© 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. 13545
    Prev
    Next
    INDEX
    Explanations

    mathematical expressions and sequences with specific punctuation

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

    The neuron is triggered by mathematical content—particularly numbers, numeric expressions, and formula‐like tokens (e.g. “How,” “integral,” “length,” numeric values, and operators).

    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
     congregate
    0.78
     paperwork
    0.76
     ERISA
    0.76
    欉
    0.73
    Plaintiffs
    0.72
     सर्जरी
    0.70
    齙
    0.70
     litigation
    0.69
    holding
    0.69
    ectomy
    0.69
    POSITIVE LOGITS
     đồng
    0.63
    <
    0.63
     రెండో
    0.62
     Jeżeli
    0.59
     <
    0.59
     położ
    0.59
     czyli
    0.58
    っていう
    0.58
     Schüler
    0.58
     jeżeli
    0.57
    Activations Density 0.005%

    No Known Activations