llm
3 artigo(s).
- AI SecurityDistillation Attack on LLMs (Model Extraction via Knowledge Distillation)In normal machine learning, knowledge distillation is a training method. A large "teacher" model teaches a smaller "student" model. The student learns to copy the teacher's outputs. This makes the student faster and chea
- AI SecurityLLM Security: A Scientific Taxonomy of Attack VectorsA technical overview of how LLM security research evolved from adversarial examples to systemic risks: jailbreaks, prompt injection, memorization, backdoors, and supply chain issues.
- SoftwareHow a Large Language Model (LLM) WorksA clear overview of how LLMs work: Transformer architecture, data processing, training, alignment (SFT, RLHF, RLAIF), and efficiency techniques like MoE, RAG, and FlashAttention.