© 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. 11937
    Prev
    Next
    INDEX
    Explanations

    HTML injection

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

    The neuron isn’t looking for any particular words or concepts at all but rather encodes where a token sits in the text (strongly firing on tokens at or near the beginnings of segments).

    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
    ದಯ
    0.89
    л
    0.87
     Dokter
    0.82
    تف
    0.82
    ᱽ
    0.82
    订阅
    0.81
     Persever
    0.81
    ﺸ
    0.79
    摶
    0.79
    できるように
    0.78
    POSITIVE LOGITS
     traditionally
    0.65
    Traditionally
    0.63
     numérique
    0.61
    ના
    0.60
     euro
    0.59
     digitales
    0.59
     physico
    0.59
     rectilinear
    0.58
    uber
    0.57
    opies
    0.57
    Activations Density 0.000%

    No Known Activations