© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 10806
    Prev
    Next
    INDEX
    Explanations

    s prefixes

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

    The neuron fires on blocks of SEO-style or spam-style promotional keywords and product/service names (e.g. “Download mp3,” “exhibition stand,” “furniture rental Dubai,” etc.).

    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
     са
    0.70
    サ
    0.69
    स्
    0.65
    萨
    0.63
     крат
    0.61
    са
    0.59
     glomerul
    0.59
     クロ
    0.58
    セイ
    0.58
     تحليل
    0.57
    POSITIVE LOGITS
     shortstop
    0.90
    ♠
    0.88
     sẽ
    0.86
     sendo
    0.85
     số
    0.83
     Số
    0.81
     seaside
    0.80
     szpital
    0.80
     선수
    0.79
     sông
    0.78
    Activations Density 3.216%

    No Known Activations