assistants, yet they are often designed for general purpose programming tasks and perform poorly for more specialized domains such as high performance computing. Creating specialized models and tools for these domains is crucial towards gaining the benefits of LLMs in areas such as HPC. While previous work has explored HPC-specific models, LLMs still struggle to generate parallel code and it is not at all clear what hurdles are still holding back these LLMs and what must be done to overcome them. In this work, we conduct an in-depth study along the many axes of fine-tuning a specialized HPC LLM in order to better understand the challenges. Based on our findings we fine-tune and evaluate a specialized HPC LLM that is shown to be the best performing open-source code LLM for parallel code generation to date.