Adding New Capability in Existing Scientific Application with LLM Assistance
This addresses a specific bottleneck in automated coding for scientific applications, but appears incremental as it builds on an existing tool.
The paper tackles the problem of generating code for new algorithms where no similar examples exist in training data, by proposing a methodology using LLM assistance and enhancing the Code-Scribe tool, resulting in a new capability for scientific applications.
With the emergence and rapid evolution of large language models (LLM), automating coding tasks has become an important research topic. Many efforts are underway and literature abounds about the efficacy of models and their ability to generate code. A less explored aspect of code generation is for new algorithms, where the training dataset would not have included any previous example of similar code. In this paper we propose a new methodology for writing code from scratch for a new algorithm using LLM assistance, and describe enhancement of a previously developed code-translation tool, Code-Scribe, for new code generation.