© 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. 24-GEMMASCOPE-2-RES-16K
    7. 1427
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on rare or low‐frequency tokens—uncommon proper names, obscure words, and unusual numbers.

    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
    ir
    0.94
    𝑙
    0.84
    SCH
    0.83
    GE
    0.82
    etc
    0.80
     niñas
    0.80
    GR
    0.77
    JT
    0.77
    GraphQL
    0.77
    PAP
    0.76
    POSITIVE LOGITS
     berita
    0.74
     Maritime
    0.73
     offshore
    0.72
     ك
    0.70
     وعلى
    0.69
    heimer
    0.68
     freight
    0.67
     rumah
    0.67
     bhfu
    0.66
     resmi
    0.66
    Activations Density 0.068%

    No Known Activations