© 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. 11554
    Prev
    Next
    INDEX
    Explanations

    slightly, progressively

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

    The main thing this neuron does is spot emphatic adverbs (especially those ending in “-ly”) that intensify or color a description.

    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
    s
    0.92
    imize
    0.86
     নিষেধ
    0.85
    zám
    0.80
    Би
    0.80
    Фі
    0.79
    Су
    0.78
    ন্ধে
    0.77
    ों
    0.75
    Са
    0.75
    POSITIVE LOGITS
     Grizz
    0.85
     adventurous
    0.85
     handlebar
    0.85
     eclectic
    0.83
     iridescent
    0.83
     perishable
    0.82
     humorous
    0.82
     unorthodox
    0.81
     spook
    0.81
     resilient
    0.80
    Activations Density 0.473%

    No Known Activations