© 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. 31-GEMMASCOPE-2-RES-16K
    7. 13172
    Prev
    Next
    INDEX
    Explanations

    preventing violence

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

    The neuron strongly activates on phrases naming criminal charges or legal descriptions of violence (e.g. assault, causing injury, charges in court).

    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
     superstition
    0.59
    忐
    0.55
     gossip
    0.54
     tacit
    0.54
     secretly
    0.51
     tél
    0.50
     собы
    0.50
     Typed
    0.50
    ጌ
    0.50
     suspicions
    0.50
    POSITIVE LOGITS
     wrestling
    1.30
     wrestled
    1.22
     wrestle
    1.15
     Wrestling
    1.14
     grappling
    1.13
     wrestlers
    1.09
     wrestler
    1.08
     Wrestle
    1.02
    restling
    1.00
    毆
    1.00
    Activations Density 0.211%

    No Known Activations