© 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. 13168
    Prev
    Next
    INDEX
    Explanations

    omit or void

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

    This neuron fires on transitional or conditional cue words (e.g. “however,” “before,” “included,” “have,” “size,” “September”) that typically introduce exceptions, qualifications, or timing in formal/legal or policy‐style text.

    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
     aides
    0.88
     aident
    0.75
    妠
    0.74
     atelier
    0.73
    🅔
    0.73
    વન
    0.72
    inde
    0.72
     elog
    0.71
    ጼ
    0.70
     Silvio
    0.69
    POSITIVE LOGITS
     Кы
    0.76
     صحت
    0.75
     Акы
    0.73
     Многие
    0.72
    鉈
    0.72
     Далее
    0.71
     কেশ
    0.70
    ściej
    0.70
    𝚕
    0.70
    鹜
    0.69
    Activations Density 0.000%

    No Known Activations