遇见数据集

结构化代码表示数据集

收藏
魔搭社区2026-07-15 更新2026-07-15 收录
官方服务:

资源简介:

Code Structural Representations Dataset, Preprocessed code structural representations for clone detection and vulnerability detection. Includes three datasets (BCBench, Devign, OJClone) each with four representation types (raw, AST sequence, CFG sequence, IR sequence), plus a shared SentencePiece BPE tokenizer (50K vocab) with train/val/test splits (8:1:1).

提供机构:
maas
创建时间:
2026-07-09
二维码
社区交流群
二维码
科研交流群
商业服务