© 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. 9182
    Prev
    Next
    INDEX
    Explanations

    This neuron highlights high-information content words—tokens that serve as key “keywords” or carry strong topic-specific weight.

    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
    textAllCaps
    0.63
     Schwartz
    0.55
    yne
    0.52
     удалось
    0.52
     Oc
    0.50
    ˳
    0.50
    فيق
    0.50
    ̒
    0.50
     ভাষ
    0.50
    uly
    0.50
    POSITIVE LOGITS
     vulnerable
    2.81
     vulnerability
    2.67
     vulnerabilities
    2.53
     vulner
    2.42
     Vulner
    2.41
     susceptible
    2.33
     vuln
    2.28
    脆弱
    2.22
     risky
    2.17
     susceptibility
    2.16
    Activations Density 0.708%

    No Known Activations