© 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. 24-GEMMASCOPE-2-RES-16K
    7. 13766
    Prev
    Next
    INDEX
    Explanations

    narrative intro phrases

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

    This neuron strongly fires on sentence‐boundary punctuation (especially end‐of‐sentence marks like periods, dashes, and related high-emphasis tokens).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    paragraph breaks and transitions between sections in news articles.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    indicators of hard-news reporting, such as attributed speech, dates/times, locations, official titles, acronyms, and other journalistic structure cues.

    oai_token-act-pair · gpt-5Triggered 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
     (
    1.08
    /
    0.90
    （
    0.88
     (=
    0.80
     (!
    0.77
     ("
    0.71
     (“
    0.71
     („
    0.70
    Furthermore
    0.70
     («
    0.69
    POSITIVE LOGITS
     तस्वी
    0.82
     அதிகாரிகள்
    0.80
     расчет
    0.79
     говорит
    0.78
     деньги
    0.77
    ുകയാണ്
    0.75
     значит
    0.75
    鑾
    0.74
     कामयाबी
    0.74
     있고요
    0.74
    Activations Density 0.091%

    No Known Activations