© 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-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 2870
    Prev
    Next
    INDEX
    Explanations

    after a problem

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

    This neuron isn’t detecting any particular word or meaning but rather fires on whatever token appears at a roughly fixed position in the input sequence (i.e. it’s a positional marker neuron).

    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_24_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
    があり
    0.78
    で
    0.73
    {
    0.72
    以來
    0.70
    無法
    0.67
    起源
    0.65
    以来
    0.64
     evasion
    0.64
    arantine
    0.63
     averse
    0.63
    POSITIVE LOGITS
    ladı
    0.91
    ManagerProxy
    0.88
    ಧಿಕ
    0.87
     اون
    0.86
     oun
    0.84
     כך
    0.83
    ی
    0.82
     baton
    0.82
    İŞ
    0.82
    ﯿ
    0.82
    Activations Density 0.001%

    No Known Activations