© 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. 10339
    Prev
    Next
    INDEX
    Explanations

    modern LLMs

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron is tuned to spot privacy‐/security‐policy language—i.e. words naming protective or regulatory actions (protect, notify, respect, attempt, etc.).

    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
    م
    0.79
    ので
    0.76
    在地
    0.72
    sta
    0.71
    不大
    0.70
    Các
    0.70
    Jeśli
    0.68
    STATE
    0.67
    リ
    0.67
    不太
    0.66
    POSITIVE LOGITS
     recomb
    0.95
     protease
    0.93
     reductase
    0.91
     bioqu
    0.86
     тща
    0.82
     soliton
    0.82
     vál
    0.80
     эпо
    0.80
     synchrotron
    0.79
    жнее
    0.79
    Activations Density 0.001%

    No Known Activations