© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 57018
    Prev
    Next
    INDEX
    Explanations

    building descriptions

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

    The neuron is selectively activated by words naming or describing physical structures or institutional entities (e.g. “Station,” “building,” “church,” “museum,” “market,” “hall,” “mill,” “facility,” “historic,” “oldest,” etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    newInstance
    -0.84
    tedra
    -0.81
    ocytosis
    -0.74
    estina
    -0.73
    했다
    -0.72
    jdk
    -0.71
    brido
    -0.71
    loroethane
    -0.70
    鵰
    -0.69
    łosi
    -0.69
    POSITIVE LOGITS
     buildings
    6.00
     building
    5.88
    building
    4.41
    Building
    4.16
     Building
    3.97
    buildings
    3.91
     edificio
    3.83
     Buildings
    3.77
     BUILDING
    3.66
    建物
    3.52
    Activations Density 0.152%

    No Known Activations