Dream at SemEval-2026 Task 13: SALSA for Single-Pass Machine-Generated Code Detection

Researchers propose SALSA, a single-pass autoregressive LLM structured classification method, for detecting machine-generated code in SemEval-2026 Task 13. Their system achieves an out-of-distribution F1 score of 0.789, significantly outperforming the CodeBERT baseline of 0.305.

arXiv:2606.25102v1 Announce Type: new Abstract: Large language models have transformed code generation, raising concerns around authorship, assessment integrity, and software trust. SemEval-2026 Task 13 Subtask A operationalizes detection as binary classification over code snippets, with a particular emphasis on out-of-distribution OOD generalization across unseen programming languages and application domains. We propose a SALSA-style formulation, Single-pass Autoregressive LLM Structured Classification, that maps each class to a dedicated output token and trains the model to emit a single-token label in a structured response. Rather than engineering hand-crafted features or decision rules, this formulation delegates the authorship decision to the model. To improve OOD robustness, we combine balanced sampling across languages with parameter-efficient fine-tuning and conservative training low learning rate, single epoch to avoid overfitting to the training domain. Our best system achieves OOD $F 1 = 0.789$ on the official leaderboard, substantially outperforming the CodeBERT baseline $F 1 = 0.305$ .