© 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. 7600
    Prev
    Next
    INDEX
    Explanations

    kindness, X, Y, fullName, year

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

    The neuron is simply marking tokens that occur very far into (i.e. late in) a document, acting as a “position‐in‐the‐document” detector rather than responding to any particular word meaning.

    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
     ανε
    0.76
     assent
    0.71
     Еще
    0.71
     Dateien
    0.68
     incont
    0.68
    нди
    0.68
    Mediterr
    0.68
    }')
    0.67
     deseos
    0.67
    =======
    0.67
    POSITIVE LOGITS
    被称为
    0.85
    स
    0.81
    pits
    0.79
    नावश्यक
    0.79
    与此同时
    0.78
    ó
    0.76
    אר
    0.76
    መሳሳይ
    0.75
    upakan
    0.75
     \*
    0.75
    Activations Density 0.000%

    No Known Activations