© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 7320
    Prev
    Next
    INDEX
    Explanations

    Mahsa Amini

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

    The neuron is strongly activated by uncommon capitalized proper names (in particular, people’s surnames).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    くる
    0.62
     Rutland
    0.60
    prisingly
    0.59
    उट
    0.58
    여
    0.56
     withdrawal
    0.56
     parallels
    0.56
     ゆ
    0.55
    ěstí
    0.55
     Parallel
    0.54
    POSITIVE LOGITS
    kampf
    0.70
     PAH
    0.69
    ্
    0.66
    seen
    0.64
    Gauche
    0.63
     Beh
    0.60
    rmann
    0.60
    ্‌
    0.59
    now
    0.58
    副本
    0.58
    Activations Density 0.363%

    No Known Activations