© 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. 12314
    Prev
    Next
    INDEX
    Explanations

    code punctuation and numbers

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

    The neuron fires strongly on the first “real” content word after long runs of all‐zero tokens—i.e. it marks the start of a new paragraph or section.

    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
     Thym
    0.61
    abá
    0.60
    Fock
    0.58
     源
    0.57
     때는
    0.57
    শিরোনাম
    0.54
     छठे
    0.54
    ปิด
    0.53
    ród
    0.52
     생각이
    0.52
    POSITIVE LOGITS
     desert
    0.63
     deserts
    0.56
     diets
    0.51
     divaric
    0.51
    жин
    0.51
     Desert
    0.50
    裟
    0.50
     dzieci
    0.50
     crate
    0.49
    Ẓ
    0.49
    Activations Density 0.000%

    No Known Activations