© 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. 3104
    Prev
    Next
    INDEX
    Explanations

    both examples in greetings and lists

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

    The neuron activates strongly on numeric tokens—especially decimal or large numeric values—so it’s essentially detecting numbers in the text.

    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
    ๓
    0.86
     등
    0.78
    etc
    0.78
    0
    0.77
    UNESCO
    0.75
    6
    0.75
    7
    0.74
     미술
    0.74
    ๕
    0.74
    CNN
    0.73
    POSITIVE LOGITS
     both
    0.95
     its
    0.89
     their
    0.88
     både
    0.86
     zarówno
    0.81
     zowel
    0.79
     those
    0.77
     tenancy
    0.76
     the
    0.76
     newfound
    0.75
    Activations Density 0.370%

    No Known Activations