Recently, reinforcement learning from verifiable rewards (RLVR) is practically very important, where the leading approach is group relative policy optimization (GRPO) that typically faces a trade-off between rollout efficiency and downstream performance. To improve efficiency while preserving performance, existing sele...
Zhen-Yu Zhang, Jian-Dong Zhang, Hua-Xiu Yao et al.· 1 citation
EoupCT effectively preserves both task-specific proficiency and inherent general-purpose knowledge, successfully mitigating the catastrophic forgetting, and forms a multi-objective optimization problem and introduces a first-order efficient Pareto optimizer.
Bing Wang, C. Li, Xin-Qiang Cai et al.· 2 citations
Attention with Routed Memory is proposed a novel KV caching structure that introduces a fully differentiable, fixed-size memory system organized as a hierarchical router that achieves superior performance and efficiency compared to fixed KV-caching approaches, while remaining efficient and scalable in terms of both mem...
Qiu-Hao Zeng, Jerry M. Huang, Peng Lu et al.· arXiv.org· 0 citations
DenMark is proposed, a semantic watermarking framework that injects key-dependent signals directly into the DLM denoising process and achieves the best results across all reported detection metrics in all 48 backbone-dataset-attack combinations.
Tian-Hao Ma, Weihao Xuan, Dong-Dong Wu et al.· 1 citation
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.