© 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. 8946
    Prev
    Next
    INDEX
    Explanations

    anticipate attacks and enemies

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

    The neuron fires whenever it sees a salient content word—especially key nouns or topic terms that signal what the passage is about.

    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
     قتل
    0.63
     photoshoot
    0.58
    憫
    0.57
     کشت
    0.57
     humiliating
    0.56
     autopsy
    0.56
     massac
    0.55
    嬢
    0.55
    mehndi
    0.54
     chuyện
    0.53
    POSITIVE LOGITS
     enemy
    1.94
    敌人
    1.94
    敵人
    1.84
     enemies
    1.83
     terrorists
    1.83
     robbers
    1.82
     attackers
    1.80
    enemy
    1.76
     thieves
    1.76
     intruders
    1.74
    Activations Density 0.791%

    No Known Activations