In a five-dataset benchmark, fine-tuned Llama-series LLMs match or beat smaller code models on some datasets, excel on long samples, yet class imbalance remains the dominant factor in detection performance.
Devign: Effective vulnerability identifi- cation by learning comprehensive program semantics via graph neural networks
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
cs.CL 1years
2024 1verdicts
CONDITIONAL 1roles
background 1polarities
support 1representative citing papers
citing papers explorer
-
Investigating Large Language Models for Code Vulnerability Detection: An Experimental Study
In a five-dataset benchmark, fine-tuned Llama-series LLMs match or beat smaller code models on some datasets, excel on long samples, yet class imbalance remains the dominant factor in detection performance.