© 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. 2560
    Prev
    Next
    INDEX
    Explanations

    introduces contrast after a statement

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

    This neuron is highly sensitive to sentence- or clause-boundary tokens and especially to the first words or pronouns (e.g. “I,” “And,” discourse markers) that begin a new sentence or major clause.

    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
    совой
    0.68
    これにより
    0.61
    續
    0.60
    옅
    0.58
    或是
    0.58
    𓐄
    0.57
    plier
    0.57
    ለጠ
    0.57
    etches
    0.57
    醾
    0.57
    POSITIVE LOGITS
     but
    3.31
     nhưng
    2.95
     BUT
    2.92
     pero
    2.84
     ولكن
    2.83
     لكن
    2.81
    but
    2.80
     mutta
    2.73
     But
    2.72
     אבל
    2.69
    Activations Density 1.254%

    No Known Activations