arXiv 2408.09121

Selective Prompt Anchoring for Code Generation

By Yuan Tian and Tianyi Zhang

Published 2024-08-17

Mindmap

Browse the paper's core ideas, clusters, and relationships in a structured outline.

Recent advances in large language models (LLMs) have transformed software development by automatically generating code from natural language. Yet challenges remain in generating fully correct code that aligns with user intent. Our study reveals that LLMs tend to pay less attention to user prompts as more code tokens are generated. We hypothesize that this attention dilution issue is an important reason for code gene…

View the original paper on arXiv