© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-4B
    5. 25-TRANSCODER-HP
    7. 107533
    Prev
    Next
    INDEX
    Explanations

    against

    np_max-act-logits · gemini-2.0-flash
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    mwhanna/qwen3-4b-transcoders/layer_25.safetensors
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    Features
    163,840
    Data Type
    float32
    Hook Name
    blocks.25.mlp.hook_in
    Architecture
    transcoder
    Context Size
    8,192
    Dataset
    monology/pile-uncopyrighted
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    让å®ĥ
    -0.27
     Reference
    -0.26
    è°ª
    -0.25
    使åħ¶
    -0.24
     twe
    -0.24
    lik
    -0.24
    AZE
    -0.24
     recurrence
    -0.24
    łģ
    -0.24
     fisheries
    -0.23
    POSITIVE LOGITS
    ochen
    0.28
     Shank
    0.27
    argest
    0.27
     onCreateView
    0.27
    æĬķåħ¥åĪ°
    0.27
    åĢŁåı£
    0.26
     cleanse
    0.26
    ï½Ĭ
    0.26
    æ´Ĺ
    0.26
    inger
    0.25
    Activations Density 0.891%

    No Known Activations