Model2Kernel: Model-Aware Symbolic Execution For Safe CUDA Kernels
arXiv 2603.24595•dcbb2cbb7eab95d3a01e88b96dabf58e0ae30065ca116e1922b3822e3c85d318
AI-safetyCUDAGPULLM-inferenceRAGdynamic-analysisformal-verificationmemory-corruptionmemory-safetymodel-securityprogram-synthesisself-evolving-agentsstatic-analysissymbolic-executiontrusted-AIvulnerability-discovery
Paper metadata
- arXiv ID
- 2603.24595
- Version
- Not specified by this published record
- Category
- Computer Science — Programming Languages (cs.PL)
The PDF link points to arxiv.org. Baitaphish does not expose a private stored PDF.
Evidence and limitations
- Source ID
- arxiv_cs_pl
- Record identifier
- dcbb2cbb7eab95d3a01e88b96dabf58e0ae30065ca116e1922b3822e3c85d318
- Enrichment time
- 2026-03-27T08:52:01Z
- AI-assisted enrichment
- Yes
This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.