© 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. 6997
    Prev
    Next
    INDEX
    Explanations

    programming languages

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

    The neuron strongly activates on the very first token of a line or section—whether that’s a heading word (like “modifier,” “statement,” or “added”) or a line‐number token—effectively marking the start of new lines or blocks.

    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
    acon
    0.59
     बुन्देल
    0.54
    -
    0.54
     receiver
    0.53
    endem
    0.52
    袋
    0.52
    PMC
    0.52
    queryString
    0.50
    _
    0.50
    \
    0.50
    POSITIVE LOGITS
     కుటు
    0.66
    arına
    0.66
     Eğ
    0.65
     Victims
    0.64
     marka
    0.60
     fakat
    0.60
     తల్లి
    0.59
     участия
    0.58
     אך
    0.57
     satta
    0.57
    Activations Density 0.002%

    No Known Activations