© 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. 1273
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly fires on structural punctuation and markup-style tokens (e.g. square brackets, parentheses, regex delimiters), effectively spotting code‐ or metadata‐like fragments.

    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
    विटी
    0.59
    ब्रेरी
    0.58
     راجسټ
    0.58
    iry
    0.57
    <unused2221>
    0.55
    ient
    0.55
    வ்வாற
    0.55
     formats
    0.54
    iky
    0.53
    <unused2150>
    0.53
    POSITIVE LOGITS
    sigh
    0.94
    NON
    0.89
     NON
    0.88
    não
    0.88
     NÃO
    0.83
    SOLD
    0.83
    sorry
    0.82
    youtube
    0.82
    BUT
    0.81
    THAT
    0.80
    Activations Density 0.503%

    No Known Activations