LLM-HYPER: Generative CTR Modeling for Cold-Start Ad Personalization via LLM-Based Hypernetworks

LLM-HYPER proposes using LLM-based hypernetworks to improve click-through rate prediction in cold-start ad personalization. The method combines generative modeling with hypernetwork architecture to handle scenarios where user/item history is limited. This represents an application of LLMs to advertising optimization and personalization tasks.