cs.LG（2026-01-06）

📊 共 15 篇论文 | 🔗 1 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (9 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (6)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (9 篇)

#题目一句话要点标签🔗⭐
1 Uni-FinLLM: A Unified Multimodal Large Language Model with Modular Task Heads for Micro-Level Stock Prediction and Macro-Level Systemic Risk Assessment Uni-FinLLM：统一多模态大语言模型，用于微观股票预测和宏观系统性风险评估 large language model multimodal
2 Empowering Reliable Visual-Centric Instruction Following in MLLMs 提出VC-IFEval基准，提升多模态大语言模型在视觉约束下的指令跟随能力 large language model multimodal instruction following
3 From Memorization to Creativity: LLM as a Designer of Novel Neural-Architectures 提出基于LLM的闭环神经架构设计框架，实现从记忆到创造的飞跃。 large language model
4 ATLAS: Adaptive Test-Time Latent Steering with External Verifiers for Enhancing LLMs Reasoning ATLAS：利用外部验证器进行自适应测试时潜在引导，增强LLM推理能力 large language model
5 Audit Me If You Can: Query-Efficient Active Fairness Auditing of Black-Box LLMs 提出BAFA以解决黑箱LLM的公平性审计问题 large language model
6 Joint Encoding of KV-Cache Blocks for Scalable LLM Serving 提出KV-Cache联合编码，提升LLM高并发服务吞吐量并降低内存占用 large language model ✅
7 When the Coffee Feature Activates on Coffins: An Analysis of Feature Extraction and Steering for Mechanistic Interpretability 针对Llama 3.1的稀疏自编码器特征提取与操控脆弱性分析 large language model
8 Bridging Mechanistic Interpretability and Prompt Engineering with Gradient Ascent for Interpretable Persona Control 提出基于梯度上升的可解释Prompt工程方法，实现对LLM行为Persona的精准控制 large language model
9 Prioritized Replay for RL Post-training 提出一种基于问题优先级的强化学习后训练框架，提升大型语言模型在强化学习任务中的性能。 large language model

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
10 Adversarial Contrastive Learning for LLM Quantization Attacks 提出对抗对比学习ACL，提升LLM量化攻击的成功率 contrastive learning large language model
11 Sparse Knowledge Distillation: A Mathematical Framework for Probability-Domain Temperature Scaling and Multi-Stage Compression 提出稀疏知识蒸馏框架以优化模型压缩与温度缩放问题 distillation
12 Decentralized Autoregressive Generation 提出去中心化自回归生成方法，解决多模态语言模型训练中的专家协作问题。 flow matching multimodal
13 Causal Manifold Fairness: Enforcing Geometric Invariance in Representation Learning 提出因果流形公平性(CMF)，通过几何不变性实现表征学习中的公平性。 representation learning
14 In-Context Reinforcement Learning through Bayesian Fusion of Context and Value Prior SPICE：通过上下文和价值先验的贝叶斯融合实现上下文强化学习 reinforcement learning
15 Stratified Hazard Sampling: Minimal-Variance Event Scheduling for CTMC/DTMC Discrete Diffusion and Flow Models 提出分层风险抽样(SHS)，最小化CTMC/DTMC离散扩散模型的事件调度方差，提升生成质量。 flow matching multimodal

⬅️ 返回 cs.LG 首页 · 🏠 返回主页