cs.LG（2025-12-08）

📊 共 8 篇论文 | 🔗 2 篇有代码

🎯 兴趣领域导航

#	题目	一句话要点	标签
1	The Geometry of Persona: Disentangling Personality from Reasoning in Large Language Models	提出Soul Engine框架，通过解耦人格与推理能力实现安全可控的LLM个性化	large language model
2	FOAM: Blocked State Folding for Memory-Efficient LLM Training	FOAM：面向内存高效LLM训练的块状状态折叠优化器	large language model
3	Balanced Accuracy: The Right Metric for Evaluating LLM Judges -- Explained through Youden's J statistic	提出使用Balanced Accuracy评估LLM Judge，解决传统指标对类别不平衡的敏感性问题	large language model
4	LUNE: Efficient LLM Unlearning via LoRA Fine-Tuning with Negative Examples	LUNE：基于负例LoRA微调的高效LLM知识遗忘框架	large language model
5	Recover-to-Forget: Gradient Reconstruction from LoRA for Efficient LLM Unlearning	提出Recover-to-Forget框架，通过LoRA梯度重构实现高效LLM遗忘学习。	foundation model

#	题目	一句话要点	标签	🔗
6	PlantBiMoE: A Bidirectional Foundation Model with SparseMoE for Plant Genomes	PlantBiMoE：一种用于植物基因组的双向稀疏MoE基础模型	Mamba foundation model	✅
7	KAN-Dreamer: Benchmarking Kolmogorov-Arnold Networks as Function Approximators in World Models	KAN-Dreamer：探索Kolmogorov-Arnold网络在World Models中作为函数逼近器的性能	reinforcement learning world model dreamer
8	Dual Refinement Cycle Learning: Unsupervised Text Classification of Mamba and Community Detection on Text Attributed Graph	提出DRCL：一种无监督文本分类与文本属性图社区检测双重优化框架	Mamba representation learning	✅