Year: 2025
Type: conference-paper
Abstract: Large language models (LLMs) have significantly advanced natural language processing, particularly through the integration of external tools and APIs.However, their effectiveness is frequently hampered by parameter misfilling during tool (function) calling.In this paper, we propose the Hierarchical Tool Error Checklist (HiTEC) framework to systematically diagnose and mitigate tool-calling errors without relying on extensive real-world interactions.HiTEC introduces a two-tiered approach: a global error checklist that identifies common, cross-tool issues, and a local error checklist that targets tool-specific and contextual failures.Building on this structure, we propose two deployments: HiTEC-In Context Learning (HiTEC-ICL) and HiTEC-Kahneman-Tversky Optimization (HiTEC-KTO).HiTEC-ICL embeds the global checklist in the initial prompts and leverages a two-round conversational interaction to dynamically refine parameter handling, while HiTEC-KTO generates high-quality negative examples to drive fine-tuning via preference-based optimization.Extensive experiments conducted on five public datasets show that our framework improves parameter-filling accuracy by up to 42% compared to baseline methods.Please read the following error checklist.Error 0: Wrong Tool Name Error … Error 2: Missing Required Parameter Error Query: The weather of 37°47′N 122°25′W on Jan 1st 2025 Function Calling Output: [{{"name": weather_api,"arguments": {{"lon":37.47,"lat":125.25}}}}]Error Message: {{"error": "MissingRequiredParameter","message": "The parameter "time" is required."}}Thought of Error: Parameter "time" is missing.Ensure all required parameters ("lat", "lon","time") are included in the function call. … Error 6: Redundant Information Error …Please avoid similar errors in the checklists and refine your previous answer.
Institution: Hong Kong University of Science and Technology
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
OpenAlex ID: W4412888152