© 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. 3237
    Prev
    Next
    INDEX
    Explanations

    dialogue marker

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

    The neuron fires strongly on dialogue attribution verbs (e.g. “said,” “asked,” “snapped,” “mocks,” “inquired”).

    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
     muttered
    0.85
     dizendo
    0.85
    瞥
    0.82
    咳嗽
    0.80
     аллер
    0.79
    ұ
    0.78
    𝚘
    0.77
     लड्डू
    0.77
    さて
    0.76
    হাস
    0.76
    POSITIVE LOGITS
    /
    0.87
     creates
    0.77
     luôn
    0.67
     countries
    0.67
     valuable
    0.67
    ↵
    0.65
     liberating
    0.64
     non
    0.64
     religions
    0.64
     methods
    0.64
    Activations Density 0.006%

    No Known Activations