© 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. 31-GEMMASCOPE-2-RES-16K
    7. 5677
    Prev
    Next
    INDEX
    Explanations

    links in markdown formats

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

    The neuron strongly activates on URL components—e.g. slashes, domain names, query strings—i.e. tokens that form parts of web addresses.

    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
     prachtige
    0.53
     écr
    0.52
     beheld
    0.51
     краси
    0.51
     끌
    0.50
     chinh
    0.50
     atr
    0.49
    靘
    0.49
     prestar
    0.49
     식
    0.49
    POSITIVE LOGITS
    にし
    0.53
    един
    0.50
     =========
    0.50
    若
    0.50
     juveniles
    0.49
    А
    0.48
    ・
    0.48
    来看看
    0.47
    即使
    0.47
     swirl
    0.47
    Activations Density 0.412%

    No Known Activations