© 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. 24-GEMMASCOPE-2-RES-16K
    4. 2775
    Prev
    Next
    INDEX
    Explanations

    The neuron isn’t picking up any coherent linguistic pattern or semantic category—it simply fires strongly on a handful of idiosyncratic vocabulary tokens (e.g. “vw,” “mushroom,” “incorporate,” “any,” “costs,” 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_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
    fic
    0.73
    pher
    0.71
    ᡵ
    0.69
    უფ
    0.67
    mv
    0.66
     soms
    0.66
    fh
    0.65
     verder
    0.65
    짓
    0.65
    ƣ
    0.64
    POSITIVE LOGITS
     prescribes
    0.79
    تے
    0.77
     בן
    0.76
     himself
    0.76
     مكان
    0.75
    ISE
    0.73
     شدم
    0.73
     tactile
    0.73
     postulate
    0.71
     linens
    0.71
    Activations Density 0.002%

    No Known Activations