Reproduce “LSem2Vec: A Simple yet Effective Two-Stage Approach for Source Code Embedding” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Zixiang Xian, Chenhui Cui, Rubing Huang, Chunrong Fang, Zhenyu Chen
LSEM2VEC detects code clones by asking an LLM to summarize each snippet once, embedding those summaries with a sentence model, and comparing vectors with cosine similarity. It reuses representations instead of prompting on every pair, without task-specific training or fine-tuning. On OJClone C, the paper reports 95.36% F1 for GPT-5.4 plus MiniLM-L12, with comparisons bounded to clone detection and clustering.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.