Skip to content

Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

All papers

Found 3 papers
Filter3

System type

Models and APIs3

Layer

Model layer3Training and data layer0

System component

Monitors3Reasoning chains7Fine-tuning and open weights3Vision10Audio5Image generation3Multimodal18
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.
Layer：Model layerLayer：Training and data layerSystem component：Image generationClear all

Another 233papers do not have category tags yet and are not included in filtered results.

  1. InterpretabilityarXiv 2610.06945

    超越线性表征假设：KANSteer 用非线性激活引导文本到图像模型

    提出 KANSteer，用非线性激活引导拟合文生图有序概念轨迹

    arXiv
    2610.06945
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    FLUX.1-schnell, FLUX.1-dev
    AbstractKANSteer 用曲线替代固定方向。

    KANSteer 研究文生图扩散模型里，有序视觉概念能否沿单一激活方向被遍历，并改用一维曲线做 steering。

    Paper analysisRead analysis
  2. InterpretabilityarXiv 2609.39625

    D-Scope：用稀疏自编码器分解与操控扩散 Transformer

    提出 D-Scope，用视觉质心检索 SAE 解码方向并转向扩散生成

    arXiv
    2609.39625
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    SANA-Sprint, Nitro-1-PixArt, SANA teacher and others4
    AbstractD-Scope 用视觉质心检索单个 SAE 方向做掩码转向，对比检索提高区域增益但不一致改善保持。

    D-Scope把扩散 Transformer 中 SAE 特征的高激活图像块，接到文本查询驱动的单方向生成干预上。

    Paper analysisRead analysis
  3. InterpretabilityarXiv 2609.09909

    解读文本到图像扩散模型中的对象依赖概念脆弱性

    用逐步 SAE 诊断文生图概念脆性，并推理时向类原型插值

    arXiv
    2609.09909
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Evaluated models
    SD 1.5, SD 3.5, SDXL and others5
    Abstract逐步 SAE 诊断物体依赖的概念脆性，早期向类原型插值可在五骨干上提高风格与属性指标。

    作者用逐步稀疏自编码器审计文生图扩散模型里随物体变化的概念脆性，并用推理时原型插值检验该诊断。。

    Paper analysisRead analysis
You have reached the end