© 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. 31-GEMMASCOPE-2-RES-16K
    7. 5469
    Prev
    Next
    INDEX
    Explanations

    segmentation types

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

    The neuron fires on document‐structure markers—specifically numeric outline headings (e.g. “4-1-1:”) and similar section‐number tokens.

    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
     L
    0.63
     Q
    0.60
     X
    0.59
     V
    0.58
     +
    0.57
    Helper
    0.57
     R
    0.57
     P
    0.55
     E
    0.55
     We
    0.55
    POSITIVE LOGITS
     أنواع
    0.61
    Gosudarstvennyj
    0.60
     típus
    0.57
     bedrijven
    0.57
     کشورهای
    0.56
    スタンダ
    0.56
    jenis
    0.55
     tradisional
    0.55
     военных
    0.55
     कंपनियां
    0.54
    Activations Density 0.205%

    No Known Activations