© 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. 24-GEMMASCOPE-2-RES-16K
    7. 6931
    Prev
    Next
    INDEX
    Explanations

    The neuron is selectively firing on tokens that are part of document section headings or labels (e.g. “Work Setting:”, “Desc:”) rather than running 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_24_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
    ア
    0.97
    ла
    0.82
     LA
    0.78
     tiro
    0.77
     Luxembourg
    0.76
    Soul
    0.75
     ONE
    0.73
     problematic
    0.73
     LU
    0.73
    卢
    0.73
    POSITIVE LOGITS
    ştik
    0.90
     nær
    0.86
     eyeing
    0.85
    жды
    0.82
    evším
    0.81
    ſe
    0.80
     effektiv
    0.80
     esche
    0.79
    সেই
    0.79
    \}$,
    0.78
    Activations Density 0.000%

    No Known Activations