Tag: Large Language Model

All the articles with the tag "Large Language Model".

• LoKI: Low-damage Knowledge Implanting of Large Language Models

  Published: 2 Jun, 2025 |  at  11:23 AM

  

  87.82 🤔

  本文提出LoKI，一种参数高效微调框架，通过分析Transformer FFN层的知识存储机制和层平衡参数选择策略，在下游任务适应和预训练知识保留之间实现了竞争性平衡。

• A Large-Scale Empirical Analysis of Custom GPTs' Vulnerabilities in the OpenAI Ecosystem

  Published: 16 May, 2025 |  at  11:13 AM

  

  94.41 🤔

  This paper conducts a large-scale empirical analysis of 14,904 custom GPTs in the OpenAI store, revealing over 95% lack adequate security against attacks like roleplay (96.51%) and phishing (91.22%), introduces a multi-metric popularity ranking system, and highlights the need for enhanced security in both custom and base models.

• Large Language Models are Locally Linear Mappings

  Published: 3 Jun, 2025 |  at  11:44 AM

  

  85.46 🤔

  本文提出了一种通过分离Jacobian将大型语言模型在特定输入点转化为近乎精确局部线性系统的方法，揭示了模型内部低秩语义结构，并初步探索了输出引导应用，但泛化性和实用性受限。

• RAISE: Reinforced Adaptive Instruction Selection For Large Language Models

  Published: 3 Jun, 2025 |  at  11:28 AM

  

  85.33 🤔

  本文提出 RAISE 框架，通过强化学习驱动的动态指令选择方法，根据指令对模型性能的预期影响自适应选择训练数据，仅用 1% 训练步骤即可超越全数据训练效果，并在多个基准测试中显著优于静态选择基线。

• Unraveling LoRA Interference: Orthogonal Subspaces for Robust Model Merging

  Published: 1 Jun, 2025 |  at  11:40 AM

  

  89.76 🤔

  本文提出OSRM方法，通过在微调前约束LoRA子空间以减少任务间干扰，显著提升了多个语言模型在八个GLUE数据集上的合并性能，同时保持单任务准确性。