© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 15621
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on runs of digits—especially multi‐digit numbers like years—i.e. it picks out numeric sequences.

    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
    তিকর
    0.57
    Efficient
    0.57
     Anch
    0.55
    ълга
    0.55
     უ
    0.54
     EXISTING
    0.54
    вым
    0.54
    ußen
    0.54
     ആൻ
    0.54
    Overview
    0.53
    POSITIVE LOGITS
    /'.$
    0.67
    '.$
    0.61
    /
    0.52
     slammed
    0.52
    叩
    0.52
    isk
    0.51
     incl
    0.51
     Glock
    0.49
    ंप
    0.49
    aines
    0.49
    Activations Density 0.018%

    No Known Activations