© 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. 9434
    Prev
    Next
    INDEX
    Explanations

    constructor with input whatever

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

    This neuron isn’t looking for any particular words or meanings but rather fires strongly on tokens that occur at a fixed position within each text segment (i.e. it detects the Nth token after the segment’s start).

    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
     Cũng
    0.85
     いる
    0.80
     వల్ల
    0.77
     Ở
    0.72
     इससे
    0.71
     Quanto
    0.71
     Су
    0.71
    РИ
    0.69
     Ирина
    0.69
     Při
    0.69
    POSITIVE LOGITS
    ள்ளன
    0.86
     organizers
    0.76
     furlough
    0.68
     cybersecurity
    0.68
     firefighting
    0.67
    serialization
    0.67
     cleats
    0.66
    presentations
    0.65
    skateboard
    0.65
    لكتر
    0.65
    Activations Density 0.001%

    No Known Activations