Open justin13601 opened 2 weeks ago
Given the standardized configuration language, large language models may be leveraged to provide a natural language interface for the creation of task configurations. Initial GPT-4 testing using in-context learning with the configuration language documentation yielded promising results; however, more detailed and quantitative evaluation is needed to determine the feasibility and accuracy of this interface.