© 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. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 59298
    Prev
    Next
    INDEX
    Explanations

    fictional worlds and places

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron strongly activates on tokens that are parts of fictional place names (e.g. Sanctuary, Tri-stram, Wakanda, Ankh-Morpork, Asgard, Midgard).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     or
    -1.77
     on
    -1.77
     this
    -1.76
     then
    -1.71
     now
    -1.68
     such
    -1.66
    8
    -1.63
     when
    -1.60
     today
    -1.49
     venerdì
    -1.44
    POSITIVE LOGITS
    IFFS
    1.60
     estekak
    1.38
    ховик
    1.38
    龊
    1.38
    坠
    1.35
    ⟣
    1.34
    了一口
    1.29
     championnat
    1.29
    
    1.28
     poderia
    1.27
    Activations Density 0.052%

    No Known Activations