© 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. 31-GEMMASCOPE-2-RES-16K
    7. 11354
    Prev
    Next
    INDEX
    Explanations

    spoken hesitation filler words

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

    The neuron responds strongly to very high‐frequency function words and filler/discourse tokens (e.g. “we,” “to,” “do,” “is,” “basically,” “some,” “of,” “and,” etc.).

    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_31_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
     (
    0.86
    !”
    0.83
    !"
    0.82
    !!"
    0.70
    ឲ
    0.69
     (‘
    0.66
    /
    0.66
     WWII
    0.66
    !!”
    0.66
    ！」
    0.65
    POSITIVE LOGITS
     ähm
    1.61
     uh
    1.48
     ờ
    1.26
     euh
    1.20
     เอ่อ
    1.20
     äh
    1.19
     อ่า
    1.17
     cuz
    1.11
     Uh
    1.08
     yeah
    1.06
    Activations Density 0.206%

    No Known Activations