As Large Language Models (LLMs) evolve into lifelong AI assistants, LLM personalization has become a critical frontier. However, progress is currently bottlenecked by the absence of a gold-standard evaluation benchmark. Existing benchmarks either overlook personalized information management that is critical for personalization or rely heavily on synthetic dialogues, which exhibit an inherent distribution gap from real-world dialogue. To bridge this gap, we introduce AlpsBench, An LLM PerSonalization benchmark derived from real-world human-LLM dialogues. AlpsBench comprises 2,500 long-term interaction sequences curated from WildChat, paired with human-verified structured memories that encapsulate both explicit and implicit personalization signals. We define four pivotal tasks - personalized information extraction, updating, retrieval, and utilization - and establish protocols to evaluate the entire lifecycle of memory management. Our benchmarking of frontier LLMs and memory-centric systems reveals that: (i) models struggle to reliably extract latent user traits; (ii) memory updating faces a performance ceiling even in the strongest models; (iii) retrieval accuracy declines sharply in the presence of large distractor pools; and (iv) while explicit memory mechanisms improve recall, they do not inherently guarantee more preference-aligned or emotionally resonant responses. AlpsBench aims to provide a comprehensive framework.

Transformer has widely been applied in various low-vision tasks, achieving significant strides in single image super-resolution. However, its low-pass characteristic still limits the ability of Transformer-based models to represent rich texture details in images. Additionally, the quadratic computational complexity of the attention mechanism also restricts its application in low-level tasks. In these regards, we propose a novel High-frequency awareness and Lightweight Mamba network (HiLiteMamba). It combines a lightweight Mamba module with a convolution block to model different features complementarily. Moreover, we develop the High-Freq Highway with residual connection to maintain deep propagation of high-frequency information and accelerate its modeling. Extensive experiments demonstrate that HiLiteMamba achieves state-of-the-art performance among baseline models.

Large language models (LLMs) like LLaMA, Baichuan and Bloom models show remarkable ability with instruction fine-tuning in many natural language tasks. Nevertheless, for the dialogue summarization task, which aims to generate summaries for different roles in dialogue, most of the state-of-the-art methods foucus on small models (e.g BART and BERT). Existing methods try to add task specified optimization on small models like adding global-local centrality score to models. In this paper, we propose an instruction fine-tuning model: Baichuan2-Sum, for role-oriented dialogue summarization. By setting different instructions for different roles, the model can learn from the dialogue interactions and output the desired summaries. Furthermore, we applied NEFTune technique to add suitable noise during training, improving the results. The experiments demonstrate that the proposed model achieves the new state-of-the-art results on two public dialogue summarization datasets: CSDS and SAMSUM. The Baichuan2-Sum model shows an improvement in Rouge scores on both datasets compared to the previously best-performing model. Notably, for the SAMSUM dataset, there is a 21% increase in the ROUGE-1 score, a 32% increase in the ROUGE-2 score, and a 9% increase in the ROUGE-L score. We have released our model and related codes to facilitate future studies in the dialogue summarization task.