© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 2771
    Prev
    Next
    INDEX
    Explanations

    This neuron detects formal auxiliary verbs (especially “does” and “will”) used in policy or legal statement contexts.

    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
    Doing
    0.63
     Doing
    0.60
    doing
    0.60
     sain
    0.55
    íme
    0.54
    ক্সি
    0.54
     робити
    0.53
    눙
    0.53
     típ
    0.52
    adaşlar
    0.52
    POSITIVE LOGITS
     not
    1.15
    not
    0.92
     NOT
    0.91
     indeed
    0.88
     infatti
    0.70
     Not
    0.69
    NotExist
    0.67
    NOT
    0.64
    Not
    0.64
    indeed
    0.62
    Activations Density 0.159%

    No Known Activations