XXXiong/ChatHLS-HLSFixer

ChatHLS-HLSFixer: Specialized HLS Debugging LLM

ChatHLS-HLSFixer is a highly specialized Large Language Model developed by XXXiong, fine-tuned from Qwen/Qwen2.5-14B-Instruct. Its primary purpose is to act as a core analysis agent within the ChatHLS framework, focusing on High-Level Synthesis (HLS) C/C++ error correction.

Key Capabilities

• HLS Error Identification: Pinpoints erroneous code lines in C/C++ programs that are incompatible with HLS.
• Root Cause Analysis: Explains the underlying reasons for HLS bugs by analyzing compiler error logs and comparing buggy code with expected patterns.
• Automated Fix Proposals: Generates specific, actionable modification actions to resolve identified HLS errors.
• Chain of Thought (CoT) Debugging: Employs a structured CoT strategy to guide the debugging process, addressing errors sequentially and validating hypotheses.
• Structured Output: Provides bug analysis in a standardized JSON format, detailing erroneous lines, reasons, and proposed fixes.

Good For

• Automated HLS Debugging: Ideal for developers and researchers working with High-Level Synthesis who need to quickly identify and fix HLS-incompatible errors in C/C++ code.
• Integrating into HLS Toolchains: Can be integrated into larger HLS design automation and optimization frameworks, such as the ChatHLS project.
• Educational Purposes: Useful for understanding common HLS pitfalls and debugging strategies through its detailed CoT output.