© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 2826
    Prev
    Next
    INDEX
    Explanations

    This neuron detects the major content‐bearing tokens that introduce or headline new sections or topics (e.g. “Information,” “Holidays,” “postings”) in a document.

    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
     tinh
    0.54
    랴
    0.53
     mur
    0.50
     طریقہ
    0.49
    ப்பட்டன
    0.48
     वृद्ध
    0.48
     MRC
    0.48
    ₁,
    0.48
     bilj
    0.48
    ₁
    0.47
    POSITIVE LOGITS
    SqlServer
    0.71
     স্বরে
    0.63
    asio
    0.57
     aead
    0.57
     Fils
    0.55
    ('$(
    0.54
     Бога
    0.54
    envs
    0.53
     словам
    0.52
    중
    0.52
    Activations Density 0.000%

    No Known Activations