© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 8993
    Prev
    Next
    INDEX
    Explanations

    social cues and responsiveness

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

    The neuron consistently lights up on discourse‐level transition markers—things like colons, contrastive adverbs (“despite,” “in spite”), sequence cues (“once,” “move forward”), and similar connectives.

    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
    取决
    0.73
     toxins
    0.72
     ferocious
    0.71
     costly
    0.71
     grueling
    0.71
     brutal
    0.70
     colossal
    0.68
     insurg
    0.68
     their
    0.67
     ruin
    0.66
    POSITIVE LOGITS
     вполне
    1.25
     durchaus
    1.17
     ちょっと
    1.01
     довольно
    0.97
     वगैरह
    0.97
    などが
    0.96
     వంటి
    0.96
     えっと
    0.96
    などを
    0.95
     ব্যাপারটা
    0.94
    Activations Density 0.364%

    No Known Activations