© 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. 6143
    Prev
    Next
    INDEX
    Explanations

    version is downloadable

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

    The neuron strongly activates on tokens that fall at or near sentence (or clause) boundaries—i.e. the last word or punctuation just before a period/comma/question-mark.

    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
    ósfera
    0.86
    Everybody
    0.81
    ُ
    0.79
    ِ
    0.78
    -/
    0.77
    たくさん
    0.77
    ஃப்
    0.77
     تُ
    0.75
    Asimismo
    0.75
     거예요
    0.75
    POSITIVE LOGITS
     কারন
    0.94
    非常的
    0.84
     nutzt
    0.83
     delà
    0.81
    的看着
    0.80
     কারনে
    0.77
     todays
    0.76
    떄
    0.76
     নির্মান
    0.76
     Fransiya
    0.75
    Activations Density 0.000%

    No Known Activations